AI/amd-strix-halo-vllm-toolboxes
2
0
フォーク 0
コード イシュー プルリクエスト Actions パッケージ プロジェクト リリース Wiki アクティビティ
ファイル
6875f62ccfaa634151700c3482e5edeeca645a8f
amd-strix-halo-vllm-toolboxes/docs/results.json
T
Donato Capitella e726d406fa updated benchmarks, fix start-vllm
2026-02-23 19:39:19 +00:00

814 行
21 KiB
JSON
Raw Blame 履歴

{
"runs": [
{
"model": "meta-llama/Meta-Llama-3.1-8B-Instruct",
"model_clean": "meta-llama/Meta-Llama-3.1-8B-Instruct",
"env": "TP2",
"gpu_config": "dual",
"quant": "BF16",
"params_b": 8.0,
"name_params_b": 8.0,
"backend": "Triton",
"network": "Ethernet",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 613.9321196754427
},
{
"model": "google/gemma-3-12b-it",
"model_clean": "google/gemma-3-12b-it",
"env": "TP2",
"gpu_config": "dual",
"quant": "BF16",
"params_b": 12.0,
"name_params_b": 12.0,
"backend": "Triton",
"network": "Ethernet",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 291.5155379231269
},
{
"model": "Qwen/Qwen3-14B-AWQ",
"model_clean": "Qwen/Qwen3-14B-AWQ",
"env": "TP2",
"gpu_config": "dual",
"quant": "AWQ",
"params_b": 14.0,
"name_params_b": 14.0,
"backend": "Triton",
"network": "Ethernet",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 280.05330212131406
},
{
"model": "openai/gpt-oss-20b",
"model_clean": "openai/gpt-oss-20b",
"env": "TP2",
"gpu_config": "dual",
"quant": "BF16",
"params_b": 20.0,
"name_params_b": 20.0,
"backend": "Triton",
"network": "Ethernet",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 602.6345456319963
},
{
"model": "openai/gpt-oss-120b",
"model_clean": "openai/gpt-oss-120b",
"env": "TP2",
"gpu_config": "dual",
"quant": "BF16",
"params_b": 120.0,
"name_params_b": 120.0,
"backend": "Triton",
"network": "Ethernet",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 85.2809331488931
},
{
"model": "zai-org/GLM-4.7-Flash",
"model_clean": "zai-org/GLM-4.7-Flash",
"env": "TP2",
"gpu_config": "dual",
"quant": "BF16",
"params_b": null,
"name_params_b": null,
"backend": "Triton",
"network": "Ethernet",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 300.773560320048
},
{
"model": "btbtyler09/Qwen3-Coder-30B-A3B-Instruct-gptq-4bit",
"model_clean": "btbtyler09/Qwen3-Coder-30B-A3B-Instruct-gptq-4bit",
"env": "TP2",
"gpu_config": "dual",
"quant": "GPTQ",
"params_b": 30.0,
"name_params_b": 30.0,
"backend": "Triton",
"network": "Ethernet",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 321.88057686801585
},
{
"model": "btbtyler09/Qwen3-Coder-30B-A3B-Instruct-gptq-8bit",
"model_clean": "btbtyler09/Qwen3-Coder-30B-A3B-Instruct-gptq-8bit",
"env": "TP2",
"gpu_config": "dual",
"quant": "GPTQ",
"params_b": 30.0,
"name_params_b": 30.0,
"backend": "Triton",
"network": "Ethernet",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 274.46004720922855
},
{
"model": "dazipe/Qwen3-Next-80B-A3B-Instruct-GPTQ-Int4A16",
"model_clean": "dazipe/Qwen3-Next-80B-A3B-Instruct-GPTQ-Int4A16",
"env": "TP2",
"gpu_config": "dual",
"quant": "GPTQ",
"params_b": 80.0,
"name_params_b": 80.0,
"backend": "Triton",
"network": "Ethernet",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 182.16229690959702
},
{
"model": "mratsim/MiniMax-M2.5-BF16-INT4-AWQ",
"model_clean": "mratsim/MiniMax-M2.5-BF16-INT4-AWQ",
"env": "TP2",
"gpu_config": "dual",
"quant": "BF16",
"params_b": null,
"name_params_b": null,
"backend": "Triton",
"network": "Ethernet",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 107.62460878889469
},
{
"model": "meta-llama/Meta-Llama-3.1-8B-Instruct",
"model_clean": "meta-llama/Meta-Llama-3.1-8B-Instruct",
"env": "TP2",
"gpu_config": "dual",
"quant": "BF16",
"params_b": 8.0,
"name_params_b": 8.0,
"backend": "Triton",
"network": "RoCE",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 590.0492703672895
},
{
"model": "google/gemma-3-12b-it",
"model_clean": "google/gemma-3-12b-it",
"env": "TP2",
"gpu_config": "dual",
"quant": "BF16",
"params_b": 12.0,
"name_params_b": 12.0,
"backend": "Triton",
"network": "RoCE",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 285.8275921888489
},
{
"model": "Qwen/Qwen3-14B-AWQ",
"model_clean": "Qwen/Qwen3-14B-AWQ",
"env": "TP2",
"gpu_config": "dual",
"quant": "AWQ",
"params_b": 14.0,
"name_params_b": 14.0,
"backend": "Triton",
"network": "RoCE",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 346.2003835540928
},
{
"model": "openai/gpt-oss-20b",
"model_clean": "openai/gpt-oss-20b",
"env": "TP2",
"gpu_config": "dual",
"quant": "BF16",
"params_b": 20.0,
"name_params_b": 20.0,
"backend": "Triton",
"network": "RoCE",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 607.216674264294
},
{
"model": "openai/gpt-oss-120b",
"model_clean": "openai/gpt-oss-120b",
"env": "TP2",
"gpu_config": "dual",
"quant": "BF16",
"params_b": 120.0,
"name_params_b": 120.0,
"backend": "Triton",
"network": "RoCE",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 122.62029501860121
},
{
"model": "zai-org/GLM-4.7-Flash",
"model_clean": "zai-org/GLM-4.7-Flash",
"env": "TP2",
"gpu_config": "dual",
"quant": "BF16",
"params_b": null,
"name_params_b": null,
"backend": "Triton",
"network": "RoCE",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 366.77052981888835
},
{
"model": "btbtyler09/Qwen3-Coder-30B-A3B-Instruct-gptq-4bit",
"model_clean": "btbtyler09/Qwen3-Coder-30B-A3B-Instruct-gptq-4bit",
"env": "TP2",
"gpu_config": "dual",
"quant": "GPTQ",
"params_b": 30.0,
"name_params_b": 30.0,
"backend": "Triton",
"network": "RoCE",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 320.0197833991106
},
{
"model": "btbtyler09/Qwen3-Coder-30B-A3B-Instruct-gptq-8bit",
"model_clean": "btbtyler09/Qwen3-Coder-30B-A3B-Instruct-gptq-8bit",
"env": "TP2",
"gpu_config": "dual",
"quant": "GPTQ",
"params_b": 30.0,
"name_params_b": 30.0,
"backend": "Triton",
"network": "RoCE",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 279.7240042842149
},
{
"model": "dazipe/Qwen3-Next-80B-A3B-Instruct-GPTQ-Int4A16",
"model_clean": "dazipe/Qwen3-Next-80B-A3B-Instruct-GPTQ-Int4A16",
"env": "TP2",
"gpu_config": "dual",
"quant": "GPTQ",
"params_b": 80.0,
"name_params_b": 80.0,
"backend": "Triton",
"network": "RoCE",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 196.2262690032198
},
{
"model": "mratsim/MiniMax-M2.5-BF16-INT4-AWQ",
"model_clean": "mratsim/MiniMax-M2.5-BF16-INT4-AWQ",
"env": "TP2",
"gpu_config": "dual",
"quant": "BF16",
"params_b": null,
"name_params_b": null,
"backend": "Triton",
"network": "RoCE",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 98.85048345093716
},
{
"model": "meta-llama/Meta-Llama-3.1-8B-Instruct",
"model_clean": "meta-llama/Meta-Llama-3.1-8B-Instruct",
"env": "TP1",
"gpu_config": "single",
"quant": "BF16",
"params_b": 8.0,
"name_params_b": 8.0,
"backend": "Triton",
"network": "RoCE",
"error": false,
"test": "Throughput",
"tp": 1,
"tps_mean": 369.23212230245684
},
{
"model": "Qwen/Qwen3-14B-AWQ",
"model_clean": "Qwen/Qwen3-14B-AWQ",
"env": "TP1",
"gpu_config": "single",
"quant": "AWQ",
"params_b": 14.0,
"name_params_b": 14.0,
"backend": "Triton",
"network": "RoCE",
"error": false,
"test": "Throughput",
"tp": 1,
"tps_mean": 159.8857312165796
},
{
"model": "openai/gpt-oss-20b",
"model_clean": "openai/gpt-oss-20b",
"env": "TP1",
"gpu_config": "single",
"quant": "BF16",
"params_b": 20.0,
"name_params_b": 20.0,
"backend": "Triton",
"network": "RoCE",
"error": false,
"test": "Throughput",
"tp": 1,
"tps_mean": 411.03557062490586
},
{
"model": "openai/gpt-oss-120b",
"model_clean": "openai/gpt-oss-120b",
"env": "TP1",
"gpu_config": "single",
"quant": "BF16",
"params_b": 120.0,
"name_params_b": 120.0,
"backend": "Triton",
"network": "RoCE",
"error": false,
"test": "Throughput",
"tp": 1,
"tps_mean": 75.0407548829671
},
{
"model": "zai-org/GLM-4.7-Flash",
"model_clean": "zai-org/GLM-4.7-Flash",
"env": "TP1",
"gpu_config": "single",
"quant": "BF16",
"params_b": null,
"name_params_b": null,
"backend": "Triton",
"network": "RoCE",
"error": false,
"test": "Throughput",
"tp": 1,
"tps_mean": 239.57478116575834
},
{
"model": "btbtyler09/Qwen3-Coder-30B-A3B-Instruct-gptq-4bit",
"model_clean": "btbtyler09/Qwen3-Coder-30B-A3B-Instruct-gptq-4bit",
"env": "TP1",
"gpu_config": "single",
"quant": "GPTQ",
"params_b": 30.0,
"name_params_b": 30.0,
"backend": "Triton",
"network": "RoCE",
"error": false,
"test": "Throughput",
"tp": 1,
"tps_mean": 213.74630950782364
},
{
"model": "btbtyler09/Qwen3-Coder-30B-A3B-Instruct-gptq-8bit",
"model_clean": "btbtyler09/Qwen3-Coder-30B-A3B-Instruct-gptq-8bit",
"env": "TP1",
"gpu_config": "single",
"quant": "GPTQ",
"params_b": 30.0,
"name_params_b": 30.0,
"backend": "Triton",
"network": "RoCE",
"error": false,
"test": "Throughput",
"tp": 1,
"tps_mean": 186.03115379827653
},
{
"model": "dazipe/Qwen3-Next-80B-A3B-Instruct-GPTQ-Int4A16",
"model_clean": "dazipe/Qwen3-Next-80B-A3B-Instruct-GPTQ-Int4A16",
"env": "TP1",
"gpu_config": "single",
"quant": "GPTQ",
"params_b": 80.0,
"name_params_b": 80.0,
"backend": "Triton",
"network": "RoCE",
"error": false,
"test": "Throughput",
"tp": 1,
"tps_mean": 125.65027253668944
},
{
"model": "google/gemma-3-12b-it",
"model_clean": "google/gemma-3-12b-it",
"env": "TP1",
"gpu_config": "single",
"quant": "BF16",
"params_b": 12.0,
"name_params_b": 12.0,
"backend": "Triton",
"network": "RoCE",
"error": false,
"test": "Throughput",
"tp": 1,
"tps_mean": 159.95620436815713
},
{
"model": "meta-llama/Meta-Llama-3.1-8B-Instruct",
"model_clean": "meta-llama/Meta-Llama-3.1-8B-Instruct",
"env": "TP2",
"gpu_config": "dual",
"quant": "BF16",
"params_b": 8.0,
"name_params_b": 8.0,
"backend": "ROCm",
"network": "Ethernet",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 682.4360360233941
},
{
"model": "google/gemma-3-12b-it",
"model_clean": "google/gemma-3-12b-it",
"env": "TP2",
"gpu_config": "dual",
"quant": "BF16",
"params_b": 12.0,
"name_params_b": 12.0,
"backend": "ROCm",
"network": "Ethernet",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 398.09474781142933
},
{
"model": "Qwen/Qwen3-14B-AWQ",
"model_clean": "Qwen/Qwen3-14B-AWQ",
"env": "TP2",
"gpu_config": "dual",
"quant": "AWQ",
"params_b": 14.0,
"name_params_b": 14.0,
"backend": "ROCm",
"network": "Ethernet",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 295.31575874126105
},
{
"model": "openai/gpt-oss-20b",
"model_clean": "openai/gpt-oss-20b",
"env": "TP2",
"gpu_config": "dual",
"quant": "BF16",
"params_b": 20.0,
"name_params_b": 20.0,
"backend": "ROCm",
"network": "Ethernet",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 490.93757442090305
},
{
"model": "openai/gpt-oss-120b",
"model_clean": "openai/gpt-oss-120b",
"env": "TP2",
"gpu_config": "dual",
"quant": "BF16",
"params_b": 120.0,
"name_params_b": 120.0,
"backend": "ROCm",
"network": "Ethernet",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 86.0910643999307
},
{
"model": "btbtyler09/Qwen3-Coder-30B-A3B-Instruct-gptq-4bit",
"model_clean": "btbtyler09/Qwen3-Coder-30B-A3B-Instruct-gptq-4bit",
"env": "TP2",
"gpu_config": "dual",
"quant": "GPTQ",
"params_b": 30.0,
"name_params_b": 30.0,
"backend": "ROCm",
"network": "Ethernet",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 321.6166453306162
},
{
"model": "btbtyler09/Qwen3-Coder-30B-A3B-Instruct-gptq-8bit",
"model_clean": "btbtyler09/Qwen3-Coder-30B-A3B-Instruct-gptq-8bit",
"env": "TP2",
"gpu_config": "dual",
"quant": "GPTQ",
"params_b": 30.0,
"name_params_b": 30.0,
"backend": "ROCm",
"network": "Ethernet",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 283.6309502128471
},
{
"model": "dazipe/Qwen3-Next-80B-A3B-Instruct-GPTQ-Int4A16",
"model_clean": "dazipe/Qwen3-Next-80B-A3B-Instruct-GPTQ-Int4A16",
"env": "TP2",
"gpu_config": "dual",
"quant": "GPTQ",
"params_b": 80.0,
"name_params_b": 80.0,
"backend": "ROCm",
"network": "Ethernet",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 182.9186467257061
},
{
"model": "mratsim/MiniMax-M2.5-BF16-INT4-AWQ",
"model_clean": "mratsim/MiniMax-M2.5-BF16-INT4-AWQ",
"env": "TP2",
"gpu_config": "dual",
"quant": "BF16",
"params_b": null,
"name_params_b": null,
"backend": "ROCm",
"network": "Ethernet",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 98.67941666807306
},
{
"model": "meta-llama/Meta-Llama-3.1-8B-Instruct",
"model_clean": "meta-llama/Meta-Llama-3.1-8B-Instruct",
"env": "TP2",
"gpu_config": "dual",
"quant": "BF16",
"params_b": 8.0,
"name_params_b": 8.0,
"backend": "ROCm",
"network": "RoCE",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 650.1471716939323
},
{
"model": "google/gemma-3-12b-it",
"model_clean": "google/gemma-3-12b-it",
"env": "TP2",
"gpu_config": "dual",
"quant": "BF16",
"params_b": 12.0,
"name_params_b": 12.0,
"backend": "ROCm",
"network": "RoCE",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 397.9710386242193
},
{
"model": "Qwen/Qwen3-14B-AWQ",
"model_clean": "Qwen/Qwen3-14B-AWQ",
"env": "TP2",
"gpu_config": "dual",
"quant": "AWQ",
"params_b": 14.0,
"name_params_b": 14.0,
"backend": "ROCm",
"network": "RoCE",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 367.5704596781314
},
{
"model": "openai/gpt-oss-20b",
"model_clean": "openai/gpt-oss-20b",
"env": "TP2",
"gpu_config": "dual",
"quant": "BF16",
"params_b": 20.0,
"name_params_b": 20.0,
"backend": "ROCm",
"network": "RoCE",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 601.2567608739705
},
{
"model": "openai/gpt-oss-120b",
"model_clean": "openai/gpt-oss-120b",
"env": "TP2",
"gpu_config": "dual",
"quant": "BF16",
"params_b": 120.0,
"name_params_b": 120.0,
"backend": "ROCm",
"network": "RoCE",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 118.34229353876268
},
{
"model": "btbtyler09/Qwen3-Coder-30B-A3B-Instruct-gptq-4bit",
"model_clean": "btbtyler09/Qwen3-Coder-30B-A3B-Instruct-gptq-4bit",
"env": "TP2",
"gpu_config": "dual",
"quant": "GPTQ",
"params_b": 30.0,
"name_params_b": 30.0,
"backend": "ROCm",
"network": "RoCE",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 333.147212194374
},
{
"model": "btbtyler09/Qwen3-Coder-30B-A3B-Instruct-gptq-8bit",
"model_clean": "btbtyler09/Qwen3-Coder-30B-A3B-Instruct-gptq-8bit",
"env": "TP2",
"gpu_config": "dual",
"quant": "GPTQ",
"params_b": 30.0,
"name_params_b": 30.0,
"backend": "ROCm",
"network": "RoCE",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 295.0301359026215
},
{
"model": "dazipe/Qwen3-Next-80B-A3B-Instruct-GPTQ-Int4A16",
"model_clean": "dazipe/Qwen3-Next-80B-A3B-Instruct-GPTQ-Int4A16",
"env": "TP2",
"gpu_config": "dual",
"quant": "GPTQ",
"params_b": 80.0,
"name_params_b": 80.0,
"backend": "ROCm",
"network": "RoCE",
"error": false,
"test": "Throughput",
"tp": 2,
"tps_mean": 193.87438091607942
},
{
"model": "meta-llama/Meta-Llama-3.1-8B-Instruct",
"model_clean": "meta-llama/Meta-Llama-3.1-8B-Instruct",
"env": "TP1",
"gpu_config": "single",
"quant": "BF16",
"params_b": 8.0,
"name_params_b": 8.0,
"backend": "ROCm",
"network": "RoCE",
"error": false,
"test": "Throughput",
"tp": 1,
"tps_mean": 433.5736973626181
},
{
"model": "Qwen/Qwen3-14B-AWQ",
"model_clean": "Qwen/Qwen3-14B-AWQ",
"env": "TP1",
"gpu_config": "single",
"quant": "AWQ",
"params_b": 14.0,
"name_params_b": 14.0,
"backend": "ROCm",
"network": "RoCE",
"error": false,
"test": "Throughput",
"tp": 1,
"tps_mean": 180.43566315423652
},
{
"model": "openai/gpt-oss-20b",
"model_clean": "openai/gpt-oss-20b",
"env": "TP1",
"gpu_config": "single",
"quant": "BF16",
"params_b": 20.0,
"name_params_b": 20.0,
"backend": "ROCm",
"network": "RoCE",
"error": false,
"test": "Throughput",
"tp": 1,
"tps_mean": 405.0974544317216
},
{
"model": "openai/gpt-oss-120b",
"model_clean": "openai/gpt-oss-120b",
"env": "TP1",
"gpu_config": "single",
"quant": "BF16",
"params_b": 120.0,
"name_params_b": 120.0,
"backend": "ROCm",
"network": "RoCE",
"error": false,
"test": "Throughput",
"tp": 1,
"tps_mean": 74.75385852312364
},
{
"model": "btbtyler09/Qwen3-Coder-30B-A3B-Instruct-gptq-4bit",
"model_clean": "btbtyler09/Qwen3-Coder-30B-A3B-Instruct-gptq-4bit",
"env": "TP1",
"gpu_config": "single",
"quant": "GPTQ",
"params_b": 30.0,
"name_params_b": 30.0,
"backend": "ROCm",
"network": "RoCE",
"error": false,
"test": "Throughput",
"tp": 1,
"tps_mean": 214.65152188564062
},
{
"model": "btbtyler09/Qwen3-Coder-30B-A3B-Instruct-gptq-8bit",
"model_clean": "btbtyler09/Qwen3-Coder-30B-A3B-Instruct-gptq-8bit",
"env": "TP1",
"gpu_config": "single",
"quant": "GPTQ",
"params_b": 30.0,
"name_params_b": 30.0,
"backend": "ROCm",
"network": "RoCE",
"error": false,
"test": "Throughput",
"tp": 1,
"tps_mean": 188.17083503449163
},
{
"model": "dazipe/Qwen3-Next-80B-A3B-Instruct-GPTQ-Int4A16",
"model_clean": "dazipe/Qwen3-Next-80B-A3B-Instruct-GPTQ-Int4A16",
"env": "TP1",
"gpu_config": "single",
"quant": "GPTQ",
"params_b": 80.0,
"name_params_b": 80.0,
"backend": "ROCm",
"network": "RoCE",
"error": false,
"test": "Throughput",
"tp": 1,
"tps_mean": 128.30078036872973
},
{
"model": "google/gemma-3-12b-it",
"model_clean": "google/gemma-3-12b-it",
"env": "TP1",
"gpu_config": "single",
"quant": "BF16",
"params_b": 12.0,
"name_params_b": 12.0,
"backend": "ROCm",
"network": "RoCE",
"error": false,
"test": "Throughput",
"tp": 1,
"tps_mean": 267.99881204205957
}
]
}
新しいイシューから参照 Git Blameを表示 パーマリンクをコピー