{ "status": "measurements_complete_published", "updated": "2026-09-05T22:21:14.012602+08:00", "build_times": { "4090_base_build_seconds": 793.47, "4090_trt_layer_build_seconds": 522.8, "note": "Concurrent Orin build; dependency cache used; TRT build log subsequently overwritten by cached retry", "orin_first_deploy_seconds": 4526.42, "orin_first_sherpa_step_seconds": 1942.1, "orin_dynamic_sherpa_step_seconds": 1892.7, "orin_dynamic_deploy_seconds": 2308.95 }, "platforms": { "4090": { "recognizer-trt-3s": { "arguments": { "self_check": false, "mode": "recognizer", "provider": "trt", "model": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/model.onnx", "tokens": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/tokens.txt", "cache": "/bench/cache", "audio": "/bench/audio", "length": 3, "frames": "50,100,250", "warmup": 10, "repeats": 3, "threads": 2, "profile": false, "dump_context": null, "out": "/bench/4090/recognizer-trt-3s.json" }, "utc": "2026-09-05T13:07:33Z", "machine": "x86_64", "ort": "1.23.2", "numpy": "2.2.6", "model_sha256": "719559b0512db928e18eb3c130a122a55dc10166fb6bd2c8c90f2b6f51064994", "start": { "VmHWM_MiB": 47.41796875, "VmRSS_MiB": 45.9609375, "Pss_MiB": 45.95703125, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93826.515625, "memory.current_MiB": 28.3671875, "memory.peak_MiB": 30.52734375, "gpu_total_memory_util_clock_power": "9569, 0, 210, 25.13" }, "rows": [ { "seconds": 3, "p50_ms": 9.952828986570239, "p95_ms": 10.328458156436682, "samples_ms": [ 10.328458156436682, 9.833354968577623, 9.952828986570239 ], "texts": [ "帮我导航到内湖科学园区，然后。", "帮我导航到内湖科学园区，然后。", "帮我导航到内湖科学园区，然后。" ], "memory": { "VmHWM_MiB": 9419.62890625, "VmRSS_MiB": 7546.06640625, "Pss_MiB": 7541.5810546875, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 88025.75, "memory.current_MiB": 5450.9609375, "memory.peak_MiB": 7340.0546875, "gpu_total_memory_util_clock_power": "11096, 1, 2535, 67.71" }, "audio_sha256": "11034b66e2ae87bdec98dfe7401642387b85262b2e9c222b9e201030dc397a1c" } ], "load_s": 5.880446331109852, "ready": { "VmHWM_MiB": 9419.62890625, "VmRSS_MiB": 7536.2421875, "Pss_MiB": 7531.7568359375, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 88048.47265625, "memory.current_MiB": 5448.91015625, "memory.peak_MiB": 7340.0546875, "gpu_total_memory_util_clock_power": "11076, 1, 2535, 65.92" }, "after_warmup": { "VmHWM_MiB": 9419.62890625, "VmRSS_MiB": 7546.06640625, "Pss_MiB": 7541.5810546875, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 88037.37890625, "memory.current_MiB": 5450.93359375, "memory.peak_MiB": 7340.0546875, "gpu_total_memory_util_clock_power": "11096, 1, 2535, 66.63" }, "end": { "VmHWM_MiB": 9419.62890625, "VmRSS_MiB": 7546.06640625, "Pss_MiB": 7541.5810546875, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 88193.92578125, "memory.current_MiB": 5451.0, "memory.peak_MiB": 7340.0546875, "gpu_total_memory_util_clock_power": "11096, 1, 2535, 67.71" } }, "recognizer-cuda-3s": { "arguments": { "self_check": false, "mode": "recognizer", "provider": "cuda", "model": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/model.onnx", "tokens": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/tokens.txt", "cache": "/bench/cache", "audio": "/bench/audio", "length": 3, "frames": "50,100,250", "warmup": 10, "repeats": 3, "threads": 2, "profile": false, "dump_context": null, "out": "/bench/4090/recognizer-cuda-3s.json" }, "utc": "2026-09-05T12:43:02Z", "machine": "x86_64", "ort": "1.23.2", "numpy": "2.2.6", "model_sha256": "719559b0512db928e18eb3c130a122a55dc10166fb6bd2c8c90f2b6f51064994", "start": { "VmHWM_MiB": 46.0703125, "VmRSS_MiB": 45.765625, "Pss_MiB": 45.76171875, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 94615.38671875, "memory.current_MiB": 48.1328125, "memory.peak_MiB": 49.9921875, "gpu_total_memory_util_clock_power": "9569, 0, 210, 25.44" }, "rows": [ { "seconds": 3, "p50_ms": 12.696039164438844, "p95_ms": 12.77677109465003, "samples_ms": [ 12.77677109465003, 12.696039164438844, 12.600649846717715 ], "texts": [ "帮我导航到内湖科学园区，然后。", "帮我导航到内湖科学园区，然后。", "帮我导航到内湖科学园区，然后。" ], "memory": { "VmHWM_MiB": 1353.4921875, "VmRSS_MiB": 1343.7734375, "Pss_MiB": 1339.2314453125, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93844.37890625, "memory.current_MiB": 1300.125, "memory.peak_MiB": 1315.5859375, "gpu_total_memory_util_clock_power": "11008, 0, 2535, 65.54" }, "audio_sha256": "11034b66e2ae87bdec98dfe7401642387b85262b2e9c222b9e201030dc397a1c" } ], "load_s": 2.2766437819227576, "ready": { "VmHWM_MiB": 1353.4921875, "VmRSS_MiB": 1291.15234375, "Pss_MiB": 1286.6572265625, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93341.68359375, "memory.current_MiB": 1233.875, "memory.peak_MiB": 1299.078125, "gpu_total_memory_util_clock_power": "10998, 0, 2535, 64.23" }, "after_warmup": { "VmHWM_MiB": 1353.4921875, "VmRSS_MiB": 1343.7734375, "Pss_MiB": 1339.2314453125, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93667.83203125, "memory.current_MiB": 1300.09765625, "memory.peak_MiB": 1300.5078125, "gpu_total_memory_util_clock_power": "11008, 0, 2535, 63.71" }, "end": { "VmHWM_MiB": 1353.4921875, "VmRSS_MiB": 1343.7734375, "Pss_MiB": 1339.2314453125, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93719.484375, "memory.current_MiB": 1300.17578125, "memory.peak_MiB": 1315.59375, "gpu_total_memory_util_clock_power": "11008, 0, 2535, 65.54" } }, "environment": { "platform": "Linux-6.8.0-136-generic-x86_64-with-glibc2.36", "python": "3.10.19", "packages": { "onnxruntime-gpu": "1.23.2", "numpy": "2.2.6", "sherpa-onnx": "1.12.36", "torch": "2.9.1", "kitt-core": "0.3.7", "pipecat-ai": "0.0.108", "tensorrt-cu12": "10.13.3.9" }, "sha256": { "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/model.onnx": "719559b0512db928e18eb3c130a122a55dc10166fb6bd2c8c90f2b6f51064994", "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/tokens.txt": "f449eb28dc567533d7fa59be34e2abca8784f771850c78a47fb731a31429a1dc", "/app/src/asr/batch_asr_manager.py": "fb9a00f6cdbc5bd972d3bb82c85b44a70277a593d1694460071bac124ef25815", "/app/src/kitt_stt_service.py": "5283feaee7a213feb529afe1d98308c7fec447ccd18aa9f737a7632ea08b83bc" } }, "service-cuda-end-memory": { "utc": 1788616988.194794, "VmHWM_MiB": 2649.08984375, "VmRSS_MiB": 2610.45703125, "Pss_MiB": 2603.20703125, "memory.current_MiB": 2275.3984375, "memory.peak_MiB": 2799.4765625 }, "graph-trt": { "arguments": { "self_check": false, "mode": "graph", "provider": "trt", "model": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/model.onnx", "tokens": null, "cache": "/bench/cache", "audio": null, "length": 3, "frames": "50,100,250", "warmup": 10, "repeats": 50, "threads": 2, "profile": false, "dump_context": null, "out": "/bench/4090/graph-trt.json" }, "utc": "2026-09-05T13:08:40Z", "machine": "x86_64", "ort": "1.23.2", "numpy": "2.2.6", "model_sha256": "719559b0512db928e18eb3c130a122a55dc10166fb6bd2c8c90f2b6f51064994", "start": { "VmHWM_MiB": 47.16796875, "VmRSS_MiB": 45.73828125, "Pss_MiB": 45.734375, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93795.44921875, "memory.current_MiB": 28.2421875, "memory.peak_MiB": 30.19921875, "gpu_total_memory_util_clock_power": "9569, 0, 900, 64.13" }, "rows": [ { "frames": 50, "p50_ms": 6.0869010630995035, "p95_ms": 6.377147976309061, "samples_ms": [ 6.232210900634527, 6.172513822093606, 6.105597130954266, 6.08363002538681, 6.063100881874561, 6.157600088045001, 6.0095819644629955, 5.991270998492837, 5.9595550410449505, 6.377147976309061, 6.393870105966926, 6.403503008186817, 6.263872142881155, 6.193594075739384, 6.099425023421645, 6.119034951552749, 6.215111119672656, 6.088780937716365, 6.035469006747007, 6.060915999114513, 6.032600067555904, 6.217160960659385, 6.091483868658543, 6.057668011635542, 6.045526824891567, 6.085521075874567, 6.268962984904647, 6.086979061365128, 6.154519971460104, 6.185833131894469, 6.1594590079039335, 6.146664964035153, 6.0868230648338795, 6.09168503433466, 6.072461837902665, 6.097041070461273, 6.186744198203087, 6.151210051029921, 5.88117097504437, 5.889215972274542, 5.871551111340523, 5.975241074338555, 6.021746899932623, 5.912852007895708, 5.825829925015569, 5.800541955977678, 5.789068061858416, 5.990200908854604, 5.8707790449261665, 5.82902692258358 ], "memory": { "VmHWM_MiB": 9399.046875, "VmRSS_MiB": 7525.75390625, "Pss_MiB": 7521.2724609375, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 87526.36328125, "memory.current_MiB": 5441.66015625, "memory.peak_MiB": 7330.41015625, "gpu_total_memory_util_clock_power": "11106, 2, 2535, 93.17" }, "finite": true, "input_sha256": "472ba979ad46ce462cc2cceeb0d218314362876e03b63f06a136a1c19b3dbab7" }, { "frames": 100, "p50_ms": 5.878260941244662, "p95_ms": 6.150828907266259, "samples_ms": [ 5.831962218508124, 5.909692961722612, 6.124439183622599, 5.881454097107053, 6.015687948092818, 5.849621957167983, 5.875067785382271, 6.291236961260438, 6.458970019593835, 6.150828907266259, 5.890442058444023, 5.872976966202259, 5.931149004027247, 6.140192039310932, 6.00607693195343, 5.881513003259897, 5.849750014021993, 6.01952918805182, 6.087054032832384, 5.930023966357112, 5.862530088052154, 5.910641048103571, 5.915131885558367, 5.927688907831907, 5.950170801952481, 5.8406449388712645, 5.825205007568002, 5.814623087644577, 5.856079049408436, 5.990796023979783, 5.81294996663928, 5.783551139757037, 5.7526088785380125, 5.743007175624371, 5.923336837440729, 5.803894018754363, 5.774869816377759, 5.762081826105714, 5.901560885831714, 5.945226177573204, 5.970196099951863, 5.800989922136068, 5.773531971499324, 5.7534039951860905, 5.744708003476262, 6.005807081237435, 5.8594960719347, 5.816797958686948, 5.769510054960847, 5.78154600225389 ], "memory": { "VmHWM_MiB": 9399.046875, "VmRSS_MiB": 7525.859375, "Pss_MiB": 7521.3779296875, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 87451.15625, "memory.current_MiB": 5441.71484375, "memory.peak_MiB": 7330.41015625, "gpu_total_memory_util_clock_power": "11122, 2, 2535, 141.75" }, "finite": true, "input_sha256": "dda0d977bf96004405c6cbeac7cd57cfb466001cc340939617503e06dff71881" }, { "frames": 250, "p50_ms": 9.90213151089847, "p95_ms": 10.815754998475313, "samples_ms": [ 9.775936836376786, 10.06799703463912, 9.803483029827476, 9.799935854971409, 10.099423117935658, 9.886711835861206, 9.824865963310003, 9.805625071749091, 9.794787038117647, 9.837169200181961, 9.774973150342703, 10.128733003512025, 9.803519118577242, 10.029213968664408, 10.326871182769537, 9.839707985520363, 9.797341888770461, 10.183613048866391, 9.813320124521852, 9.850846137851477, 9.829391026869416, 10.815754998475313, 10.519578121602535, 10.09326707571745, 9.944404941052198, 10.087731992825866, 9.964150143787265, 10.404102969914675, 9.867944056168199, 9.836965007707477, 10.628944030031562, 9.923711884766817, 10.319409891963005, 11.334575014188886, 9.92819294333458, 9.81951504945755, 10.84224903024733, 9.968555998057127, 9.873359929770231, 10.072617093101144, 9.88934189081192, 9.914921130985022, 10.002356953918934, 9.940766962245107, 9.80430911295116, 9.789014933630824, 9.871133835986257, 9.835986187681556, 9.762719040736556, 10.394158074632287 ], "memory": { "VmHWM_MiB": 9399.046875, "VmRSS_MiB": 7525.875, "Pss_MiB": 7521.3896484375, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 87337.6484375, "memory.current_MiB": 5441.73828125, "memory.peak_MiB": 7330.41015625, "gpu_total_memory_util_clock_power": "11154, 76, 2715, 208.14" }, "finite": true, "input_sha256": "726cc736d46a4b8ca3c2d2831abc8c8e3a59489168194774d072d5fc1aa502b5" } ], "load_s": 5.807886126916856, "ready": { "VmHWM_MiB": 9399.046875, "VmRSS_MiB": 7523.38671875, "Pss_MiB": 7518.9013671875, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 87670.36328125, "memory.current_MiB": 5441.5, "memory.peak_MiB": 7330.41015625, "gpu_total_memory_util_clock_power": "11086, 27, 2535, 66.67" }, "providers": [ "TensorrtExecutionProvider", "CUDAExecutionProvider", "CPUExecutionProvider" ], "end": { "VmHWM_MiB": 9399.046875, "VmRSS_MiB": 7525.875, "Pss_MiB": 7521.3896484375, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 87333.95703125, "memory.current_MiB": 5441.93359375, "memory.peak_MiB": 7330.41015625, "gpu_total_memory_util_clock_power": "11154, 76, 2715, 208.14" } }, "service-cuda-ready-memory": { "utc": 1788616781.5262446, "VmHWM_MiB": 2583.98046875, "VmRSS_MiB": 2477.28515625, "Pss_MiB": 2470.041015625, "memory.current_MiB": 2142.265625, "memory.peak_MiB": 2256.34765625 }, "graph-cuda": { "arguments": { "self_check": false, "mode": "graph", "provider": "cuda", "model": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/model.onnx", "tokens": null, "cache": "/bench/cache", "audio": null, "length": 3, "frames": "50,100,250", "warmup": 10, "repeats": 50, "threads": 2, "profile": false, "dump_context": null, "out": "/bench/4090/graph-cuda.json" }, "utc": "2026-09-05T12:44:19Z", "machine": "x86_64", "ort": "1.23.2", "numpy": "2.2.6", "model_sha256": "719559b0512db928e18eb3c130a122a55dc10166fb6bd2c8c90f2b6f51064994", "start": { "VmHWM_MiB": 46.72265625, "VmRSS_MiB": 45.7578125, "Pss_MiB": 45.75390625, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 94363.1328125, "memory.current_MiB": 29.7578125, "memory.peak_MiB": 31.71484375, "gpu_total_memory_util_clock_power": "9569, 0, 210, 25.38" }, "rows": [ { "frames": 50, "p50_ms": 8.172842091880739, "p95_ms": 8.643068838864565, "samples_ms": [ 8.527150144800544, 8.481200085952878, 8.560962975025177, 8.489476051181555, 8.578567998483777, 8.645408088341355, 8.650887990370393, 8.561576949432492, 8.594484999775887, 8.643068838864565, 8.574220119044185, 8.489442989230156, 8.357185870409012, 8.069282863289118, 8.012234000489116, 7.905936799943447, 7.927268976345658, 8.041779045015574, 7.971392013132572, 7.934102090075612, 8.020943962037563, 8.093304000794888, 7.976610912010074, 7.879026001319289, 7.878228090703487, 7.9693240113556385, 7.935520028695464, 7.906096987426281, 7.933028042316437, 8.281433023512363, 8.428520988672972, 8.395337965339422, 8.227469166740775, 8.121690945699811, 8.06975201703608, 8.217249996960163, 8.14719800837338, 8.10133689083159, 8.081771899014711, 8.311423938721418, 8.275034837424755, 8.236921858042479, 8.122666971758008, 8.35025985725224, 8.20566387847066, 8.087004069238901, 8.085719076916575, 8.27315985225141, 8.196765091270208, 8.148919092491269 ], "memory": { "VmHWM_MiB": 1337.25, "VmRSS_MiB": 1249.69921875, "Pss_MiB": 1245.1650390625, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93248.890625, "memory.current_MiB": 1037.40625, "memory.peak_MiB": 1179.08203125, "gpu_total_memory_util_clock_power": "11008, 96, 2715, 94.26" }, "finite": true, "input_sha256": "472ba979ad46ce462cc2cceeb0d218314362876e03b63f06a136a1c19b3dbab7" }, { "frames": 100, "p50_ms": 8.167131571099162, "p95_ms": 8.49536620080471, "samples_ms": [ 8.470481028780341, 8.220166899263859, 8.219011127948761, 8.235972840338945, 8.379911072552204, 8.26529785990715, 8.44961917027831, 8.465796010568738, 8.223028853535652, 8.120185928419232, 8.086116053164005, 8.316634921357036, 8.152869995683432, 8.126328000798821, 8.048854069784284, 8.513472974300385, 8.30240617506206, 8.312301011756063, 8.175408001989126, 8.514976827427745, 8.308883057907224, 8.128295186907053, 8.080547209829092, 8.139024022966623, 8.210645988583565, 8.125566877424717, 8.063697954639792, 8.068103808909655, 8.188550127670169, 8.154212962836027, 8.148467866703868, 8.068480994552374, 8.115720003843307, 8.132621878758073, 8.054840844124556, 8.08222801424563, 8.251148043200374, 8.175170049071312, 8.35245382040739, 8.083882043138146, 8.132907096296549, 8.082648040726781, 8.49536620080471, 8.15618084743619, 8.159093093127012, 8.082743035629392, 8.2451649941504, 8.276771986857057, 8.202970959246159, 8.15177010372281 ], "memory": { "VmHWM_MiB": 1337.25, "VmRSS_MiB": 1250.0234375, "Pss_MiB": 1245.4853515625, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93232.3359375, "memory.current_MiB": 1037.328125, "memory.peak_MiB": 1179.08203125, "gpu_total_memory_util_clock_power": "11008, 96, 2715, 137.59" }, "finite": true, "input_sha256": "dda0d977bf96004405c6cbeac7cd57cfb466001cc340939617503e06dff71881" }, { "frames": 250, "p50_ms": 10.545336059294641, "p95_ms": 11.397663038223982, "samples_ms": [ 10.346119059249759, 11.035120114684105, 10.39771898649633, 10.345060843974352, 10.879555949941278, 10.750374058261514, 10.603742208331823, 11.397663038223982, 10.646871058270335, 10.459722951054573, 11.316381860524416, 10.518630035221577, 10.453202994540334, 11.43207703717053, 10.626694187521935, 10.407175868749619, 13.248879928141832, 11.390739120543003, 10.627188952639699, 10.87002013809979, 10.520662181079388, 10.482548968866467, 11.100440053269267, 10.626795934513211, 10.512762004509568, 11.165613075718284, 11.015312047675252, 10.65000705420971, 11.087868129834533, 10.425246087834239, 10.349211050197482, 10.35677408799529, 10.508568957448006, 10.37365198135376, 10.518023045733571, 10.406506014987826, 10.490281041711569, 10.744059924036264, 10.472094174474478, 10.324110044166446, 10.62650210224092, 10.45611291192472, 10.355989914387465, 10.890493867918849, 10.476930998265743, 10.497616836801171, 11.065457947552204, 10.570009937509894, 10.438109980896115, 10.881124995648861 ], "memory": { "VmHWM_MiB": 1337.25, "VmRSS_MiB": 1250.03125, "Pss_MiB": 1245.4892578125, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93191.5234375, "memory.current_MiB": 1037.13671875, "memory.peak_MiB": 1179.08203125, "gpu_total_memory_util_clock_power": "11008, 95, 2715, 186.99" }, "finite": true, "input_sha256": "726cc736d46a4b8ca3c2d2831abc8c8e3a59489168194774d072d5fc1aa502b5" } ], "load_s": 9.63523994688876, "ready": { "VmHWM_MiB": 1337.25, "VmRSS_MiB": 1194.55859375, "Pss_MiB": 1190.0634765625, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93821.203125, "memory.current_MiB": 1027.3125, "memory.peak_MiB": 1179.08203125, "gpu_total_memory_util_clock_power": "10998, 0, 2535, 63.97" }, "providers": [ "CUDAExecutionProvider", "CPUExecutionProvider" ], "end": { "VmHWM_MiB": 1337.25, "VmRSS_MiB": 1250.03125, "Pss_MiB": 1245.4892578125, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93172.6875, "memory.current_MiB": 1037.4453125, "memory.peak_MiB": 1179.08203125, "gpu_total_memory_util_clock_power": "11008, 95, 2715, 184.19" } }, "service-trt-ready-memory": { "utc": 1788617016.0895221, "VmHWM_MiB": 9795.2578125, "VmRSS_MiB": 9165.34375, "Pss_MiB": 9158.05078125, "memory.current_MiB": 7885.70703125, "memory.peak_MiB": 8526.69921875 }, "service-trt-end-memory": { "utc": 1788617233.0222907, "VmHWM_MiB": 9795.2578125, "VmRSS_MiB": 9288.94140625, "Pss_MiB": 9143.419921875, "memory.current_MiB": 8009.484375, "memory.peak_MiB": 8533.44921875 }, "recognizer-cuda-6s": { "arguments": { "self_check": false, "mode": "recognizer", "provider": "cuda", "model": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/model.onnx", "tokens": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/tokens.txt", "cache": "/bench/cache", "audio": "/bench/audio", "length": 6, "frames": "50,100,250", "warmup": 10, "repeats": 3, "threads": 2, "profile": false, "dump_context": null, "out": "/bench/4090/recognizer-cuda-6s.json" }, "utc": "2026-09-05T12:43:40Z", "machine": "x86_64", "ort": "1.23.2", "numpy": "2.2.6", "model_sha256": "719559b0512db928e18eb3c130a122a55dc10166fb6bd2c8c90f2b6f51064994", "start": { "VmHWM_MiB": 46.453125, "VmRSS_MiB": 46.12109375, "Pss_MiB": 46.1171875, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 94240.55859375, "memory.current_MiB": 30.77734375, "memory.peak_MiB": 32.72265625, "gpu_total_memory_util_clock_power": "9569, 0, 210, 24.95" }, "rows": [ { "seconds": 6, "p50_ms": 18.740060972049832, "p95_ms": 37.363274954259396, "samples_ms": [ 37.363274954259396, 16.705831978470087, 18.740060972049832 ], "texts": [ "帮我导航到内湖科学园区，然后选最快的路线，不要走高架桥。", "帮我导航到内湖科学园区，然后选最快的路线，不要走高架桥。", "帮我导航到内湖科学园区，然后选最快的路线，不要走高架桥。" ], "memory": { "VmHWM_MiB": 1350.796875, "VmRSS_MiB": 1318.4375, "Pss_MiB": 1313.8955078125, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93534.0390625, "memory.current_MiB": 1086.8984375, "memory.peak_MiB": 1175.015625, "gpu_total_memory_util_clock_power": "11008, 0, 2535, 65.04" }, "audio_sha256": "9566be6ccf6c7307830514e0fe5be86250c493efaef2800f27bfccc1ca1850b0" } ], "load_s": 3.0798465500120074, "ready": { "VmHWM_MiB": 1350.796875, "VmRSS_MiB": 1265.10546875, "Pss_MiB": 1260.6103515625, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93091.40625, "memory.current_MiB": 1080.43359375, "memory.peak_MiB": 1175.015625, "gpu_total_memory_util_clock_power": "10998, 0, 2535, 64.54" }, "after_warmup": { "VmHWM_MiB": 1350.796875, "VmRSS_MiB": 1318.359375, "Pss_MiB": 1313.8173828125, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 92721.37890625, "memory.current_MiB": 1086.91796875, "memory.peak_MiB": 1175.015625, "gpu_total_memory_util_clock_power": "11008, 0, 2535, 64.36" }, "end": { "VmHWM_MiB": 1350.796875, "VmRSS_MiB": 1318.4375, "Pss_MiB": 1313.8955078125, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93524.19921875, "memory.current_MiB": 1087.75, "memory.peak_MiB": 1175.015625, "gpu_total_memory_util_clock_power": "11008, 0, 2535, 65.79" } }, "recognizer-cuda-15s": { "arguments": { "self_check": false, "mode": "recognizer", "provider": "cuda", "model": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/model.onnx", "tokens": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/tokens.txt", "cache": "/bench/cache", "audio": "/bench/audio", "length": 15, "frames": "50,100,250", "warmup": 10, "repeats": 3, "threads": 2, "profile": false, "dump_context": null, "out": "/bench/4090/recognizer-cuda-15s.json" }, "utc": "2026-09-05T12:43:58Z", "machine": "x86_64", "ort": "1.23.2", "numpy": "2.2.6", "model_sha256": "719559b0512db928e18eb3c130a122a55dc10166fb6bd2c8c90f2b6f51064994", "start": { "VmHWM_MiB": 47.1328125, "VmRSS_MiB": 45.91796875, "Pss_MiB": 45.9140625, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 94590.37109375, "memory.current_MiB": 28.25, "memory.peak_MiB": 30.40234375, "gpu_total_memory_util_clock_power": "9569, 0, 210, 25.33" }, "rows": [ { "seconds": 15, "p50_ms": 29.02595279738307, "p95_ms": 35.68060998804867, "samples_ms": [ 35.68060998804867, 29.02595279738307, 28.76228094100952 ], "texts": [ "帮我导航到内湖科学园区，然后选最快的路线，不要走高架桥，接着把冷气打开，温度调到24度，风量开大。帮我导航到内湖科学园区，然后选最快的路线。不要走。", "帮我导航到内湖科学园区，然后选最快的路线，不要走高架桥，接着把冷气打开，温度调到24度，风量开大。帮我导航到内湖科学园区，然后选最快的路线。不要走。", "帮我导航到内湖科学园区，然后选最快的路线，不要走高架桥，接着把冷气打开，温度调到24度，风量开大。帮我导航到内湖科学园区，然后选最快的路线。不要走。" ], "memory": { "VmHWM_MiB": 1358.328125, "VmRSS_MiB": 1324.21875, "Pss_MiB": 1319.6767578125, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93389.7421875, "memory.current_MiB": 1090.578125, "memory.peak_MiB": 1179.39453125, "gpu_total_memory_util_clock_power": "11006, 0, 2535, 67.55" }, "audio_sha256": "4950a411664da365f4b8cfe13838f881430f01f2fed8b2c3335719f6b791ff9f" } ], "load_s": 1.240162475965917, "ready": { "VmHWM_MiB": 1358.328125, "VmRSS_MiB": 1270.671875, "Pss_MiB": 1266.1767578125, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93406.62109375, "memory.current_MiB": 1083.6875, "memory.peak_MiB": 1179.39453125, "gpu_total_memory_util_clock_power": "10996, 0, 2535, 64.88" }, "after_warmup": { "VmHWM_MiB": 1358.328125, "VmRSS_MiB": 1324.12890625, "Pss_MiB": 1319.5869140625, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93390.21875, "memory.current_MiB": 1090.21875, "memory.peak_MiB": 1179.39453125, "gpu_total_memory_util_clock_power": "11006, 0, 2535, 65.19" }, "end": { "VmHWM_MiB": 1358.328125, "VmRSS_MiB": 1324.21875, "Pss_MiB": 1319.6767578125, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93389.7421875, "memory.current_MiB": 1090.4609375, "memory.peak_MiB": 1179.39453125, "gpu_total_memory_util_clock_power": "11006, 0, 2535, 67.55" } }, "recognizer-trt-15s": { "arguments": { "self_check": false, "mode": "recognizer", "provider": "trt", "model": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/model.onnx", "tokens": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/tokens.txt", "cache": "/bench/cache", "audio": "/bench/audio", "length": 15, "frames": "50,100,250", "warmup": 10, "repeats": 3, "threads": 2, "profile": false, "dump_context": null, "out": "/bench/4090/recognizer-trt-15s.json" }, "utc": "2026-09-05T13:08:32Z", "machine": "x86_64", "ort": "1.23.2", "numpy": "2.2.6", "model_sha256": "719559b0512db928e18eb3c130a122a55dc10166fb6bd2c8c90f2b6f51064994", "start": { "VmHWM_MiB": 46.9453125, "VmRSS_MiB": 45.875, "Pss_MiB": 45.87109375, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 94762.30859375, "memory.current_MiB": 28.2734375, "memory.peak_MiB": 30.54296875, "gpu_total_memory_util_clock_power": "9569, 0, 210, 24.78" }, "rows": [ { "seconds": 15, "p50_ms": 28.46145606599748, "p95_ms": 28.62507291138172, "samples_ms": [ 28.46145606599748, 28.62507291138172, 28.0569139868021 ], "texts": [ "帮我导航到内湖科学园区，然后选最快的路线，不要走高架桥，接着把冷气打开，温度调到24度，风量开大。帮我导航到内湖科学园区，然后选最快的路线。不要走。", "帮我导航到内湖科学园区，然后选最快的路线，不要走高架桥，接着把冷气打开，温度调到24度，风量开大。帮我导航到内湖科学园区，然后选最快的路线。不要走。", "帮我导航到内湖科学园区，然后选最快的路线，不要走高架桥，接着把冷气打开，温度调到24度，风量开大。帮我导航到内湖科学园区，然后选最快的路线。不要走。" ], "memory": { "VmHWM_MiB": 9418.8984375, "VmRSS_MiB": 7546.0625, "Pss_MiB": 7541.5771484375, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 88685.19921875, "memory.current_MiB": 5450.85546875, "memory.peak_MiB": 7340.00390625, "gpu_total_memory_util_clock_power": "11128, 1, 2535, 70.03" }, "audio_sha256": "4950a411664da365f4b8cfe13838f881430f01f2fed8b2c3335719f6b791ff9f" } ], "load_s": 5.809297051047906, "ready": { "VmHWM_MiB": 9418.8984375, "VmRSS_MiB": 7536.15234375, "Pss_MiB": 7531.6669921875, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 88234.67578125, "memory.current_MiB": 5448.6328125, "memory.peak_MiB": 7340.00390625, "gpu_total_memory_util_clock_power": "11076, 1, 2535, 66.35" }, "after_warmup": { "VmHWM_MiB": 9418.8984375, "VmRSS_MiB": 7546.0625, "Pss_MiB": 7541.5771484375, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 88261.5, "memory.current_MiB": 5451.23046875, "memory.peak_MiB": 7340.00390625, "gpu_total_memory_util_clock_power": "11096, 1, 2535, 66.82" }, "end": { "VmHWM_MiB": 9418.8984375, "VmRSS_MiB": 7546.0625, "Pss_MiB": 7541.5771484375, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 88669.49609375, "memory.current_MiB": 5450.92578125, "memory.peak_MiB": 7340.00390625, "gpu_total_memory_util_clock_power": "11128, 1, 2535, 70.03" } }, "recognizer-trt-6s": { "arguments": { "self_check": false, "mode": "recognizer", "provider": "trt", "model": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/model.onnx", "tokens": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/tokens.txt", "cache": "/bench/cache", "audio": "/bench/audio", "length": 6, "frames": "50,100,250", "warmup": 10, "repeats": 3, "threads": 2, "profile": false, "dump_context": null, "out": "/bench/4090/recognizer-trt-6s.json" }, "utc": "2026-09-05T13:07:41Z", "machine": "x86_64", "ort": "1.23.2", "numpy": "2.2.6", "model_sha256": "719559b0512db928e18eb3c130a122a55dc10166fb6bd2c8c90f2b6f51064994", "start": { "VmHWM_MiB": 46.64453125, "VmRSS_MiB": 45.69921875, "Pss_MiB": 45.6953125, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93011.8671875, "memory.current_MiB": 28.27734375, "memory.peak_MiB": 30.32421875, "gpu_total_memory_util_clock_power": "9569, 0, 2535, 65.26" }, "rows": [ { "seconds": 6, "p50_ms": 14.208432985469699, "p95_ms": 16.890024999156594, "samples_ms": [ 16.890024999156594, 14.030528021976352, 14.208432985469699 ], "texts": [ "帮我导航到内湖科学园区，然后选最快的路线，不要走高架桥。", "帮我导航到内湖科学园区，然后选最快的路线，不要走高架桥。", "帮我导航到内湖科学园区，然后选最快的路线，不要走高架桥。" ], "memory": { "VmHWM_MiB": 9418.51171875, "VmRSS_MiB": 7545.78125, "Pss_MiB": 7541.2958984375, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 87763.0546875, "memory.current_MiB": 5451.08984375, "memory.peak_MiB": 7339.99609375, "gpu_total_memory_util_clock_power": "11112, 1, 2535, 66.99" }, "audio_sha256": "9566be6ccf6c7307830514e0fe5be86250c493efaef2800f27bfccc1ca1850b0" } ], "load_s": 5.8533999510109425, "ready": { "VmHWM_MiB": 9418.51171875, "VmRSS_MiB": 7535.75390625, "Pss_MiB": 7531.2685546875, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 87766.51171875, "memory.current_MiB": 5448.91015625, "memory.peak_MiB": 7339.99609375, "gpu_total_memory_util_clock_power": "11076, 1, 2535, 66.01" }, "after_warmup": { "VmHWM_MiB": 9418.51171875, "VmRSS_MiB": 7545.78125, "Pss_MiB": 7541.2958984375, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 87770.63671875, "memory.current_MiB": 5451.01171875, "memory.peak_MiB": 7339.99609375, "gpu_total_memory_util_clock_power": "11096, 1, 2535, 65.96" }, "end": { "VmHWM_MiB": 9418.51171875, "VmRSS_MiB": 7545.78125, "Pss_MiB": 7541.2958984375, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 87969.4765625, "memory.current_MiB": 5451.40625, "memory.peak_MiB": 7339.99609375, "gpu_total_memory_util_clock_power": "11112, 1, 2535, 67.51" } }, "services": { "service-cuda-1": { "environment": { "target": "ws://127.0.0.1:19242/stt/sensevoice", "uri": "ws://127.0.0.1:19242/stt/sensevoice", "warmup_runs": 3, "single_connection_series": true }, "generated": "2026-09-05T14:00:50", "results": [ { "feed": "bulk", "target_s": 3.0, "duration_s": 3.0, "status": "ok", "server_latency_ms": 22.45, "server_rtf": 0.01, "server_ttft_ms": 31.97, "client_final_latency_ms": 28.46, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.03, "text": "幫我導航到內湖科學園區，然後。", "error": null }, { "feed": "bulk", "target_s": 6.0, "duration_s": 6.0, "status": "ok", "server_latency_ms": 32.92, "server_rtf": 0.006, "server_ttft_ms": 23.45, "client_final_latency_ms": 38.64, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.05, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。", "error": null }, { "feed": "bulk", "target_s": 15.0, "duration_s": 15.0, "status": "ok", "server_latency_ms": 99.66, "server_rtf": 0.004, "server_ttft_ms": 25.28, "client_final_latency_ms": 105.85, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.13, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開大。幫我導航到內湖科學園區，然後選最快的路線。不要走。", "error": null }, { "feed": "realtime", "target_s": 3.0, "duration_s": 3.0, "status": "ok", "server_latency_ms": 42.77, "server_rtf": 0.05, "server_ttft_ms": 40.76, "client_final_latency_ms": 46.04, "interim_count": 6, "max_interim_gap_s": 0.51, "feed_secs": 3.14, "text": "幫我導航到內湖科學園區，然後。", "error": null }, { "feed": "realtime", "target_s": 6.0, "duration_s": 6.0, "status": "ok", "server_latency_ms": 23.32, "server_rtf": 0.051, "server_ttft_ms": 36.03, "client_final_latency_ms": 38.31, "interim_count": 10, "max_interim_gap_s": 1.093, "feed_secs": 6.29, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。", "error": null }, { "feed": "realtime", "target_s": 15.0, "duration_s": 15.0, "status": "ok", "server_latency_ms": 74.55, "server_rtf": 0.053, "server_ttft_ms": 38.67, "client_final_latency_ms": 78.33, "interim_count": 27, "max_interim_gap_s": 1.253, "feed_secs": 15.75, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開大。幫我導航到內湖科學園區，然後選最快的路線。不要走。", "error": null } ] }, "service-trt-1": { "environment": { "target": "ws://127.0.0.1:19242/stt/sensevoice", "uri": "ws://127.0.0.1:19242/stt/sensevoice", "warmup_runs": 3, "single_connection_series": true }, "generated": "2026-09-05T14:04:46", "results": [ { "feed": "bulk", "target_s": 3.0, "duration_s": 3.0, "status": "ok", "server_latency_ms": 13.18, "server_rtf": 0.004, "server_ttft_ms": 12.2, "client_final_latency_ms": 20.39, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.02, "text": "幫我導航到內湖科學園區，然後。", "error": null }, { "feed": "bulk", "target_s": 6.0, "duration_s": 6.0, "status": "ok", "server_latency_ms": 19.47, "server_rtf": 0.003, "server_ttft_ms": 14.92, "client_final_latency_ms": 31.78, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.06, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。", "error": null }, { "feed": "bulk", "target_s": 15.0, "duration_s": 15.0, "status": "ok", "server_latency_ms": 75.1, "server_rtf": 0.002, "server_ttft_ms": 15.28, "client_final_latency_ms": 128.63, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.11, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開大。幫我導航到內湖科學園區，然後選最快的路線。不要走。", "error": null }, { "feed": "realtime", "target_s": 3.0, "duration_s": 3.0, "status": "ok", "server_latency_ms": 17.98, "server_rtf": 0.02, "server_ttft_ms": 12.54, "client_final_latency_ms": 37.62, "interim_count": 6, "max_interim_gap_s": 0.525, "feed_secs": 3.16, "text": "幫我導航到內湖科學園區，然後。", "error": null }, { "feed": "realtime", "target_s": 6.0, "duration_s": 6.0, "status": "ok", "server_latency_ms": 13.37, "server_rtf": 0.023, "server_ttft_ms": 11.8, "client_final_latency_ms": 25.71, "interim_count": 12, "max_interim_gap_s": 1.032, "feed_secs": 6.31, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。", "error": null }, { "feed": "realtime", "target_s": 15.0, "duration_s": 15.0, "status": "ok", "server_latency_ms": 53.66, "server_rtf": 0.034, "server_ttft_ms": 13.24, "client_final_latency_ms": 55.5, "interim_count": 28, "max_interim_gap_s": 1.03, "feed_secs": 15.79, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開大。幫我導航到內湖科學園區，然後選最快的路線。不要走。", "error": null } ] }, "service-trt-3": { "environment": { "target": "ws://127.0.0.1:19242/stt/sensevoice", "uri": "ws://127.0.0.1:19242/stt/sensevoice", "warmup_runs": 3, "single_connection_series": true }, "generated": "2026-09-05T14:07:12", "results": [ { "feed": "bulk", "target_s": 3.0, "duration_s": 3.0, "status": "ok", "server_latency_ms": 13.34, "server_rtf": 0.005, "server_ttft_ms": 13.7, "client_final_latency_ms": 31.61, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.04, "text": "幫我導航到內湖科學園區，然後。", "error": null }, { "feed": "bulk", "target_s": 6.0, "duration_s": 6.0, "status": "ok", "server_latency_ms": 17.97, "server_rtf": 0.003, "server_ttft_ms": 11.39, "client_final_latency_ms": 29.91, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.04, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。", "error": null }, { "feed": "bulk", "target_s": 15.0, "duration_s": 15.0, "status": "ok", "server_latency_ms": 108.68, "server_rtf": 0.002, "server_ttft_ms": 20.59, "client_final_latency_ms": 114.93, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.13, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開大。幫我導航到內湖科學園區，然後選最快的路線。不要走。", "error": null }, { "feed": "realtime", "target_s": 3.0, "duration_s": 3.0, "status": "ok", "server_latency_ms": 21.24, "server_rtf": 0.024, "server_ttft_ms": 10.27, "client_final_latency_ms": 26.31, "interim_count": 6, "max_interim_gap_s": 0.99, "feed_secs": 3.13, "text": "幫我導航到內湖科學園區，然後。", "error": null }, { "feed": "realtime", "target_s": 6.0, "duration_s": 6.0, "status": "ok", "server_latency_ms": 21.1, "server_rtf": 0.029, "server_ttft_ms": 20.32, "client_final_latency_ms": 44.2, "interim_count": 12, "max_interim_gap_s": 1.011, "feed_secs": 6.3, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。", "error": null }, { "feed": "realtime", "target_s": 15.0, "duration_s": 15.0, "status": "ok", "server_latency_ms": 54.44, "server_rtf": 0.032, "server_ttft_ms": 23.32, "client_final_latency_ms": 56.21, "interim_count": 28, "max_interim_gap_s": 1.351, "feed_secs": 15.77, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開大。幫我導航到內湖科學園區，然後選最快的路線。不要走。", "error": null } ] }, "service-cuda-2": { "environment": { "target": "ws://127.0.0.1:19242/stt/sensevoice", "uri": "ws://127.0.0.1:19242/stt/sensevoice", "warmup_runs": 3, "single_connection_series": true }, "generated": "2026-09-05T14:01:59", "results": [ { "feed": "bulk", "target_s": 3.0, "duration_s": 3.0, "status": "ok", "server_latency_ms": 23.1, "server_rtf": 0.011, "server_ttft_ms": 35.58, "client_final_latency_ms": 28.6, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.04, "text": "幫我導航到內湖科學園區，然後。", "error": null }, { "feed": "bulk", "target_s": 6.0, "duration_s": 6.0, "status": "ok", "server_latency_ms": 26.65, "server_rtf": 0.006, "server_ttft_ms": 31.79, "client_final_latency_ms": 32.25, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.06, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。", "error": null }, { "feed": "bulk", "target_s": 15.0, "duration_s": 15.0, "status": "ok", "server_latency_ms": 72.64, "server_rtf": 0.003, "server_ttft_ms": 21.74, "client_final_latency_ms": 78.52, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.12, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開大。幫我導航到內湖科學園區，然後選最快的路線。不要走。", "error": null }, { "feed": "realtime", "target_s": 3.0, "duration_s": 3.0, "status": "ok", "server_latency_ms": 28.21, "server_rtf": 0.044, "server_ttft_ms": 19.63, "client_final_latency_ms": 33.03, "interim_count": 7, "max_interim_gap_s": 0.523, "feed_secs": 3.13, "text": "幫我導航到內湖科學園區，然後。", "error": null }, { "feed": "realtime", "target_s": 6.0, "duration_s": 6.0, "status": "ok", "server_latency_ms": 30.55, "server_rtf": 0.056, "server_ttft_ms": 32.08, "client_final_latency_ms": 33.05, "interim_count": 11, "max_interim_gap_s": 1.039, "feed_secs": 6.32, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。", "error": null }, { "feed": "realtime", "target_s": 15.0, "duration_s": 15.0, "status": "ok", "server_latency_ms": 67.57, "server_rtf": 0.051, "server_ttft_ms": 28.99, "client_final_latency_ms": 91.67, "interim_count": 27, "max_interim_gap_s": 1.486, "feed_secs": 15.76, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開大。幫我導航到內湖科學園區，然後選最快的路線。不要走。", "error": null } ] }, "service-cuda-3": { "environment": { "target": "ws://127.0.0.1:19242/stt/sensevoice", "uri": "ws://127.0.0.1:19242/stt/sensevoice", "warmup_runs": 3, "single_connection_series": true }, "generated": "2026-09-05T14:03:07", "results": [ { "feed": "bulk", "target_s": 3.0, "duration_s": 3.0, "status": "ok", "server_latency_ms": 21.72, "server_rtf": 0.011, "server_ttft_ms": 34.91, "client_final_latency_ms": 27.42, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.04, "text": "幫我導航到內湖科學園區，然後。", "error": null }, { "feed": "bulk", "target_s": 6.0, "duration_s": 6.0, "status": "ok", "server_latency_ms": 26.83, "server_rtf": 0.006, "server_ttft_ms": 28.07, "client_final_latency_ms": 32.86, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.05, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。", "error": null }, { "feed": "bulk", "target_s": 15.0, "duration_s": 15.0, "status": "ok", "server_latency_ms": 69.4, "server_rtf": 0.003, "server_ttft_ms": 21.9, "client_final_latency_ms": 75.41, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.12, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開。幫我導航到內湖科學園區，然後選最快的路線，不要走。", "error": null }, { "feed": "realtime", "target_s": 3.0, "duration_s": 3.0, "status": "ok", "server_latency_ms": 21.07, "server_rtf": 0.046, "server_ttft_ms": 25.25, "client_final_latency_ms": 28.77, "interim_count": 7, "max_interim_gap_s": 0.528, "feed_secs": 3.15, "text": "幫我導航到內湖科學園區，然後。", "error": null }, { "feed": "realtime", "target_s": 6.0, "duration_s": 6.0, "status": "ok", "server_latency_ms": 33.77, "server_rtf": 0.05, "server_ttft_ms": 20.86, "client_final_latency_ms": 36.3, "interim_count": 11, "max_interim_gap_s": 1.028, "feed_secs": 6.29, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。", "error": null }, { "feed": "realtime", "target_s": 15.0, "duration_s": 15.0, "status": "ok", "server_latency_ms": 85.37, "server_rtf": 0.051, "server_ttft_ms": 20.11, "client_final_latency_ms": 88.97, "interim_count": 28, "max_interim_gap_s": 1.049, "feed_secs": 15.74, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開大。幫我導航到內湖科學園區，然後選最快的路線。不要走。", "error": null } ] }, "service-trt-2": { "environment": { "target": "ws://127.0.0.1:19242/stt/sensevoice", "uri": "ws://127.0.0.1:19242/stt/sensevoice", "warmup_runs": 3, "single_connection_series": true }, "generated": "2026-09-05T14:06:01", "results": [ { "feed": "bulk", "target_s": 3.0, "duration_s": 3.0, "status": "ok", "server_latency_ms": 16.32, "server_rtf": 0.005, "server_ttft_ms": 12.31, "client_final_latency_ms": 22.27, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.02, "text": "幫我導航到內湖科學園區，然後。", "error": null }, { "feed": "bulk", "target_s": 6.0, "duration_s": 6.0, "status": "ok", "server_latency_ms": 16.9, "server_rtf": 0.003, "server_ttft_ms": 11.77, "client_final_latency_ms": 32.43, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.06, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。", "error": null }, { "feed": "bulk", "target_s": 15.0, "duration_s": 15.0, "status": "ok", "server_latency_ms": 73.87, "server_rtf": 0.002, "server_ttft_ms": 17.24, "client_final_latency_ms": 80.45, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.12, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開大。幫我導航到內湖科學園區，然後選最快的路線。不要走。", "error": null }, { "feed": "realtime", "target_s": 3.0, "duration_s": 3.0, "status": "ok", "server_latency_ms": 11.29, "server_rtf": 0.027, "server_ttft_ms": 11.0, "client_final_latency_ms": 17.98, "interim_count": 7, "max_interim_gap_s": 0.528, "feed_secs": 3.16, "text": "幫我導航到內湖科學園區，然後。", "error": null }, { "feed": "realtime", "target_s": 6.0, "duration_s": 6.0, "status": "ok", "server_latency_ms": 12.29, "server_rtf": 0.02, "server_ttft_ms": 10.5, "client_final_latency_ms": 25.26, "interim_count": 10, "max_interim_gap_s": 1.543, "feed_secs": 6.28, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。", "error": null }, { "feed": "realtime", "target_s": 15.0, "duration_s": 15.0, "status": "ok", "server_latency_ms": 56.77, "server_rtf": 0.035, "server_ttft_ms": 28.08, "client_final_latency_ms": 58.73, "interim_count": 28, "max_interim_gap_s": 1.027, "feed_secs": 15.74, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開大。幫我導航到內湖科學園區，然後選最快的路線。不要走。", "error": null } ] } } }, "orin": { "recognizer-trt-3s": { "arguments": { "self_check": false, "mode": "recognizer", "provider": "trt", "model": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/model.onnx", "tokens": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/tokens.txt", "cache": "/bench/cache", "audio": "/bench/audio", "length": 3, "frames": "50,100,250", "warmup": 10, "repeats": 3, "threads": 2, "profile": false, "dump_context": null, "out": "/bench/result.json" }, "utc": "2026-09-05T13:37:48Z", "machine": "aarch64", "ort": "1.24.0", "numpy": "1.26.4", "model_sha256": "719559b0512db928e18eb3c130a122a55dc10166fb6bd2c8c90f2b6f51064994", "start": { "VmHWM_MiB": 49.12109375, "VmRSS_MiB": 47.2421875, "Pss_MiB": 34.7998046875, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 19147.0078125, "memory.current_MiB": 994.9765625 }, "rows": [ { "seconds": 3, "p50_ms": 28.385848738253117, "p95_ms": 28.61195197328925, "samples_ms": [ 28.61195197328925, 28.385848738253117, 28.341433964669704 ], "texts": [ "帮我导航到内湖科学园区，然后。", "帮我导航到内湖科学园区，然后。", "帮我导航到内湖科学园区，然后。" ], "memory": { "VmHWM_MiB": 8597.359375, "VmRSS_MiB": 6977.66015625, "Pss_MiB": 5510.9384765625, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 12951.3359375, "memory.current_MiB": 7291.09375 }, "audio_sha256": "11034b66e2ae87bdec98dfe7401642387b85262b2e9c222b9e201030dc397a1c" } ], "load_s": 16.197763889096677, "ready": { "VmHWM_MiB": 8597.359375, "VmRSS_MiB": 6948.2890625, "Pss_MiB": 5505.3720703125, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 12972.84765625, "memory.current_MiB": 7267.9375 }, "after_warmup": { "VmHWM_MiB": 8597.359375, "VmRSS_MiB": 6977.66015625, "Pss_MiB": 5510.9384765625, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 12951.265625, "memory.current_MiB": 7291.05078125 }, "end": { "VmHWM_MiB": 8597.359375, "VmRSS_MiB": 6977.66015625, "Pss_MiB": 5510.9384765625, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 12951.10546875, "memory.current_MiB": 7291.09375 } }, "recognizer-cuda-3s": { "arguments": { "self_check": false, "mode": "recognizer", "provider": "cuda", "model": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/model.onnx", "tokens": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/tokens.txt", "cache": "/bench/cache", "audio": "/bench/audio", "length": 3, "frames": "50,100,250", "warmup": 10, "repeats": 3, "threads": 2, "profile": false, "dump_context": null, "out": "/bench/result.json" }, "utc": "2026-09-05T13:34:58Z", "machine": "aarch64", "ort": "1.24.0", "numpy": "1.26.4", "model_sha256": "719559b0512db928e18eb3c130a122a55dc10166fb6bd2c8c90f2b6f51064994", "start": { "VmHWM_MiB": 51.1328125, "VmRSS_MiB": 49.25390625, "Pss_MiB": 36.8818359375, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 20043.2421875, "memory.current_MiB": 997.8046875 }, "rows": [ { "seconds": 3, "p50_ms": 45.1835123822093, "p95_ms": 45.341553166508675, "samples_ms": [ 45.16981588676572, 45.1835123822093, 45.341553166508675 ], "texts": [ "帮我导航到内湖科学园区，然后。", "帮我导航到内湖科学园区，然后。", "帮我导航到内湖科学园区，然后。" ], "memory": { "VmHWM_MiB": 2362.70703125, "VmRSS_MiB": 2362.703125, "Pss_MiB": 1152.6904296875, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 17863.05078125, "memory.current_MiB": 3169.73828125 }, "audio_sha256": "11034b66e2ae87bdec98dfe7401642387b85262b2e9c222b9e201030dc397a1c" } ], "load_s": 2.936227814294398, "ready": { "VmHWM_MiB": 2312.17578125, "VmRSS_MiB": 2312.17578125, "Pss_MiB": 1128.3603515625, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 17873.203125, "memory.current_MiB": 3155.11328125 }, "after_warmup": { "VmHWM_MiB": 2362.70703125, "VmRSS_MiB": 2362.70703125, "Pss_MiB": 1152.6884765625, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 17863.05078125, "memory.current_MiB": 3169.73828125 }, "end": { "VmHWM_MiB": 2362.70703125, "VmRSS_MiB": 2362.703125, "Pss_MiB": 1152.6904296875, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 17863.05078125, "memory.current_MiB": 3169.73828125 } }, "environment": { "platform": "Linux-5.15.148-tegra-aarch64-with-glibc2.35", "python": "3.10.12", "packages": { "onnxruntime-gpu": "1.24.0", "numpy": "1.26.4", "sherpa-onnx": "1.13.0+cuda", "torch": "2.8.0", "kitt-core": "0.3.7", "pipecat-ai": "0.0.108" }, "sha256": { "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/model.onnx": "719559b0512db928e18eb3c130a122a55dc10166fb6bd2c8c90f2b6f51064994", "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/tokens.txt": "f449eb28dc567533d7fa59be34e2abca8784f771850c78a47fb731a31429a1dc", "/app/src/asr/batch_asr_manager.py": "fb9a00f6cdbc5bd972d3bb82c85b44a70277a593d1694460071bac124ef25815", "/app/src/kitt_stt_service.py": "5283feaee7a213feb529afe1d98308c7fec447ccd18aa9f737a7632ea08b83bc" } }, "service-cuda-end-memory": { "utc": 1788617445.0855966, "VmHWM_MiB": 4214.390625, "VmRSS_MiB": 4174.65234375, "Pss_MiB": 1908.93359375, "memory.current_MiB": 4118.53515625 }, "graph-trt": { "arguments": { "self_check": false, "mode": "graph", "provider": "trt", "model": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/model.onnx", "tokens": null, "cache": "/bench/cache", "audio": null, "length": 3, "frames": "50,100,250", "warmup": 10, "repeats": 50, "threads": 2, "profile": false, "dump_context": null, "out": "/bench/result.json" }, "utc": "2026-09-05T13:40:29Z", "machine": "aarch64", "ort": "1.24.0", "numpy": "1.26.4", "model_sha256": "719559b0512db928e18eb3c130a122a55dc10166fb6bd2c8c90f2b6f51064994", "start": { "VmHWM_MiB": 46.06640625, "VmRSS_MiB": 44.41796875, "Pss_MiB": 34.568359375, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 18593.3125, "memory.current_MiB": 1008.08203125 }, "rows": [ { "frames": 50, "p50_ms": 19.858619896695018, "p95_ms": 19.920490216463804, "samples_ms": [ 19.912714138627052, 19.950921181589365, 19.94298491626978, 19.911241717636585, 19.86532425507903, 19.865259993821383, 19.920490216463804, 19.88893887028098, 19.902025815099478, 19.89316288381815, 19.905610010027885, 19.872714765369892, 19.875403959304094, 19.85070016235113, 19.882666878402233, 19.842605106532574, 19.854316022247076, 19.831916783005, 19.851500168442726, 19.83166066929698, 19.86922835931182, 19.82762897387147, 19.864588044583797, 19.848620053380728, 19.87681072205305, 19.854635931551456, 19.84253991395235, 19.833837170153856, 19.85645992681384, 19.824716728180647, 19.91229923442006, 19.844493363052607, 19.915785640478134, 19.850573036819696, 19.874636083841324, 19.847148098051548, 19.850059878081083, 19.879691302776337, 19.86769074574113, 19.87326005473733, 19.852907862514257, 19.860779866576195, 19.845292903482914, 19.81140486896038, 19.82574164867401, 19.812558311969042, 19.870092160999775, 19.807182252407074, 19.855821039527655, 19.81070125475526 ], "memory": { "VmHWM_MiB": 8586.3203125, "VmRSS_MiB": 6974.390625, "Pss_MiB": 5504.3388671875, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 13452.703125, "memory.current_MiB": 6448.82421875 }, "finite": true, "input_sha256": "472ba979ad46ce462cc2cceeb0d218314362876e03b63f06a136a1c19b3dbab7" }, { "frames": 100, "p50_ms": 20.036390284076333, "p95_ms": 20.10256377980113, "samples_ms": [ 20.072197075933218, 20.187233109027147, 20.089827943593264, 20.08419716730714, 20.079493056982756, 20.024423021823168, 20.1111719943583, 20.10256377980113, 20.05379693582654, 20.03722172230482, 20.053190179169178, 20.070981234312057, 20.061605144292116, 20.021382719278336, 20.03805385902524, 20.028389990329742, 20.052645821124315, 20.047142170369625, 20.060325972735882, 20.03722218796611, 20.024999044835567, 20.020999014377594, 20.039654336869717, 20.031366031616926, 20.06307803094387, 20.012584049254656, 20.01421619206667, 19.997959956526756, 20.01216821372509, 20.011751912534237, 20.06381331011653, 19.987144973129034, 20.026438869535923, 20.005512051284313, 20.057189743965864, 20.018599927425385, 20.0549503788352, 20.014632027596235, 20.02496737986803, 19.987881183624268, 20.052038598805666, 20.035558845847845, 20.021831151098013, 20.039366092532873, 20.065669901669025, 19.99562419950962, 20.01127228140831, 19.993416033685207, 20.026918966323137, 19.98301688581705 ], "memory": { "VmHWM_MiB": 8586.3203125, "VmRSS_MiB": 6990.34375, "Pss_MiB": 5504.5546875, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 13436.65625, "memory.current_MiB": 6465.07421875 }, "finite": true, "input_sha256": "dda0d977bf96004405c6cbeac7cd57cfb466001cc340939617503e06dff71881" }, { "frames": 250, "p50_ms": 38.81613560952246, "p95_ms": 38.88523671776056, "samples_ms": [ 38.79992803558707, 38.92027586698532, 38.824919145554304, 38.861429784446955, 38.75890513882041, 38.829046767205, 38.825302850455046, 38.787481375038624, 38.840183056890965, 38.88126974925399, 38.81698288023472, 38.81336795166135, 38.79426419734955, 38.86299813166261, 38.8082480058074, 38.81490370258689, 38.822358939796686, 38.80434390157461, 38.80415204912424, 38.83560700342059, 38.803352043032646, 38.777528796344995, 38.826551754027605, 38.87102892622352, 38.7193551287055, 38.88523671776056, 38.84197538718581, 38.832024205476046, 38.80495298653841, 38.80674485117197, 38.804665207862854, 38.815288338810205, 38.77138579264283, 38.79487281665206, 38.799993228167295, 38.833752274513245, 38.83064817637205, 38.8538152910769, 38.76092191785574, 38.83567079901695, 38.828855846077204, 38.80728781223297, 38.9167251996696, 38.86670991778374, 38.81797706708312, 38.821880239993334, 38.77509804442525, 38.80978422239423, 38.80866523832083, 38.78460172563791 ], "memory": { "VmHWM_MiB": 8586.3203125, "VmRSS_MiB": 7022.33984375, "Pss_MiB": 5504.4775390625, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 13402.98828125, "memory.current_MiB": 6497.13671875 }, "finite": true, "input_sha256": "726cc736d46a4b8ca3c2d2831abc8c8e3a59489168194774d072d5fc1aa502b5" } ], "load_s": 14.90848413715139, "ready": { "VmHWM_MiB": 8586.3203125, "VmRSS_MiB": 6953.2109375, "Pss_MiB": 5502.65625, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 13472.92578125, "memory.current_MiB": 6428.82421875 }, "providers": [ "TensorrtExecutionProvider", "CUDAExecutionProvider", "CPUExecutionProvider" ], "end": { "VmHWM_MiB": 8586.3203125, "VmRSS_MiB": 7022.33984375, "Pss_MiB": 5504.4853515625, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 13402.98828125, "memory.current_MiB": 6497.13671875 } }, "service-cuda-ready-memory": { "utc": 1788617161.5043106, "VmHWM_MiB": 4147.34375, "VmRSS_MiB": 4035.86328125, "Pss_MiB": 1792.1669921875, "memory.current_MiB": 3976.8515625 }, "graph-cuda": { "arguments": { "self_check": false, "mode": "graph", "provider": "cuda", "model": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/model.onnx", "tokens": null, "cache": "/bench/cache", "audio": null, "length": 3, "frames": "50,100,250", "warmup": 10, "repeats": 50, "threads": 2, "profile": false, "dump_context": null, "out": "/bench/result.json" }, "utc": "2026-09-05T13:37:00Z", "machine": "aarch64", "ort": "1.24.0", "numpy": "1.26.4", "model_sha256": "719559b0512db928e18eb3c130a122a55dc10166fb6bd2c8c90f2b6f51064994", "start": { "VmHWM_MiB": 48.87890625, "VmRSS_MiB": 47.23046875, "Pss_MiB": 34.7880859375, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 18908.6875, "memory.current_MiB": 995.0078125 }, "rows": [ { "frames": 50, "p50_ms": 37.34521381556988, "p95_ms": 37.70603286102414, "samples_ms": [ 37.66494616866112, 37.356380838900805, 37.37545199692249, 37.15529991313815, 37.5465820543468, 37.59672399610281, 37.576116155833006, 37.402810994535685, 37.5421978533268, 37.27126494050026, 36.82155394926667, 37.132902070879936, 36.840945947915316, 37.32579108327627, 37.285856902599335, 37.42421884089708, 37.70603286102414, 37.5781012699008, 35.79068090766668, 37.05236176028848, 37.24128194153309, 37.359326146543026, 37.380957044661045, 37.54075895994902, 37.66283392906189, 37.17418015003204, 37.54034265875816, 37.4896889552474, 37.14598994702101, 37.314462941139936, 37.070760037750006, 37.495736964046955, 36.98676312342286, 36.826514173299074, 37.19875495880842, 37.45154710486531, 38.009926211088896, 37.67950600013137, 37.98570344224572, 37.04116167500615, 37.519289180636406, 37.537079770118, 37.263970356434584, 37.01057191938162, 36.88551997765899, 37.33404679223895, 36.87134524807334, 36.615259014070034, 37.520761135965586, 37.12122282013297 ], "memory": { "VmHWM_MiB": 2358.8828125, "VmRSS_MiB": 2358.84765625, "Pss_MiB": 1154.9814453125, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 18129.6796875, "memory.current_MiB": 2064.81640625 }, "finite": true, "input_sha256": "472ba979ad46ce462cc2cceeb0d218314362876e03b63f06a136a1c19b3dbab7" }, { "frames": 100, "p50_ms": 38.48082176409662, "p95_ms": 52.81285662204027, "samples_ms": [ 38.875078316777945, 38.23427110910416, 38.975970819592476, 38.81181590259075, 54.100616835057735, 52.81285662204027, 54.10061590373516, 43.62232377752662, 41.754265781491995, 41.34541703388095, 38.18083228543401, 38.649966940283775, 38.81674399599433, 38.700364995747805, 39.126460906118155, 38.789354264736176, 38.46991015598178, 38.45416568219662, 38.491733372211456, 38.16707292571664, 38.25737303122878, 38.424566853791475, 38.44491904601455, 37.851245142519474, 38.57857873663306, 38.64632034674287, 38.666767068207264, 38.63656101748347, 38.57365110889077, 38.56894699856639, 38.13424287363887, 38.5427069850266, 38.59835397452116, 38.125507067888975, 37.83115139231086, 38.130627013742924, 38.155170157551765, 38.32367621362209, 38.3633547462523, 38.21766469627619, 38.33308443427086, 38.021222688257694, 38.51010138168931, 38.365531247109175, 38.44319190829992, 38.125250954180956, 38.42239314690232, 38.064037915319204, 38.15619507804513, 38.54162013158202 ], "memory": { "VmHWM_MiB": 2359.0703125, "VmRSS_MiB": 2359.0703125, "Pss_MiB": 1155.1669921875, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 18128.90234375, "memory.current_MiB": 2064.83203125 }, "finite": true, "input_sha256": "dda0d977bf96004405c6cbeac7cd57cfb466001cc340939617503e06dff71881" }, { "frames": 250, "p50_ms": 45.933760004118085, "p95_ms": 46.03730887174606, "samples_ms": [ 46.02041095495224, 45.99858773872256, 45.96243007108569, 45.94143806025386, 45.90175999328494, 45.919007156044245, 46.01039597764611, 45.89670384302735, 45.987932942807674, 46.00700316950679, 45.994813088327646, 45.91449582949281, 45.99676514044404, 45.83734693005681, 46.08060186728835, 45.9295678883791, 46.037083957344294, 45.88413005694747, 45.877665746957064, 46.08889017254114, 45.989662408828735, 45.90492881834507, 45.96774186939001, 45.99647782742977, 45.96249479800463, 45.857345685362816, 45.896385330706835, 45.887712854892015, 45.92291172593832, 45.850147027522326, 45.86246609687805, 45.90713605284691, 45.8882269449532, 45.87430693209171, 45.77293526381254, 45.9773107431829, 45.87248293682933, 45.93795211985707, 45.95078434795141, 46.03730887174606, 45.91891309246421, 45.951231848448515, 45.9248973056674, 45.98687868565321, 45.88752193376422, 46.00988607853651, 45.90873792767525, 45.9677753970027, 45.90928228572011, 45.939935836941004 ], "memory": { "VmHWM_MiB": 2383.51953125, "VmRSS_MiB": 2383.51953125, "Pss_MiB": 1179.4951171875, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 18103.8125, "memory.current_MiB": 2089.171875 }, "finite": true, "input_sha256": "726cc736d46a4b8ca3c2d2831abc8c8e3a59489168194774d072d5fc1aa502b5" } ], "load_s": 2.839484474156052, "ready": { "VmHWM_MiB": 2308.671875, "VmRSS_MiB": 2308.671875, "Pss_MiB": 1131.1162109375, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 18146.30078125, "memory.current_MiB": 2050.41015625 }, "providers": [ "CUDAExecutionProvider", "CPUExecutionProvider" ], "end": { "VmHWM_MiB": 2383.51953125, "VmRSS_MiB": 2383.51953125, "Pss_MiB": 1179.5029296875, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 18103.8125, "memory.current_MiB": 2089.171875 } }, "service-trt-ready-memory": { "utc": 1788617477.679606, "VmHWM_MiB": 9452.35546875, "VmRSS_MiB": 8723.1484375, "Pss_MiB": 6303.3876953125, "memory.current_MiB": 7114.625 }, "service-trt-end-memory": { "utc": 1788617722.855794, "VmHWM_MiB": 9452.35546875, "VmRSS_MiB": 8895.08203125, "Pss_MiB": 6400.7470703125, "memory.current_MiB": 7285.36328125 }, "recognizer-cuda-6s": { "arguments": { "self_check": false, "mode": "recognizer", "provider": "cuda", "model": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/model.onnx", "tokens": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/tokens.txt", "cache": "/bench/cache", "audio": "/bench/audio", "length": 6, "frames": "50,100,250", "warmup": 10, "repeats": 3, "threads": 2, "profile": false, "dump_context": null, "out": "/bench/result.json" }, "utc": "2026-09-05T13:35:37Z", "machine": "aarch64", "ort": "1.24.0", "numpy": "1.26.4", "model_sha256": "719559b0512db928e18eb3c130a122a55dc10166fb6bd2c8c90f2b6f51064994", "start": { "VmHWM_MiB": 48.8828125, "VmRSS_MiB": 47.234375, "Pss_MiB": 34.8759765625, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 18920.51171875, "memory.current_MiB": 994.99609375 }, "rows": [ { "seconds": 6, "p50_ms": 53.98238077759743, "p95_ms": 77.93578784912825, "samples_ms": [ 77.93578784912825, 53.98238077759743, 53.93521627411246 ], "texts": [ "帮我导航到内湖科学园区，然后选最快的路线，不要走高架桥。", "帮我导航到内湖科学园区，然后选最快的路线，不要走高架桥。", "帮我导航到内湖科学园区，然后选最快的路线，不要走高架桥。" ], "memory": { "VmHWM_MiB": 2382.0546875, "VmRSS_MiB": 2382.0546875, "Pss_MiB": 1171.7529296875, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 17858.95703125, "memory.current_MiB": 2066.3046875 }, "audio_sha256": "9566be6ccf6c7307830514e0fe5be86250c493efaef2800f27bfccc1ca1850b0" } ], "load_s": 2.6790056759491563, "ready": { "VmHWM_MiB": 2379.671875, "VmRSS_MiB": 2331.1953125, "Pss_MiB": 1147.2939453125, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 17861.96484375, "memory.current_MiB": 2059.4140625 }, "after_warmup": { "VmHWM_MiB": 2381.72265625, "VmRSS_MiB": 2381.72265625, "Pss_MiB": 1171.6181640625, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 17859.203125, "memory.current_MiB": 2066.4140625 }, "end": { "VmHWM_MiB": 2382.0546875, "VmRSS_MiB": 2382.0546875, "Pss_MiB": 1171.7529296875, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 17858.95703125, "memory.current_MiB": 2066.3046875 } }, "recognizer-cuda-15s": { "arguments": { "self_check": false, "mode": "recognizer", "provider": "cuda", "model": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/model.onnx", "tokens": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/tokens.txt", "cache": "/bench/cache", "audio": "/bench/audio", "length": 15, "frames": "50,100,250", "warmup": 10, "repeats": 3, "threads": 2, "profile": false, "dump_context": null, "out": "/bench/result.json" }, "utc": "2026-09-05T13:36:21Z", "machine": "aarch64", "ort": "1.24.0", "numpy": "1.26.4", "model_sha256": "719559b0512db928e18eb3c130a122a55dc10166fb6bd2c8c90f2b6f51064994", "start": { "VmHWM_MiB": 51.11328125, "VmRSS_MiB": 49.234375, "Pss_MiB": 36.7880859375, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 18908.2578125, "memory.current_MiB": 996.96484375 }, "rows": [ { "seconds": 15, "p50_ms": 87.55026711151004, "p95_ms": 105.87390791624784, "samples_ms": [ 105.87390791624784, 87.55026711151004, 87.53967611119151 ], "texts": [ "帮我导航到内湖科学园区，然后选最快的路线，不要走高架桥，接着把冷气打开，温度调到24度，风量开大。帮我导航到内湖科学园区，然后选最快的路线。不要走。", "帮我导航到内湖科学园区，然后选最快的路线，不要走高架桥，接着把冷气打开，温度调到24度，风量开大。帮我导航到内湖科学园区，然后选最快的路线。不要走。", "帮我导航到内湖科学园区，然后选最快的路线，不要走高架桥，接着把冷气打开，温度调到24度，风量开大。帮我导航到内湖科学园区，然后选最快的路线。不要走。" ], "memory": { "VmHWM_MiB": 2355.83203125, "VmRSS_MiB": 2342.15625, "Pss_MiB": 1132.0048828125, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 17900.5390625, "memory.current_MiB": 2025.375 }, "audio_sha256": "4950a411664da365f4b8cfe13838f881430f01f2fed8b2c3335719f6b791ff9f" } ], "load_s": 2.673405689187348, "ready": { "VmHWM_MiB": 2355.83203125, "VmRSS_MiB": 2291.3046875, "Pss_MiB": 1107.4912109375, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 17905.31640625, "memory.current_MiB": 2017.859375 }, "after_warmup": { "VmHWM_MiB": 2355.83203125, "VmRSS_MiB": 2341.8359375, "Pss_MiB": 1131.8583984375, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 17900.59375, "memory.current_MiB": 2024.984375 }, "end": { "VmHWM_MiB": 2355.83203125, "VmRSS_MiB": 2342.15625, "Pss_MiB": 1132.0048828125, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 17900.5390625, "memory.current_MiB": 2025.375 } }, "recognizer-trt-15s": { "arguments": { "self_check": false, "mode": "recognizer", "provider": "trt", "model": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/model.onnx", "tokens": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/tokens.txt", "cache": "/bench/cache", "audio": "/bench/audio", "length": 15, "frames": "50,100,250", "warmup": 10, "repeats": 3, "threads": 2, "profile": false, "dump_context": null, "out": "/bench/result.json" }, "utc": "2026-09-05T13:39:36Z", "machine": "aarch64", "ort": "1.24.0", "numpy": "1.26.4", "model_sha256": "719559b0512db928e18eb3c130a122a55dc10166fb6bd2c8c90f2b6f51064994", "start": { "VmHWM_MiB": 46.30859375, "VmRSS_MiB": 44.4296875, "Pss_MiB": 34.58203125, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 18367.31640625, "memory.current_MiB": 1022.66796875 }, "rows": [ { "seconds": 15, "p50_ms": 81.36147679761052, "p95_ms": 89.36421200633049, "samples_ms": [ 89.36421200633049, 81.22202614322305, 81.36147679761052 ], "texts": [ "帮我导航到内湖科学园区，然后选最快的路线，不要走高架桥，接着把冷气打开，温度调到24度，风量开大。帮我导航到内湖科学园区，然后选最快的路线。不要走。", "帮我导航到内湖科学园区，然后选最快的路线，不要走高架桥，接着把冷气打开，温度调到24度，风量开大。帮我导航到内湖科学园区，然后选最快的路线。不要走。", "帮我导航到内湖科学园区，然后选最快的路线，不要走高架桥，接着把冷气打开，温度调到24度，风量开大。帮我导航到内湖科学园区，然后选最快的路线。不要走。" ], "memory": { "VmHWM_MiB": 8601.02734375, "VmRSS_MiB": 7013.0390625, "Pss_MiB": 5517.12890625, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 13335.734375, "memory.current_MiB": 6565.74609375 }, "audio_sha256": "4950a411664da365f4b8cfe13838f881430f01f2fed8b2c3335719f6b791ff9f" } ], "load_s": 15.166109941899776, "ready": { "VmHWM_MiB": 8601.02734375, "VmRSS_MiB": 6952.390625, "Pss_MiB": 5512.09765625, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 13389.1796875, "memory.current_MiB": 6510.99609375 }, "after_warmup": { "VmHWM_MiB": 8601.02734375, "VmRSS_MiB": 6981.04296875, "Pss_MiB": 5517.126953125, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 13367.48046875, "memory.current_MiB": 6533.24609375 }, "end": { "VmHWM_MiB": 8601.02734375, "VmRSS_MiB": 7013.0390625, "Pss_MiB": 5517.12890625, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 13335.734375, "memory.current_MiB": 6565.74609375 } }, "recognizer-trt-6s": { "arguments": { "self_check": false, "mode": "recognizer", "provider": "trt", "model": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/model.onnx", "tokens": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/tokens.txt", "cache": "/bench/cache", "audio": "/bench/audio", "length": 6, "frames": "50,100,250", "warmup": 10, "repeats": 3, "threads": 2, "profile": false, "dump_context": null, "out": "/bench/result.json" }, "utc": "2026-09-05T13:38:41Z", "machine": "aarch64", "ort": "1.24.0", "numpy": "1.26.4", "model_sha256": "719559b0512db928e18eb3c130a122a55dc10166fb6bd2c8c90f2b6f51064994", "start": { "VmHWM_MiB": 44.21484375, "VmRSS_MiB": 42.33984375, "Pss_MiB": 32.515625, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 18206.48046875, "memory.current_MiB": 1038.95703125 }, "rows": [ { "seconds": 6, "p50_ms": 39.396277628839016, "p95_ms": 54.19284477829933, "samples_ms": [ 54.19284477829933, 39.396277628839016, 39.381461683660746 ], "texts": [ "帮我导航到内湖科学园区，然后选最快的路线，不要走高架桥。", "帮我导航到内湖科学园区，然后选最快的路线，不要走高架桥。", "帮我导航到内湖科学园区，然后选最快的路线，不要走高架桥。" ], "memory": { "VmHWM_MiB": 8598.453125, "VmRSS_MiB": 6994.83203125, "Pss_MiB": 5514.84375, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 13136.14453125, "memory.current_MiB": 6802.125 }, "audio_sha256": "9566be6ccf6c7307830514e0fe5be86250c493efaef2800f27bfccc1ca1850b0" } ], "load_s": 15.272963133640587, "ready": { "VmHWM_MiB": 8598.453125, "VmRSS_MiB": 6949.703125, "Pss_MiB": 5509.443359375, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 13173.984375, "memory.current_MiB": 6763.1328125 }, "after_warmup": { "VmHWM_MiB": 8598.453125, "VmRSS_MiB": 6978.83203125, "Pss_MiB": 5514.84375, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 13152.49609375, "memory.current_MiB": 6786.125 }, "end": { "VmHWM_MiB": 8598.453125, "VmRSS_MiB": 6994.83203125, "Pss_MiB": 5514.84375, "MemTotal_MiB": 62840.6796875, "MemAvailable_MiB": 13136.14453125, "memory.current_MiB": 6802.125 } }, "services": { "service-cuda-1": { "environment": { "target": "ws://127.0.0.1:19342/stt/sensevoice", "uri": "ws://127.0.0.1:19342/stt/sensevoice", "warmup_runs": 3, "single_connection_series": true }, "generated": "2026-09-05T14:07:50", "results": [ { "feed": "bulk", "target_s": 3.0, "duration_s": 3.0, "status": "ok", "server_latency_ms": 89.43, "server_rtf": 0.019, "server_ttft_ms": 170.51, "client_final_latency_ms": 146.28, "interim_count": 0, "max_interim_gap_s": null, "feed_secs": 0.03, "text": "幫我導航到內湖科學園區，然後。", "error": null }, { "feed": "bulk", "target_s": 6.0, "duration_s": 6.0, "status": "ok", "server_latency_ms": 113.12, "server_rtf": 0.013, "server_ttft_ms": 223.12, "client_final_latency_ms": 194.65, "interim_count": 0, "max_interim_gap_s": null, "feed_secs": 0.03, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。", "error": null }, { "feed": "bulk", "target_s": 15.0, "duration_s": 15.0, "status": "ok", "server_latency_ms": 242.13, "server_rtf": 0.01, "server_ttft_ms": 85.16, "client_final_latency_ms": 302.42, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.19, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開大。幫我導航到內湖科學園區，然後選最快的路線。不要走。", "error": null }, { "feed": "realtime", "target_s": 3.0, "duration_s": 3.0, "status": "ok", "server_latency_ms": 80.41, "server_rtf": 0.144, "server_ttft_ms": 70.44, "client_final_latency_ms": 89.21, "interim_count": 7, "max_interim_gap_s": 0.71, "feed_secs": 3.17, "text": "幫我導航到內湖科學園區，然後。", "error": null }, { "feed": "realtime", "target_s": 6.0, "duration_s": 6.0, "status": "ok", "server_latency_ms": 80.34, "server_rtf": 0.136, "server_ttft_ms": 70.39, "client_final_latency_ms": 94.65, "interim_count": 12, "max_interim_gap_s": 1.022, "feed_secs": 6.32, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。", "error": null }, { "feed": "realtime", "target_s": 15.0, "duration_s": 15.0, "status": "ok", "server_latency_ms": 191.6, "server_rtf": 0.144, "server_ttft_ms": 69.78, "client_final_latency_ms": 198.13, "interim_count": 27, "max_interim_gap_s": 1.164, "feed_secs": 15.85, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開大。幫我導航到內湖科學園區，然後選最快的路線。不要走。", "error": null } ] }, "service-trt-1": { "environment": { "target": "ws://127.0.0.1:19342/stt/sensevoice", "uri": "ws://127.0.0.1:19342/stt/sensevoice", "warmup_runs": 3, "single_connection_series": true }, "generated": "2026-09-05T14:12:41", "results": [ { "feed": "bulk", "target_s": 3.0, "duration_s": 3.0, "status": "ok", "server_latency_ms": 56.73, "server_rtf": 0.011, "server_ttft_ms": 119.33, "client_final_latency_ms": 93.73, "interim_count": 0, "max_interim_gap_s": null, "feed_secs": 0.03, "text": "幫我導航到內湖科學園區，然後。", "error": null }, { "feed": "bulk", "target_s": 6.0, "duration_s": 6.0, "status": "ok", "server_latency_ms": 63.69, "server_rtf": 0.012, "server_ttft_ms": 62.44, "client_final_latency_ms": 155.59, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.05, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。", "error": null }, { "feed": "bulk", "target_s": 15.0, "duration_s": 15.0, "status": "ok", "server_latency_ms": 227.01, "server_rtf": 0.007, "server_ttft_ms": 60.5, "client_final_latency_ms": 369.26, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.09, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開大。幫我導航到內湖科學園區，然後選最快的路線。不要走。", "error": null }, { "feed": "realtime", "target_s": 3.0, "duration_s": 3.0, "status": "ok", "server_latency_ms": 46.68, "server_rtf": 0.068, "server_ttft_ms": 37.86, "client_final_latency_ms": 195.76, "interim_count": 6, "max_interim_gap_s": 0.515, "feed_secs": 3.17, "text": "幫我導航到內湖科學園區，然後。", "error": null }, { "feed": "realtime", "target_s": 6.0, "duration_s": 6.0, "status": "ok", "server_latency_ms": 45.47, "server_rtf": 0.066, "server_ttft_ms": 38.16, "client_final_latency_ms": 82.21, "interim_count": 12, "max_interim_gap_s": 1.03, "feed_secs": 6.34, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。", "error": null }, { "feed": "realtime", "target_s": 15.0, "duration_s": 15.0, "status": "ok", "server_latency_ms": 172.99, "server_rtf": 0.077, "server_ttft_ms": 37.62, "client_final_latency_ms": 178.06, "interim_count": 27, "max_interim_gap_s": 1.016, "feed_secs": 15.83, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開大。幫我導航到內湖科學園區，然後選最快的路線。不要走。", "error": null } ] }, "service-trt-3": { "environment": { "target": "ws://127.0.0.1:19342/stt/sensevoice", "uri": "ws://127.0.0.1:19342/stt/sensevoice", "warmup_runs": 3, "single_connection_series": true }, "generated": "2026-09-05T14:15:19", "results": [ { "feed": "bulk", "target_s": 3.0, "duration_s": 3.0, "status": "ok", "server_latency_ms": 55.17, "server_rtf": 0.01, "server_ttft_ms": 125.76, "client_final_latency_ms": 114.23, "interim_count": 0, "max_interim_gap_s": null, "feed_secs": 0.01, "text": "幫我導航到內湖科學園區，然後。", "error": null }, { "feed": "bulk", "target_s": 6.0, "duration_s": 6.0, "status": "ok", "server_latency_ms": 59.14, "server_rtf": 0.011, "server_ttft_ms": 86.24, "client_final_latency_ms": 152.22, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.03, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。", "error": null }, { "feed": "bulk", "target_s": 15.0, "duration_s": 15.0, "status": "ok", "server_latency_ms": 210.9, "server_rtf": 0.006, "server_ttft_ms": 49.61, "client_final_latency_ms": 305.02, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.14, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開大。幫我導航到內湖科學園區，然後選最快的路線。不要走。", "error": null }, { "feed": "realtime", "target_s": 3.0, "duration_s": 3.0, "status": "ok", "server_latency_ms": 29.52, "server_rtf": 0.054, "server_ttft_ms": 37.66, "client_final_latency_ms": 54.85, "interim_count": 6, "max_interim_gap_s": 1.061, "feed_secs": 3.14, "text": "幫我導航到內湖科學園區，然後。", "error": null }, { "feed": "realtime", "target_s": 6.0, "duration_s": 6.0, "status": "ok", "server_latency_ms": 45.55, "server_rtf": 0.068, "server_ttft_ms": 37.84, "client_final_latency_ms": 79.45, "interim_count": 11, "max_interim_gap_s": 1.018, "feed_secs": 6.34, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。", "error": null }, { "feed": "realtime", "target_s": 15.0, "duration_s": 15.0, "status": "ok", "server_latency_ms": 172.45, "server_rtf": 0.075, "server_ttft_ms": 37.87, "client_final_latency_ms": 178.2, "interim_count": 26, "max_interim_gap_s": 1.19, "feed_secs": 15.83, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開大。幫我導航到內湖科學園區，然後選最快的路線。不要走。", "error": null } ] }, "service-cuda-2": { "environment": { "target": "ws://127.0.0.1:19342/stt/sensevoice", "uri": "ws://127.0.0.1:19342/stt/sensevoice", "warmup_runs": 3, "single_connection_series": true }, "generated": "2026-09-05T14:09:20", "results": [ { "feed": "bulk", "target_s": 3.0, "duration_s": 3.0, "status": "ok", "server_latency_ms": 90.42, "server_rtf": 0.019, "server_ttft_ms": 161.48, "client_final_latency_ms": 115.5, "interim_count": 0, "max_interim_gap_s": null, "feed_secs": 0.05, "text": "幫我導航到內湖科學園區，然後。", "error": null }, { "feed": "bulk", "target_s": 6.0, "duration_s": 6.0, "status": "ok", "server_latency_ms": 99.14, "server_rtf": 0.021, "server_ttft_ms": 94.57, "client_final_latency_ms": 160.19, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.05, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。", "error": null }, { "feed": "bulk", "target_s": 15.0, "duration_s": 15.0, "status": "ok", "server_latency_ms": 228.9, "server_rtf": 0.01, "server_ttft_ms": 92.31, "client_final_latency_ms": 363.05, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.11, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開大。幫我導航到內湖科學園區，然後選最快的路線。不要走。", "error": null }, { "feed": "realtime", "target_s": 3.0, "duration_s": 3.0, "status": "ok", "server_latency_ms": 83.05, "server_rtf": 0.126, "server_ttft_ms": 66.47, "client_final_latency_ms": 89.04, "interim_count": 7, "max_interim_gap_s": 0.521, "feed_secs": 3.16, "text": "幫我導航到內湖科學園區，然後。", "error": null }, { "feed": "realtime", "target_s": 6.0, "duration_s": 6.0, "status": "ok", "server_latency_ms": 86.71, "server_rtf": 0.128, "server_ttft_ms": 67.35, "client_final_latency_ms": 102.76, "interim_count": 11, "max_interim_gap_s": 1.002, "feed_secs": 6.31, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。", "error": null }, { "feed": "realtime", "target_s": 15.0, "duration_s": 15.0, "status": "ok", "server_latency_ms": 191.45, "server_rtf": 0.13, "server_ttft_ms": 66.44, "client_final_latency_ms": 196.54, "interim_count": 28, "max_interim_gap_s": 1.024, "feed_secs": 15.78, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開大。幫我導航到內湖科學園區，然後選最快的路線。不要走。", "error": null } ] }, "service-cuda-3": { "environment": { "target": "ws://127.0.0.1:19342/stt/sensevoice", "uri": "ws://127.0.0.1:19342/stt/sensevoice", "warmup_runs": 3, "single_connection_series": true }, "generated": "2026-09-05T14:10:35", "results": [ { "feed": "bulk", "target_s": 3.0, "duration_s": 3.0, "status": "ok", "server_latency_ms": 92.96, "server_rtf": 0.023, "server_ttft_ms": 131.28, "client_final_latency_ms": 706.32, "interim_count": 0, "max_interim_gap_s": null, "feed_secs": 0.01, "text": "幫我導航到內湖科學園區，然後。", "error": null }, { "feed": "bulk", "target_s": 6.0, "duration_s": 6.0, "status": "ok", "server_latency_ms": 95.75, "server_rtf": 0.022, "server_ttft_ms": 94.46, "client_final_latency_ms": 166.04, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.05, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。", "error": null }, { "feed": "bulk", "target_s": 15.0, "duration_s": 15.0, "status": "ok", "server_latency_ms": 222.27, "server_rtf": 0.01, "server_ttft_ms": 106.68, "client_final_latency_ms": 361.4, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.11, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開大。幫我導航到內湖科學園區，然後選最快的路線。不要走。", "error": null }, { "feed": "realtime", "target_s": 3.0, "duration_s": 3.0, "status": "ok", "server_latency_ms": 81.78, "server_rtf": 0.131, "server_ttft_ms": 69.27, "client_final_latency_ms": 86.32, "interim_count": 7, "max_interim_gap_s": 0.519, "feed_secs": 3.17, "text": "幫我導航到內湖科學園區，然後。", "error": null }, { "feed": "realtime", "target_s": 6.0, "duration_s": 6.0, "status": "ok", "server_latency_ms": 89.49, "server_rtf": 0.13, "server_ttft_ms": 76.86, "client_final_latency_ms": 104.94, "interim_count": 11, "max_interim_gap_s": 1.02, "feed_secs": 6.35, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。", "error": null }, { "feed": "realtime", "target_s": 15.0, "duration_s": 15.0, "status": "ok", "server_latency_ms": 189.38, "server_rtf": 0.128, "server_ttft_ms": 69.24, "client_final_latency_ms": 194.51, "interim_count": 24, "max_interim_gap_s": 1.321, "feed_secs": 15.8, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開大。幫我導航到內湖科學園區，然後選最快的路線。不要走。", "error": null } ] }, "service-trt-2": { "environment": { "target": "ws://127.0.0.1:19342/stt/sensevoice", "uri": "ws://127.0.0.1:19342/stt/sensevoice", "warmup_runs": 3, "single_connection_series": true }, "generated": "2026-09-05T14:13:56", "results": [ { "feed": "bulk", "target_s": 3.0, "duration_s": 3.0, "status": "ok", "server_latency_ms": 55.89, "server_rtf": 0.01, "server_ttft_ms": 126.8, "client_final_latency_ms": 111.89, "interim_count": 0, "max_interim_gap_s": null, "feed_secs": 0.02, "text": "幫我導航到內湖科學園區，然後。", "error": null }, { "feed": "bulk", "target_s": 6.0, "duration_s": 6.0, "status": "ok", "server_latency_ms": 59.5, "server_rtf": 0.011, "server_ttft_ms": 73.38, "client_final_latency_ms": 154.94, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.03, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。", "error": null }, { "feed": "bulk", "target_s": 15.0, "duration_s": 15.0, "status": "ok", "server_latency_ms": 211.27, "server_rtf": 0.007, "server_ttft_ms": 101.81, "client_final_latency_ms": 351.11, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.1, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開大。幫我導航到內湖科學園區，然後選最快的路線。不要走。", "error": null }, { "feed": "realtime", "target_s": 3.0, "duration_s": 3.0, "status": "ok", "server_latency_ms": 44.54, "server_rtf": 0.06, "server_ttft_ms": 37.8, "client_final_latency_ms": 58.49, "interim_count": 5, "max_interim_gap_s": 0.995, "feed_secs": 3.16, "text": "幫我導航到內湖科學園區，然後。", "error": null }, { "feed": "realtime", "target_s": 6.0, "duration_s": 6.0, "status": "ok", "server_latency_ms": 45.32, "server_rtf": 0.062, "server_ttft_ms": 37.77, "client_final_latency_ms": 84.25, "interim_count": 11, "max_interim_gap_s": 1.012, "feed_secs": 6.33, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。", "error": null }, { "feed": "realtime", "target_s": 15.0, "duration_s": 15.0, "status": "ok", "server_latency_ms": 214.99, "server_rtf": 0.073, "server_ttft_ms": 37.9, "client_final_latency_ms": 232.96, "interim_count": 27, "max_interim_gap_s": 1.061, "feed_secs": 15.83, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開大。幫我導航到內湖科學園區，然後選最快的路線。不要走。", "error": null } ] } }, "image.txt": "registry.devops.foxconn.com/srdc/agent/stt/kitt-stt:v5.8.0-recheck-20260905-orin@sha256:d63d99616d88c72e4398e987a11191c212237ad68ec137810dfdb534fad6809c", "deployed-image.txt": "registry.devops.foxconn.com/srdc/agent/stt/kitt-stt:v5.8.0-recheck-ncpu-20260905-orin@sha256:742803510d4299f38c580136db078aee011de2ca0a6161c4f8e2b57c41762cca", "environment-final-deployment": { "platform": "Linux-5.15.148-tegra-aarch64-with-glibc2.35", "python": "3.10.12", "packages": { "onnxruntime-gpu": "1.24.0", "numpy": "1.26.4", "sherpa-onnx": "1.13.0+cuda", "torch": "2.8.0", "kitt-core": "0.3.7", "pipecat-ai": "0.0.108" }, "sha256": { "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/model.onnx": "719559b0512db928e18eb3c130a122a55dc10166fb6bd2c8c90f2b6f51064994", "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/tokens.txt": "f449eb28dc567533d7fa59be34e2abca8784f771850c78a47fb731a31429a1dc", "/app/src/asr/batch_asr_manager.py": "fb9a00f6cdbc5bd972d3bb82c85b44a70277a593d1694460071bac124ef25815", "/app/src/kitt_stt_service.py": "5283feaee7a213feb529afe1d98308c7fec447ccd18aa9f737a7632ea08b83bc" } } }, "4090-concurrent": { "recognizer-trt-3s": { "arguments": { "self_check": false, "mode": "recognizer", "provider": "trt", "model": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/model.onnx", "tokens": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/tokens.txt", "cache": "/bench/cache", "audio": "/bench/audio", "length": 3, "frames": "50,100,250", "warmup": 10, "repeats": 3, "threads": 2, "profile": false, "dump_context": null, "out": "/bench/4090/recognizer-trt-3s.json" }, "utc": "2026-09-05T13:07:33Z", "machine": "x86_64", "ort": "1.23.2", "numpy": "2.2.6", "model_sha256": "719559b0512db928e18eb3c130a122a55dc10166fb6bd2c8c90f2b6f51064994", "start": { "VmHWM_MiB": 47.41796875, "VmRSS_MiB": 45.9609375, "Pss_MiB": 45.95703125, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93826.515625, "memory.current_MiB": 28.3671875, "memory.peak_MiB": 30.52734375, "gpu_total_memory_util_clock_power": "9569, 0, 210, 25.13" }, "rows": [ { "seconds": 3, "p50_ms": 9.952828986570239, "p95_ms": 10.328458156436682, "samples_ms": [ 10.328458156436682, 9.833354968577623, 9.952828986570239 ], "texts": [ "帮我导航到内湖科学园区，然后。", "帮我导航到内湖科学园区，然后。", "帮我导航到内湖科学园区，然后。" ], "memory": { "VmHWM_MiB": 9419.62890625, "VmRSS_MiB": 7546.06640625, "Pss_MiB": 7541.5810546875, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 88025.75, "memory.current_MiB": 5450.9609375, "memory.peak_MiB": 7340.0546875, "gpu_total_memory_util_clock_power": "11096, 1, 2535, 67.71" }, "audio_sha256": "11034b66e2ae87bdec98dfe7401642387b85262b2e9c222b9e201030dc397a1c" } ], "load_s": 5.880446331109852, "ready": { "VmHWM_MiB": 9419.62890625, "VmRSS_MiB": 7536.2421875, "Pss_MiB": 7531.7568359375, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 88048.47265625, "memory.current_MiB": 5448.91015625, "memory.peak_MiB": 7340.0546875, "gpu_total_memory_util_clock_power": "11076, 1, 2535, 65.92" }, "after_warmup": { "VmHWM_MiB": 9419.62890625, "VmRSS_MiB": 7546.06640625, "Pss_MiB": 7541.5810546875, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 88037.37890625, "memory.current_MiB": 5450.93359375, "memory.peak_MiB": 7340.0546875, "gpu_total_memory_util_clock_power": "11096, 1, 2535, 66.63" }, "end": { "VmHWM_MiB": 9419.62890625, "VmRSS_MiB": 7546.06640625, "Pss_MiB": 7541.5810546875, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 88193.92578125, "memory.current_MiB": 5451.0, "memory.peak_MiB": 7340.0546875, "gpu_total_memory_util_clock_power": "11096, 1, 2535, 67.71" } }, "recognizer-cuda-3s": { "arguments": { "self_check": false, "mode": "recognizer", "provider": "cuda", "model": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/model.onnx", "tokens": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/tokens.txt", "cache": "/bench/cache", "audio": "/bench/audio", "length": 3, "frames": "50,100,250", "warmup": 10, "repeats": 3, "threads": 2, "profile": false, "dump_context": null, "out": "/bench/4090/recognizer-cuda-3s.json" }, "utc": "2026-09-05T12:43:02Z", "machine": "x86_64", "ort": "1.23.2", "numpy": "2.2.6", "model_sha256": "719559b0512db928e18eb3c130a122a55dc10166fb6bd2c8c90f2b6f51064994", "start": { "VmHWM_MiB": 46.0703125, "VmRSS_MiB": 45.765625, "Pss_MiB": 45.76171875, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 94615.38671875, "memory.current_MiB": 48.1328125, "memory.peak_MiB": 49.9921875, "gpu_total_memory_util_clock_power": "9569, 0, 210, 25.44" }, "rows": [ { "seconds": 3, "p50_ms": 12.696039164438844, "p95_ms": 12.77677109465003, "samples_ms": [ 12.77677109465003, 12.696039164438844, 12.600649846717715 ], "texts": [ "帮我导航到内湖科学园区，然后。", "帮我导航到内湖科学园区，然后。", "帮我导航到内湖科学园区，然后。" ], "memory": { "VmHWM_MiB": 1353.4921875, "VmRSS_MiB": 1343.7734375, "Pss_MiB": 1339.2314453125, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93844.37890625, "memory.current_MiB": 1300.125, "memory.peak_MiB": 1315.5859375, "gpu_total_memory_util_clock_power": "11008, 0, 2535, 65.54" }, "audio_sha256": "11034b66e2ae87bdec98dfe7401642387b85262b2e9c222b9e201030dc397a1c" } ], "load_s": 2.2766437819227576, "ready": { "VmHWM_MiB": 1353.4921875, "VmRSS_MiB": 1291.15234375, "Pss_MiB": 1286.6572265625, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93341.68359375, "memory.current_MiB": 1233.875, "memory.peak_MiB": 1299.078125, "gpu_total_memory_util_clock_power": "10998, 0, 2535, 64.23" }, "after_warmup": { "VmHWM_MiB": 1353.4921875, "VmRSS_MiB": 1343.7734375, "Pss_MiB": 1339.2314453125, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93667.83203125, "memory.current_MiB": 1300.09765625, "memory.peak_MiB": 1300.5078125, "gpu_total_memory_util_clock_power": "11008, 0, 2535, 63.71" }, "end": { "VmHWM_MiB": 1353.4921875, "VmRSS_MiB": 1343.7734375, "Pss_MiB": 1339.2314453125, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93719.484375, "memory.current_MiB": 1300.17578125, "memory.peak_MiB": 1315.59375, "gpu_total_memory_util_clock_power": "11008, 0, 2535, 65.54" } }, "environment": { "platform": "Linux-6.8.0-136-generic-x86_64-with-glibc2.36", "python": "3.10.19", "packages": { "onnxruntime-gpu": "1.23.2", "numpy": "2.2.6", "sherpa-onnx": "1.12.36", "torch": "2.9.1", "kitt-core": "0.3.7", "pipecat-ai": "0.0.108", "tensorrt-cu12": "10.13.3.9" }, "sha256": { "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/model.onnx": "719559b0512db928e18eb3c130a122a55dc10166fb6bd2c8c90f2b6f51064994", "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/tokens.txt": "f449eb28dc567533d7fa59be34e2abca8784f771850c78a47fb731a31429a1dc", "/app/src/asr/batch_asr_manager.py": "fb9a00f6cdbc5bd972d3bb82c85b44a70277a593d1694460071bac124ef25815", "/app/src/kitt_stt_service.py": "5283feaee7a213feb529afe1d98308c7fec447ccd18aa9f737a7632ea08b83bc" } }, "service-cuda-end-memory": { "utc": 1788613321.549029, "VmHWM_MiB": 2705.203125, "VmRSS_MiB": 2665.21484375, "Pss_MiB": 2658.03515625, "memory.current_MiB": 2455.81640625, "memory.peak_MiB": 2981.484375 }, "graph-trt": { "arguments": { "self_check": false, "mode": "graph", "provider": "trt", "model": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/model.onnx", "tokens": null, "cache": "/bench/cache", "audio": null, "length": 3, "frames": "50,100,250", "warmup": 10, "repeats": 50, "threads": 2, "profile": false, "dump_context": null, "out": "/bench/4090/graph-trt.json" }, "utc": "2026-09-05T13:08:40Z", "machine": "x86_64", "ort": "1.23.2", "numpy": "2.2.6", "model_sha256": "719559b0512db928e18eb3c130a122a55dc10166fb6bd2c8c90f2b6f51064994", "start": { "VmHWM_MiB": 47.16796875, "VmRSS_MiB": 45.73828125, "Pss_MiB": 45.734375, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93795.44921875, "memory.current_MiB": 28.2421875, "memory.peak_MiB": 30.19921875, "gpu_total_memory_util_clock_power": "9569, 0, 900, 64.13" }, "rows": [ { "frames": 50, "p50_ms": 6.0869010630995035, "p95_ms": 6.377147976309061, "samples_ms": [ 6.232210900634527, 6.172513822093606, 6.105597130954266, 6.08363002538681, 6.063100881874561, 6.157600088045001, 6.0095819644629955, 5.991270998492837, 5.9595550410449505, 6.377147976309061, 6.393870105966926, 6.403503008186817, 6.263872142881155, 6.193594075739384, 6.099425023421645, 6.119034951552749, 6.215111119672656, 6.088780937716365, 6.035469006747007, 6.060915999114513, 6.032600067555904, 6.217160960659385, 6.091483868658543, 6.057668011635542, 6.045526824891567, 6.085521075874567, 6.268962984904647, 6.086979061365128, 6.154519971460104, 6.185833131894469, 6.1594590079039335, 6.146664964035153, 6.0868230648338795, 6.09168503433466, 6.072461837902665, 6.097041070461273, 6.186744198203087, 6.151210051029921, 5.88117097504437, 5.889215972274542, 5.871551111340523, 5.975241074338555, 6.021746899932623, 5.912852007895708, 5.825829925015569, 5.800541955977678, 5.789068061858416, 5.990200908854604, 5.8707790449261665, 5.82902692258358 ], "memory": { "VmHWM_MiB": 9399.046875, "VmRSS_MiB": 7525.75390625, "Pss_MiB": 7521.2724609375, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 87526.36328125, "memory.current_MiB": 5441.66015625, "memory.peak_MiB": 7330.41015625, "gpu_total_memory_util_clock_power": "11106, 2, 2535, 93.17" }, "finite": true, "input_sha256": "472ba979ad46ce462cc2cceeb0d218314362876e03b63f06a136a1c19b3dbab7" }, { "frames": 100, "p50_ms": 5.878260941244662, "p95_ms": 6.150828907266259, "samples_ms": [ 5.831962218508124, 5.909692961722612, 6.124439183622599, 5.881454097107053, 6.015687948092818, 5.849621957167983, 5.875067785382271, 6.291236961260438, 6.458970019593835, 6.150828907266259, 5.890442058444023, 5.872976966202259, 5.931149004027247, 6.140192039310932, 6.00607693195343, 5.881513003259897, 5.849750014021993, 6.01952918805182, 6.087054032832384, 5.930023966357112, 5.862530088052154, 5.910641048103571, 5.915131885558367, 5.927688907831907, 5.950170801952481, 5.8406449388712645, 5.825205007568002, 5.814623087644577, 5.856079049408436, 5.990796023979783, 5.81294996663928, 5.783551139757037, 5.7526088785380125, 5.743007175624371, 5.923336837440729, 5.803894018754363, 5.774869816377759, 5.762081826105714, 5.901560885831714, 5.945226177573204, 5.970196099951863, 5.800989922136068, 5.773531971499324, 5.7534039951860905, 5.744708003476262, 6.005807081237435, 5.8594960719347, 5.816797958686948, 5.769510054960847, 5.78154600225389 ], "memory": { "VmHWM_MiB": 9399.046875, "VmRSS_MiB": 7525.859375, "Pss_MiB": 7521.3779296875, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 87451.15625, "memory.current_MiB": 5441.71484375, "memory.peak_MiB": 7330.41015625, "gpu_total_memory_util_clock_power": "11122, 2, 2535, 141.75" }, "finite": true, "input_sha256": "dda0d977bf96004405c6cbeac7cd57cfb466001cc340939617503e06dff71881" }, { "frames": 250, "p50_ms": 9.90213151089847, "p95_ms": 10.815754998475313, "samples_ms": [ 9.775936836376786, 10.06799703463912, 9.803483029827476, 9.799935854971409, 10.099423117935658, 9.886711835861206, 9.824865963310003, 9.805625071749091, 9.794787038117647, 9.837169200181961, 9.774973150342703, 10.128733003512025, 9.803519118577242, 10.029213968664408, 10.326871182769537, 9.839707985520363, 9.797341888770461, 10.183613048866391, 9.813320124521852, 9.850846137851477, 9.829391026869416, 10.815754998475313, 10.519578121602535, 10.09326707571745, 9.944404941052198, 10.087731992825866, 9.964150143787265, 10.404102969914675, 9.867944056168199, 9.836965007707477, 10.628944030031562, 9.923711884766817, 10.319409891963005, 11.334575014188886, 9.92819294333458, 9.81951504945755, 10.84224903024733, 9.968555998057127, 9.873359929770231, 10.072617093101144, 9.88934189081192, 9.914921130985022, 10.002356953918934, 9.940766962245107, 9.80430911295116, 9.789014933630824, 9.871133835986257, 9.835986187681556, 9.762719040736556, 10.394158074632287 ], "memory": { "VmHWM_MiB": 9399.046875, "VmRSS_MiB": 7525.875, "Pss_MiB": 7521.3896484375, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 87337.6484375, "memory.current_MiB": 5441.73828125, "memory.peak_MiB": 7330.41015625, "gpu_total_memory_util_clock_power": "11154, 76, 2715, 208.14" }, "finite": true, "input_sha256": "726cc736d46a4b8ca3c2d2831abc8c8e3a59489168194774d072d5fc1aa502b5" } ], "load_s": 5.807886126916856, "ready": { "VmHWM_MiB": 9399.046875, "VmRSS_MiB": 7523.38671875, "Pss_MiB": 7518.9013671875, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 87670.36328125, "memory.current_MiB": 5441.5, "memory.peak_MiB": 7330.41015625, "gpu_total_memory_util_clock_power": "11086, 27, 2535, 66.67" }, "providers": [ "TensorrtExecutionProvider", "CUDAExecutionProvider", "CPUExecutionProvider" ], "end": { "VmHWM_MiB": 9399.046875, "VmRSS_MiB": 7525.875, "Pss_MiB": 7521.3896484375, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 87333.95703125, "memory.current_MiB": 5441.93359375, "memory.peak_MiB": 7330.41015625, "gpu_total_memory_util_clock_power": "11154, 76, 2715, 208.14" } }, "service-cuda-ready-memory": { "utc": 1788613078.142293, "VmHWM_MiB": 2548.1953125, "VmRSS_MiB": 2531.5703125, "Pss_MiB": 2524.392578125, "memory.current_MiB": 2251.34765625, "memory.peak_MiB": 2278.34375 }, "graph-cuda": { "arguments": { "self_check": false, "mode": "graph", "provider": "cuda", "model": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/model.onnx", "tokens": null, "cache": "/bench/cache", "audio": null, "length": 3, "frames": "50,100,250", "warmup": 10, "repeats": 50, "threads": 2, "profile": false, "dump_context": null, "out": "/bench/4090/graph-cuda.json" }, "utc": "2026-09-05T12:44:19Z", "machine": "x86_64", "ort": "1.23.2", "numpy": "2.2.6", "model_sha256": "719559b0512db928e18eb3c130a122a55dc10166fb6bd2c8c90f2b6f51064994", "start": { "VmHWM_MiB": 46.72265625, "VmRSS_MiB": 45.7578125, "Pss_MiB": 45.75390625, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 94363.1328125, "memory.current_MiB": 29.7578125, "memory.peak_MiB": 31.71484375, "gpu_total_memory_util_clock_power": "9569, 0, 210, 25.38" }, "rows": [ { "frames": 50, "p50_ms": 8.172842091880739, "p95_ms": 8.643068838864565, "samples_ms": [ 8.527150144800544, 8.481200085952878, 8.560962975025177, 8.489476051181555, 8.578567998483777, 8.645408088341355, 8.650887990370393, 8.561576949432492, 8.594484999775887, 8.643068838864565, 8.574220119044185, 8.489442989230156, 8.357185870409012, 8.069282863289118, 8.012234000489116, 7.905936799943447, 7.927268976345658, 8.041779045015574, 7.971392013132572, 7.934102090075612, 8.020943962037563, 8.093304000794888, 7.976610912010074, 7.879026001319289, 7.878228090703487, 7.9693240113556385, 7.935520028695464, 7.906096987426281, 7.933028042316437, 8.281433023512363, 8.428520988672972, 8.395337965339422, 8.227469166740775, 8.121690945699811, 8.06975201703608, 8.217249996960163, 8.14719800837338, 8.10133689083159, 8.081771899014711, 8.311423938721418, 8.275034837424755, 8.236921858042479, 8.122666971758008, 8.35025985725224, 8.20566387847066, 8.087004069238901, 8.085719076916575, 8.27315985225141, 8.196765091270208, 8.148919092491269 ], "memory": { "VmHWM_MiB": 1337.25, "VmRSS_MiB": 1249.69921875, "Pss_MiB": 1245.1650390625, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93248.890625, "memory.current_MiB": 1037.40625, "memory.peak_MiB": 1179.08203125, "gpu_total_memory_util_clock_power": "11008, 96, 2715, 94.26" }, "finite": true, "input_sha256": "472ba979ad46ce462cc2cceeb0d218314362876e03b63f06a136a1c19b3dbab7" }, { "frames": 100, "p50_ms": 8.167131571099162, "p95_ms": 8.49536620080471, "samples_ms": [ 8.470481028780341, 8.220166899263859, 8.219011127948761, 8.235972840338945, 8.379911072552204, 8.26529785990715, 8.44961917027831, 8.465796010568738, 8.223028853535652, 8.120185928419232, 8.086116053164005, 8.316634921357036, 8.152869995683432, 8.126328000798821, 8.048854069784284, 8.513472974300385, 8.30240617506206, 8.312301011756063, 8.175408001989126, 8.514976827427745, 8.308883057907224, 8.128295186907053, 8.080547209829092, 8.139024022966623, 8.210645988583565, 8.125566877424717, 8.063697954639792, 8.068103808909655, 8.188550127670169, 8.154212962836027, 8.148467866703868, 8.068480994552374, 8.115720003843307, 8.132621878758073, 8.054840844124556, 8.08222801424563, 8.251148043200374, 8.175170049071312, 8.35245382040739, 8.083882043138146, 8.132907096296549, 8.082648040726781, 8.49536620080471, 8.15618084743619, 8.159093093127012, 8.082743035629392, 8.2451649941504, 8.276771986857057, 8.202970959246159, 8.15177010372281 ], "memory": { "VmHWM_MiB": 1337.25, "VmRSS_MiB": 1250.0234375, "Pss_MiB": 1245.4853515625, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93232.3359375, "memory.current_MiB": 1037.328125, "memory.peak_MiB": 1179.08203125, "gpu_total_memory_util_clock_power": "11008, 96, 2715, 137.59" }, "finite": true, "input_sha256": "dda0d977bf96004405c6cbeac7cd57cfb466001cc340939617503e06dff71881" }, { "frames": 250, "p50_ms": 10.545336059294641, "p95_ms": 11.397663038223982, "samples_ms": [ 10.346119059249759, 11.035120114684105, 10.39771898649633, 10.345060843974352, 10.879555949941278, 10.750374058261514, 10.603742208331823, 11.397663038223982, 10.646871058270335, 10.459722951054573, 11.316381860524416, 10.518630035221577, 10.453202994540334, 11.43207703717053, 10.626694187521935, 10.407175868749619, 13.248879928141832, 11.390739120543003, 10.627188952639699, 10.87002013809979, 10.520662181079388, 10.482548968866467, 11.100440053269267, 10.626795934513211, 10.512762004509568, 11.165613075718284, 11.015312047675252, 10.65000705420971, 11.087868129834533, 10.425246087834239, 10.349211050197482, 10.35677408799529, 10.508568957448006, 10.37365198135376, 10.518023045733571, 10.406506014987826, 10.490281041711569, 10.744059924036264, 10.472094174474478, 10.324110044166446, 10.62650210224092, 10.45611291192472, 10.355989914387465, 10.890493867918849, 10.476930998265743, 10.497616836801171, 11.065457947552204, 10.570009937509894, 10.438109980896115, 10.881124995648861 ], "memory": { "VmHWM_MiB": 1337.25, "VmRSS_MiB": 1250.03125, "Pss_MiB": 1245.4892578125, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93191.5234375, "memory.current_MiB": 1037.13671875, "memory.peak_MiB": 1179.08203125, "gpu_total_memory_util_clock_power": "11008, 95, 2715, 186.99" }, "finite": true, "input_sha256": "726cc736d46a4b8ca3c2d2831abc8c8e3a59489168194774d072d5fc1aa502b5" } ], "load_s": 9.63523994688876, "ready": { "VmHWM_MiB": 1337.25, "VmRSS_MiB": 1194.55859375, "Pss_MiB": 1190.0634765625, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93821.203125, "memory.current_MiB": 1027.3125, "memory.peak_MiB": 1179.08203125, "gpu_total_memory_util_clock_power": "10998, 0, 2535, 63.97" }, "providers": [ "CUDAExecutionProvider", "CPUExecutionProvider" ], "end": { "VmHWM_MiB": 1337.25, "VmRSS_MiB": 1250.03125, "Pss_MiB": 1245.4892578125, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93172.6875, "memory.current_MiB": 1037.4453125, "memory.peak_MiB": 1179.08203125, "gpu_total_memory_util_clock_power": "11008, 95, 2715, 184.19" } }, "service-trt-ready-memory": { "utc": 1788613403.6617377, "VmHWM_MiB": 11960.37890625, "VmRSS_MiB": 10985.31640625, "Pss_MiB": 10978.08203125, "memory.current_MiB": 10038.34375, "memory.peak_MiB": 10962.77734375 }, "service-trt-end-memory": { "utc": 1788613616.869852, "VmHWM_MiB": 11960.37890625, "VmRSS_MiB": 11104.43359375, "Pss_MiB": 11097.171875, "memory.current_MiB": 10157.36328125, "memory.peak_MiB": 10962.77734375 }, "recognizer-cuda-6s": { "arguments": { "self_check": false, "mode": "recognizer", "provider": "cuda", "model": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/model.onnx", "tokens": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/tokens.txt", "cache": "/bench/cache", "audio": "/bench/audio", "length": 6, "frames": "50,100,250", "warmup": 10, "repeats": 3, "threads": 2, "profile": false, "dump_context": null, "out": "/bench/4090/recognizer-cuda-6s.json" }, "utc": "2026-09-05T12:43:40Z", "machine": "x86_64", "ort": "1.23.2", "numpy": "2.2.6", "model_sha256": "719559b0512db928e18eb3c130a122a55dc10166fb6bd2c8c90f2b6f51064994", "start": { "VmHWM_MiB": 46.453125, "VmRSS_MiB": 46.12109375, "Pss_MiB": 46.1171875, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 94240.55859375, "memory.current_MiB": 30.77734375, "memory.peak_MiB": 32.72265625, "gpu_total_memory_util_clock_power": "9569, 0, 210, 24.95" }, "rows": [ { "seconds": 6, "p50_ms": 18.740060972049832, "p95_ms": 37.363274954259396, "samples_ms": [ 37.363274954259396, 16.705831978470087, 18.740060972049832 ], "texts": [ "帮我导航到内湖科学园区，然后选最快的路线，不要走高架桥。", "帮我导航到内湖科学园区，然后选最快的路线，不要走高架桥。", "帮我导航到内湖科学园区，然后选最快的路线，不要走高架桥。" ], "memory": { "VmHWM_MiB": 1350.796875, "VmRSS_MiB": 1318.4375, "Pss_MiB": 1313.8955078125, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93534.0390625, "memory.current_MiB": 1086.8984375, "memory.peak_MiB": 1175.015625, "gpu_total_memory_util_clock_power": "11008, 0, 2535, 65.04" }, "audio_sha256": "9566be6ccf6c7307830514e0fe5be86250c493efaef2800f27bfccc1ca1850b0" } ], "load_s": 3.0798465500120074, "ready": { "VmHWM_MiB": 1350.796875, "VmRSS_MiB": 1265.10546875, "Pss_MiB": 1260.6103515625, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93091.40625, "memory.current_MiB": 1080.43359375, "memory.peak_MiB": 1175.015625, "gpu_total_memory_util_clock_power": "10998, 0, 2535, 64.54" }, "after_warmup": { "VmHWM_MiB": 1350.796875, "VmRSS_MiB": 1318.359375, "Pss_MiB": 1313.8173828125, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 92721.37890625, "memory.current_MiB": 1086.91796875, "memory.peak_MiB": 1175.015625, "gpu_total_memory_util_clock_power": "11008, 0, 2535, 64.36" }, "end": { "VmHWM_MiB": 1350.796875, "VmRSS_MiB": 1318.4375, "Pss_MiB": 1313.8955078125, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93524.19921875, "memory.current_MiB": 1087.75, "memory.peak_MiB": 1175.015625, "gpu_total_memory_util_clock_power": "11008, 0, 2535, 65.79" } }, "recognizer-cuda-15s": { "arguments": { "self_check": false, "mode": "recognizer", "provider": "cuda", "model": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/model.onnx", "tokens": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/tokens.txt", "cache": "/bench/cache", "audio": "/bench/audio", "length": 15, "frames": "50,100,250", "warmup": 10, "repeats": 3, "threads": 2, "profile": false, "dump_context": null, "out": "/bench/4090/recognizer-cuda-15s.json" }, "utc": "2026-09-05T12:43:58Z", "machine": "x86_64", "ort": "1.23.2", "numpy": "2.2.6", "model_sha256": "719559b0512db928e18eb3c130a122a55dc10166fb6bd2c8c90f2b6f51064994", "start": { "VmHWM_MiB": 47.1328125, "VmRSS_MiB": 45.91796875, "Pss_MiB": 45.9140625, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 94590.37109375, "memory.current_MiB": 28.25, "memory.peak_MiB": 30.40234375, "gpu_total_memory_util_clock_power": "9569, 0, 210, 25.33" }, "rows": [ { "seconds": 15, "p50_ms": 29.02595279738307, "p95_ms": 35.68060998804867, "samples_ms": [ 35.68060998804867, 29.02595279738307, 28.76228094100952 ], "texts": [ "帮我导航到内湖科学园区，然后选最快的路线，不要走高架桥，接着把冷气打开，温度调到24度，风量开大。帮我导航到内湖科学园区，然后选最快的路线。不要走。", "帮我导航到内湖科学园区，然后选最快的路线，不要走高架桥，接着把冷气打开，温度调到24度，风量开大。帮我导航到内湖科学园区，然后选最快的路线。不要走。", "帮我导航到内湖科学园区，然后选最快的路线，不要走高架桥，接着把冷气打开，温度调到24度，风量开大。帮我导航到内湖科学园区，然后选最快的路线。不要走。" ], "memory": { "VmHWM_MiB": 1358.328125, "VmRSS_MiB": 1324.21875, "Pss_MiB": 1319.6767578125, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93389.7421875, "memory.current_MiB": 1090.578125, "memory.peak_MiB": 1179.39453125, "gpu_total_memory_util_clock_power": "11006, 0, 2535, 67.55" }, "audio_sha256": "4950a411664da365f4b8cfe13838f881430f01f2fed8b2c3335719f6b791ff9f" } ], "load_s": 1.240162475965917, "ready": { "VmHWM_MiB": 1358.328125, "VmRSS_MiB": 1270.671875, "Pss_MiB": 1266.1767578125, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93406.62109375, "memory.current_MiB": 1083.6875, "memory.peak_MiB": 1179.39453125, "gpu_total_memory_util_clock_power": "10996, 0, 2535, 64.88" }, "after_warmup": { "VmHWM_MiB": 1358.328125, "VmRSS_MiB": 1324.12890625, "Pss_MiB": 1319.5869140625, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93390.21875, "memory.current_MiB": 1090.21875, "memory.peak_MiB": 1179.39453125, "gpu_total_memory_util_clock_power": "11006, 0, 2535, 65.19" }, "end": { "VmHWM_MiB": 1358.328125, "VmRSS_MiB": 1324.21875, "Pss_MiB": 1319.6767578125, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93389.7421875, "memory.current_MiB": 1090.4609375, "memory.peak_MiB": 1179.39453125, "gpu_total_memory_util_clock_power": "11006, 0, 2535, 67.55" } }, "recognizer-trt-15s": { "arguments": { "self_check": false, "mode": "recognizer", "provider": "trt", "model": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/model.onnx", "tokens": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/tokens.txt", "cache": "/bench/cache", "audio": "/bench/audio", "length": 15, "frames": "50,100,250", "warmup": 10, "repeats": 3, "threads": 2, "profile": false, "dump_context": null, "out": "/bench/4090/recognizer-trt-15s.json" }, "utc": "2026-09-05T13:08:32Z", "machine": "x86_64", "ort": "1.23.2", "numpy": "2.2.6", "model_sha256": "719559b0512db928e18eb3c130a122a55dc10166fb6bd2c8c90f2b6f51064994", "start": { "VmHWM_MiB": 46.9453125, "VmRSS_MiB": 45.875, "Pss_MiB": 45.87109375, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 94762.30859375, "memory.current_MiB": 28.2734375, "memory.peak_MiB": 30.54296875, "gpu_total_memory_util_clock_power": "9569, 0, 210, 24.78" }, "rows": [ { "seconds": 15, "p50_ms": 28.46145606599748, "p95_ms": 28.62507291138172, "samples_ms": [ 28.46145606599748, 28.62507291138172, 28.0569139868021 ], "texts": [ "帮我导航到内湖科学园区，然后选最快的路线，不要走高架桥，接着把冷气打开，温度调到24度，风量开大。帮我导航到内湖科学园区，然后选最快的路线。不要走。", "帮我导航到内湖科学园区，然后选最快的路线，不要走高架桥，接着把冷气打开，温度调到24度，风量开大。帮我导航到内湖科学园区，然后选最快的路线。不要走。", "帮我导航到内湖科学园区，然后选最快的路线，不要走高架桥，接着把冷气打开，温度调到24度，风量开大。帮我导航到内湖科学园区，然后选最快的路线。不要走。" ], "memory": { "VmHWM_MiB": 9418.8984375, "VmRSS_MiB": 7546.0625, "Pss_MiB": 7541.5771484375, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 88685.19921875, "memory.current_MiB": 5450.85546875, "memory.peak_MiB": 7340.00390625, "gpu_total_memory_util_clock_power": "11128, 1, 2535, 70.03" }, "audio_sha256": "4950a411664da365f4b8cfe13838f881430f01f2fed8b2c3335719f6b791ff9f" } ], "load_s": 5.809297051047906, "ready": { "VmHWM_MiB": 9418.8984375, "VmRSS_MiB": 7536.15234375, "Pss_MiB": 7531.6669921875, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 88234.67578125, "memory.current_MiB": 5448.6328125, "memory.peak_MiB": 7340.00390625, "gpu_total_memory_util_clock_power": "11076, 1, 2535, 66.35" }, "after_warmup": { "VmHWM_MiB": 9418.8984375, "VmRSS_MiB": 7546.0625, "Pss_MiB": 7541.5771484375, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 88261.5, "memory.current_MiB": 5451.23046875, "memory.peak_MiB": 7340.00390625, "gpu_total_memory_util_clock_power": "11096, 1, 2535, 66.82" }, "end": { "VmHWM_MiB": 9418.8984375, "VmRSS_MiB": 7546.0625, "Pss_MiB": 7541.5771484375, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 88669.49609375, "memory.current_MiB": 5450.92578125, "memory.peak_MiB": 7340.00390625, "gpu_total_memory_util_clock_power": "11128, 1, 2535, 70.03" } }, "recognizer-trt-6s": { "arguments": { "self_check": false, "mode": "recognizer", "provider": "trt", "model": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/model.onnx", "tokens": "/app/models/sherpa-onnx-sense-voice-kitt-wake-lora-v1/tokens.txt", "cache": "/bench/cache", "audio": "/bench/audio", "length": 6, "frames": "50,100,250", "warmup": 10, "repeats": 3, "threads": 2, "profile": false, "dump_context": null, "out": "/bench/4090/recognizer-trt-6s.json" }, "utc": "2026-09-05T13:07:41Z", "machine": "x86_64", "ort": "1.23.2", "numpy": "2.2.6", "model_sha256": "719559b0512db928e18eb3c130a122a55dc10166fb6bd2c8c90f2b6f51064994", "start": { "VmHWM_MiB": 46.64453125, "VmRSS_MiB": 45.69921875, "Pss_MiB": 45.6953125, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 93011.8671875, "memory.current_MiB": 28.27734375, "memory.peak_MiB": 30.32421875, "gpu_total_memory_util_clock_power": "9569, 0, 2535, 65.26" }, "rows": [ { "seconds": 6, "p50_ms": 14.208432985469699, "p95_ms": 16.890024999156594, "samples_ms": [ 16.890024999156594, 14.030528021976352, 14.208432985469699 ], "texts": [ "帮我导航到内湖科学园区，然后选最快的路线，不要走高架桥。", "帮我导航到内湖科学园区，然后选最快的路线，不要走高架桥。", "帮我导航到内湖科学园区，然后选最快的路线，不要走高架桥。" ], "memory": { "VmHWM_MiB": 9418.51171875, "VmRSS_MiB": 7545.78125, "Pss_MiB": 7541.2958984375, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 87763.0546875, "memory.current_MiB": 5451.08984375, "memory.peak_MiB": 7339.99609375, "gpu_total_memory_util_clock_power": "11112, 1, 2535, 66.99" }, "audio_sha256": "9566be6ccf6c7307830514e0fe5be86250c493efaef2800f27bfccc1ca1850b0" } ], "load_s": 5.8533999510109425, "ready": { "VmHWM_MiB": 9418.51171875, "VmRSS_MiB": 7535.75390625, "Pss_MiB": 7531.2685546875, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 87766.51171875, "memory.current_MiB": 5448.91015625, "memory.peak_MiB": 7339.99609375, "gpu_total_memory_util_clock_power": "11076, 1, 2535, 66.01" }, "after_warmup": { "VmHWM_MiB": 9418.51171875, "VmRSS_MiB": 7545.78125, "Pss_MiB": 7541.2958984375, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 87770.63671875, "memory.current_MiB": 5451.01171875, "memory.peak_MiB": 7339.99609375, "gpu_total_memory_util_clock_power": "11096, 1, 2535, 65.96" }, "end": { "VmHWM_MiB": 9418.51171875, "VmRSS_MiB": 7545.78125, "Pss_MiB": 7541.2958984375, "MemTotal_MiB": 128413.15234375, "MemAvailable_MiB": 87969.4765625, "memory.current_MiB": 5451.40625, "memory.peak_MiB": 7339.99609375, "gpu_total_memory_util_clock_power": "11112, 1, 2535, 67.51" } }, "services": { "service-cuda-1": { "environment": { "target": "ws://127.0.0.1:19242/stt/sensevoice", "uri": "ws://127.0.0.1:19242/stt/sensevoice", "warmup_runs": 3, "single_connection_series": true }, "generated": "2026-09-05T12:59:29", "results": [ { "feed": "bulk", "target_s": 3.0, "duration_s": 3.0, "status": "ok", "server_latency_ms": 50.46, "server_rtf": 0.01, "server_ttft_ms": 71.01, "client_final_latency_ms": 2996.61, "interim_count": 0, "max_interim_gap_s": null, "feed_secs": 0.01, "text": "幫我導航到內湖科學園區，然後。", "error": null }, { "feed": "bulk", "target_s": 6.0, "duration_s": 6.0, "status": "ok", "server_latency_ms": 32.01, "server_rtf": 0.007, "server_ttft_ms": 29.24, "client_final_latency_ms": 38.55, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.06, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。", "error": null }, { "feed": "bulk", "target_s": 15.0, "duration_s": 15.0, "status": "ok", "server_latency_ms": 102.37, "server_rtf": 0.004, "server_ttft_ms": 29.53, "client_final_latency_ms": 113.96, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.17, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開。幫我導航到內湖科學園區，然後選最快的路線，不要走。", "error": null }, { "feed": "realtime", "target_s": 3.0, "duration_s": 3.0, "status": "ok", "server_latency_ms": 28.72, "server_rtf": 0.059, "server_ttft_ms": 48.3, "client_final_latency_ms": 30.48, "interim_count": 6, "max_interim_gap_s": 0.524, "feed_secs": 3.1, "text": "幫我導航到內湖科學園區，然後。", "error": null }, { "feed": "realtime", "target_s": 6.0, "duration_s": 6.0, "status": "ok", "server_latency_ms": 37.46, "server_rtf": 0.052, "server_ttft_ms": 19.87, "client_final_latency_ms": 40.75, "interim_count": 10, "max_interim_gap_s": 1.368, "feed_secs": 6.21, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。", "error": null }, { "feed": "realtime", "target_s": 15.0, "duration_s": 15.0, "status": "ok", "server_latency_ms": 64.04, "server_rtf": 0.064, "server_ttft_ms": 43.44, "client_final_latency_ms": 66.34, "interim_count": 26, "max_interim_gap_s": 3.09, "feed_secs": 15.53, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開。幫我導航到內湖科學園區，然後選最快的路線，不要走。", "error": null } ] }, "service-trt-1": { "environment": { "target": "ws://127.0.0.1:19242/stt/sensevoice", "uri": "ws://127.0.0.1:19242/stt/sensevoice", "warmup_runs": 3, "single_connection_series": true }, "generated": "2026-09-05T13:04:35", "results": [ { "feed": "bulk", "target_s": 3.0, "duration_s": 3.0, "status": "ok", "server_latency_ms": 12.56, "server_rtf": 0.005, "server_ttft_ms": 17.52, "client_final_latency_ms": 19.32, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.03, "text": "幫我導航到內湖科學園區，然後。", "error": null }, { "feed": "bulk", "target_s": 6.0, "duration_s": 6.0, "status": "ok", "server_latency_ms": 21.69, "server_rtf": 0.003, "server_ttft_ms": 15.08, "client_final_latency_ms": 534.48, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.06, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。", "error": null }, { "feed": "bulk", "target_s": 15.0, "duration_s": 15.0, "status": "ok", "server_latency_ms": 75.26, "server_rtf": 0.003, "server_ttft_ms": 28.74, "client_final_latency_ms": 87.12, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.2, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開大。幫我導航到內湖科學園區，然後選最快的路線。不要走。", "error": null }, { "feed": "realtime", "target_s": 3.0, "duration_s": 3.0, "status": "ok", "server_latency_ms": 16.4, "server_rtf": 0.003, "server_ttft_ms": 35.19, "client_final_latency_ms": 1465.75, "interim_count": 0, "max_interim_gap_s": null, "feed_secs": 3.08, "text": "幫我導航到內湖科學園區，然後。", "error": null }, { "feed": "realtime", "target_s": 6.0, "duration_s": 6.0, "status": "ok", "server_latency_ms": 20.94, "server_rtf": 0.023, "server_ttft_ms": 11.8, "client_final_latency_ms": 25.23, "interim_count": 12, "max_interim_gap_s": 1.006, "feed_secs": 6.24, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。", "error": null }, { "feed": "realtime", "target_s": 15.0, "duration_s": 15.0, "status": "ok", "server_latency_ms": 56.59, "server_rtf": 0.027, "server_ttft_ms": 10.33, "client_final_latency_ms": 58.94, "interim_count": 28, "max_interim_gap_s": 1.038, "feed_secs": 15.61, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開大。幫我導航到內湖科學園區，然後選最快的路線。不要走。", "error": null } ] }, "service-trt-3": { "environment": { "target": "ws://127.0.0.1:19242/stt/sensevoice", "uri": "ws://127.0.0.1:19242/stt/sensevoice", "warmup_runs": 3, "single_connection_series": true }, "generated": "2026-09-05T13:06:56", "results": [ { "feed": "bulk", "target_s": 3.0, "duration_s": 3.0, "status": "ok", "server_latency_ms": 12.37, "server_rtf": 0.004, "server_ttft_ms": 11.71, "client_final_latency_ms": 21.25, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.02, "text": "幫我導航到內湖科學園區，然後。", "error": null }, { "feed": "bulk", "target_s": 6.0, "duration_s": 6.0, "status": "ok", "server_latency_ms": 16.5, "server_rtf": 0.003, "server_ttft_ms": 10.74, "client_final_latency_ms": 29.46, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.04, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。", "error": null }, { "feed": "bulk", "target_s": 15.0, "duration_s": 15.0, "status": "ok", "server_latency_ms": 64.6, "server_rtf": 0.002, "server_ttft_ms": 11.83, "client_final_latency_ms": 70.53, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.12, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開。幫我導航到內湖科學園區，然後選最快的路線，不要走。", "error": null }, { "feed": "realtime", "target_s": 3.0, "duration_s": 3.0, "status": "ok", "server_latency_ms": 11.22, "server_rtf": 0.016, "server_ttft_ms": 11.96, "client_final_latency_ms": 14.87, "interim_count": 6, "max_interim_gap_s": 0.577, "feed_secs": 3.11, "text": "幫我導航到內湖科學園區，然後。", "error": null }, { "feed": "realtime", "target_s": 6.0, "duration_s": 6.0, "status": "ok", "server_latency_ms": 16.36, "server_rtf": 0.02, "server_ttft_ms": 10.42, "client_final_latency_ms": 27.75, "interim_count": 11, "max_interim_gap_s": 1.021, "feed_secs": 6.24, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。", "error": null }, { "feed": "realtime", "target_s": 15.0, "duration_s": 15.0, "status": "ok", "server_latency_ms": 51.02, "server_rtf": 0.028, "server_ttft_ms": 11.14, "client_final_latency_ms": 53.59, "interim_count": 29, "max_interim_gap_s": 1.031, "feed_secs": 15.57, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開。幫我導航到內湖科學園區，然後選最快的路線，不要走。", "error": null } ] }, "service-cuda-2": { "environment": { "target": "ws://127.0.0.1:19242/stt/sensevoice", "uri": "ws://127.0.0.1:19242/stt/sensevoice", "warmup_runs": 3, "single_connection_series": true }, "generated": "2026-09-05T13:00:50", "results": [ { "feed": "bulk", "target_s": 3.0, "duration_s": 3.0, "status": "ok", "server_latency_ms": 46.48, "server_rtf": 0.009, "server_ttft_ms": 91.3, "client_final_latency_ms": 69.59, "interim_count": 0, "max_interim_gap_s": null, "feed_secs": 0.02, "text": "幫我導航到內湖科學園區，然後。", "error": null }, { "feed": "bulk", "target_s": 6.0, "duration_s": 6.0, "status": "ok", "server_latency_ms": 107.2, "server_rtf": 0.019, "server_ttft_ms": 63.03, "client_final_latency_ms": 136.29, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.09, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。", "error": null }, { "feed": "bulk", "target_s": 15.0, "duration_s": 15.0, "status": "ok", "server_latency_ms": 106.64, "server_rtf": 0.006, "server_ttft_ms": 58.05, "client_final_latency_ms": 116.07, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.15, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開。幫我導航到內湖科學園區，然後選最快的路線，不要走。", "error": null }, { "feed": "realtime", "target_s": 3.0, "duration_s": 3.0, "status": "ok", "server_latency_ms": 22.98, "server_rtf": 0.057, "server_ttft_ms": 43.29, "client_final_latency_ms": 29.78, "interim_count": 6, "max_interim_gap_s": 0.533, "feed_secs": 3.13, "text": "幫我導航到內湖科學園區，然後。", "error": null }, { "feed": "realtime", "target_s": 6.0, "duration_s": 6.0, "status": "ok", "server_latency_ms": 34.26, "server_rtf": 0.088, "server_ttft_ms": 36.17, "client_final_latency_ms": 51.92, "interim_count": 11, "max_interim_gap_s": 1.021, "feed_secs": 6.26, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。", "error": null }, { "feed": "realtime", "target_s": 15.0, "duration_s": 15.0, "status": "ok", "server_latency_ms": 118.14, "server_rtf": 0.103, "server_ttft_ms": 45.38, "client_final_latency_ms": 121.73, "interim_count": 25, "max_interim_gap_s": 1.051, "feed_secs": 15.67, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開。幫我導航到內湖科學園區，然後選最快的路線，不要走。", "error": null } ] }, "service-cuda-3": { "environment": { "target": "ws://127.0.0.1:19242/stt/sensevoice", "uri": "ws://127.0.0.1:19242/stt/sensevoice", "warmup_runs": 3, "single_connection_series": true }, "generated": "2026-09-05T13:02:00", "results": [ { "feed": "bulk", "target_s": 3.0, "duration_s": 3.0, "status": "ok", "server_latency_ms": 60.08, "server_rtf": 0.025, "server_ttft_ms": 54.64, "client_final_latency_ms": 75.13, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.05, "text": "幫我導航到內湖科學園區，然後。", "error": null }, { "feed": "bulk", "target_s": 6.0, "duration_s": 6.0, "status": "ok", "server_latency_ms": 60.64, "server_rtf": 0.014, "server_ttft_ms": 50.22, "client_final_latency_ms": 72.72, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.09, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。", "error": null }, { "feed": "bulk", "target_s": 15.0, "duration_s": 15.0, "status": "ok", "server_latency_ms": 110.07, "server_rtf": 0.006, "server_ttft_ms": 59.88, "client_final_latency_ms": 121.3, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.19, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開。幫我導航到內湖科學園區，然後選最快的路線，不要走。", "error": null }, { "feed": "realtime", "target_s": 3.0, "duration_s": 3.0, "status": "ok", "server_latency_ms": 58.39, "server_rtf": 0.102, "server_ttft_ms": 52.6, "client_final_latency_ms": 62.11, "interim_count": 6, "max_interim_gap_s": 0.531, "feed_secs": 3.14, "text": "幫我導航到內湖科學園區，然後。", "error": null }, { "feed": "realtime", "target_s": 6.0, "duration_s": 6.0, "status": "ok", "server_latency_ms": 46.1, "server_rtf": 0.091, "server_ttft_ms": 50.0, "client_final_latency_ms": 53.97, "interim_count": 9, "max_interim_gap_s": 1.041, "feed_secs": 6.25, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。", "error": null }, { "feed": "realtime", "target_s": 15.0, "duration_s": 15.0, "status": "ok", "server_latency_ms": 83.43, "server_rtf": 0.083, "server_ttft_ms": 42.67, "client_final_latency_ms": 86.72, "interim_count": 27, "max_interim_gap_s": 1.218, "feed_secs": 15.65, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開。幫我導航到內湖科學園區，然後選最快的路線，不要走。", "error": null } ] }, "service-trt-2": { "environment": { "target": "ws://127.0.0.1:19242/stt/sensevoice", "uri": "ws://127.0.0.1:19242/stt/sensevoice", "warmup_runs": 3, "single_connection_series": true }, "generated": "2026-09-05T13:05:43", "results": [ { "feed": "bulk", "target_s": 3.0, "duration_s": 3.0, "status": "ok", "server_latency_ms": 13.57, "server_rtf": 0.005, "server_ttft_ms": 12.12, "client_final_latency_ms": 19.73, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.02, "text": "幫我導航到內湖科學園區，然後。", "error": null }, { "feed": "bulk", "target_s": 6.0, "duration_s": 6.0, "status": "ok", "server_latency_ms": 17.64, "server_rtf": 0.003, "server_ttft_ms": 12.32, "client_final_latency_ms": 30.67, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.05, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。", "error": null }, { "feed": "bulk", "target_s": 15.0, "duration_s": 15.0, "status": "ok", "server_latency_ms": 68.84, "server_rtf": 0.002, "server_ttft_ms": 18.24, "client_final_latency_ms": 74.78, "interim_count": 1, "max_interim_gap_s": null, "feed_secs": 0.12, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開大。幫我導航到內湖科學園區，然後選最快的路線。不要走。", "error": null }, { "feed": "realtime", "target_s": 3.0, "duration_s": 3.0, "status": "ok", "server_latency_ms": 11.04, "server_rtf": 0.026, "server_ttft_ms": 11.45, "client_final_latency_ms": 13.76, "interim_count": 6, "max_interim_gap_s": 0.523, "feed_secs": 3.15, "text": "幫我導航到內湖科學園區，然後。", "error": null }, { "feed": "realtime", "target_s": 6.0, "duration_s": 6.0, "status": "ok", "server_latency_ms": 13.26, "server_rtf": 0.014, "server_ttft_ms": 10.91, "client_final_latency_ms": 22.97, "interim_count": 9, "max_interim_gap_s": 2.429, "feed_secs": 6.22, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。", "error": null }, { "feed": "realtime", "target_s": 15.0, "duration_s": 15.0, "status": "ok", "server_latency_ms": 60.46, "server_rtf": 0.026, "server_ttft_ms": 10.46, "client_final_latency_ms": 63.13, "interim_count": 26, "max_interim_gap_s": 1.433, "feed_secs": 15.61, "text": "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開大。幫我導航到內湖科學園區，然後選最快的路線。不要走。", "error": null } ] } } } }, "audio_sha256": { "clip-3s.wav": "11034b66e2ae87bdec98dfe7401642387b85262b2e9c222b9e201030dc397a1c", "clip-15s.wav": "4950a411664da365f4b8cfe13838f881430f01f2fed8b2c3335719f6b791ff9f", "clip-6s.wav": "9566be6ccf6c7307830514e0fe5be86250c493efaef2800f27bfccc1ca1850b0" }, "final_success_count": 72, "text_variants": [ { "platform": "4090", "provider": "cuda", "feed": "bulk", "seconds": 3, "unique_texts": [ "幫我導航到內湖科學園區，然後。" ] }, { "platform": "4090", "provider": "trt", "feed": "bulk", "seconds": 3, "unique_texts": [ "幫我導航到內湖科學園區，然後。" ] }, { "platform": "4090", "provider": "cuda", "feed": "bulk", "seconds": 6, "unique_texts": [ "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。" ] }, { "platform": "4090", "provider": "trt", "feed": "bulk", "seconds": 6, "unique_texts": [ "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。" ] }, { "platform": "4090", "provider": "cuda", "feed": "bulk", "seconds": 15, "unique_texts": [ "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開。幫我導航到內湖科學園區，然後選最快的路線，不要走。", "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開大。幫我導航到內湖科學園區，然後選最快的路線。不要走。" ] }, { "platform": "4090", "provider": "trt", "feed": "bulk", "seconds": 15, "unique_texts": [ "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開大。幫我導航到內湖科學園區，然後選最快的路線。不要走。" ] }, { "platform": "4090", "provider": "cuda", "feed": "realtime", "seconds": 3, "unique_texts": [ "幫我導航到內湖科學園區，然後。" ] }, { "platform": "4090", "provider": "trt", "feed": "realtime", "seconds": 3, "unique_texts": [ "幫我導航到內湖科學園區，然後。" ] }, { "platform": "4090", "provider": "cuda", "feed": "realtime", "seconds": 6, "unique_texts": [ "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。" ] }, { "platform": "4090", "provider": "trt", "feed": "realtime", "seconds": 6, "unique_texts": [ "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。" ] }, { "platform": "4090", "provider": "cuda", "feed": "realtime", "seconds": 15, "unique_texts": [ "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開大。幫我導航到內湖科學園區，然後選最快的路線。不要走。" ] }, { "platform": "4090", "provider": "trt", "feed": "realtime", "seconds": 15, "unique_texts": [ "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開大。幫我導航到內湖科學園區，然後選最快的路線。不要走。" ] }, { "platform": "orin", "provider": "cuda", "feed": "bulk", "seconds": 3, "unique_texts": [ "幫我導航到內湖科學園區，然後。" ] }, { "platform": "orin", "provider": "trt", "feed": "bulk", "seconds": 3, "unique_texts": [ "幫我導航到內湖科學園區，然後。" ] }, { "platform": "orin", "provider": "cuda", "feed": "bulk", "seconds": 6, "unique_texts": [ "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。" ] }, { "platform": "orin", "provider": "trt", "feed": "bulk", "seconds": 6, "unique_texts": [ "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。" ] }, { "platform": "orin", "provider": "cuda", "feed": "bulk", "seconds": 15, "unique_texts": [ "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開大。幫我導航到內湖科學園區，然後選最快的路線。不要走。" ] }, { "platform": "orin", "provider": "trt", "feed": "bulk", "seconds": 15, "unique_texts": [ "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開大。幫我導航到內湖科學園區，然後選最快的路線。不要走。" ] }, { "platform": "orin", "provider": "cuda", "feed": "realtime", "seconds": 3, "unique_texts": [ "幫我導航到內湖科學園區，然後。" ] }, { "platform": "orin", "provider": "trt", "feed": "realtime", "seconds": 3, "unique_texts": [ "幫我導航到內湖科學園區，然後。" ] }, { "platform": "orin", "provider": "cuda", "feed": "realtime", "seconds": 6, "unique_texts": [ "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。" ] }, { "platform": "orin", "provider": "trt", "feed": "realtime", "seconds": 6, "unique_texts": [ "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋。" ] }, { "platform": "orin", "provider": "cuda", "feed": "realtime", "seconds": 15, "unique_texts": [ "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開大。幫我導航到內湖科學園區，然後選最快的路線。不要走。" ] }, { "platform": "orin", "provider": "trt", "feed": "realtime", "seconds": 15, "unique_texts": [ "幫我導航到內湖科學園區，然後選最快的路線，不要走高架橋，接着把冷氣打開，溫度調到24度，風量開大。幫我導航到內湖科學園區，然後選最快的路線。不要走。" ] } ], "report_url": "https://kitt-stt-orin-tensorrt.pages.dev/" }