kanshan/TensorRT-LLMs
1
0
Fork 0
mirror of https://github.com/NVIDIA/TensorRT-LLM.git synced 2026-01-14 06:27:45 +08:00
Code Issues Actions 1 Packages Projects Releases Wiki Activity

[None][test] Filter out all fp8 test case for A100. (#8420)

Browse Source 
Signed-off-by: yufeiwu <230315618+yufeiwu-nv@users.noreply.github.com>
This commit is contained in:
yufeiwu-nv 2025-10-17 11:42:50 +08:00 committed by GitHub
parent 70a0f5beb6
commit 1e1f430163
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 22 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

24
tests/integration/test_lists/qa/llm_perf_nim.yml
View File

@ -27,11 +27,31 @@ llm_perf_nim:
# cpp
- perf/test_perf.py::test_perf[phi_4_mini_instruct-bench-bfloat16-maxbs:32-maxnt:5000-input_output_len:5000,500-con:250]
- perf/test_perf.py::test_perf[phi_4_mini_instruct-bench-bfloat16-maxbs:32-input_output_len:500,2000-con:250]
- perf/test_perf.py::test_perf[phi_4_mini_instruct-bench-bfloat16-maxbs:32-input_output_len:500,2000-quant:fp8-con:250]
# reduced 'reqs' to fit timeout limit
- perf/test_perf.py::test_perf[phi_4_mini_instruct-bench-bfloat16-maxbs:32-input_output_len:500,2000-reqs:8-con:1]
# FP8 specific tests
- condition:
terms:
supports_fp8: true
ranges:
system_gpu_count:
gte: 1
wildcards:
gpu:
- '*h100*'
- '*h200*'
- '*l40s*'
- '*l20*'
- '*h20*'
tests:
# Phi-4-mini-instruct
- perf/test_perf.py::test_perf[phi_4_mini_instruct-bench-bfloat16-maxbs:32-input_output_len:500,2000-quant:fp8-con:250]
# reduced 'reqs' to fit timeout limit
- perf/test_perf.py::test_perf[phi_4_mini_instruct-bench-bfloat16-maxbs:32-input_output_len:500,2000-quant:fp8-reqs:8-con:1]
- condition:
ranges:
system_gpu_count:
@ -190,7 +210,6 @@ llm_perf_nim:
#llama_v3.2_1b
#trt backend
- perf/test_perf.py::test_perf[llama_v3.2_1b-bench-bfloat16-input_output_len:128,128-quant:fp8-gpus:2]
- perf/test_perf.py::test_perf[llama_v3.2_1b-bench-pytorch-bfloat16-input_output_len:128,128-quant:fp8-gpus:2]
- perf/test_perf.py::test_perf[llama_v3.2_1b-bench-bfloat16-input_output_len:512,32-quant:fp8-gpus:2]
- perf/test_perf.py::test_perf[llama_v3.2_1b-bench-bfloat16-input_output_len:512,200-quant:fp8-gpus:2]
#mistral_7b_v0.1
@ -199,6 +218,7 @@ llm_perf_nim:
- perf/test_perf.py::test_perf[mistral_7b_v0.1-bench-float16-input_output_len:500,2000-quant:fp8-tp:2]
# torch backend
- perf/test_perf.py::test_perf[mistral_7b_v0.1-bench-pytorch-float16-input_output_len:128,128]
- perf/test_perf.py::test_perf[llama_v3.2_1b-bench-pytorch-bfloat16-input_output_len:128,128-gpus:2]
#phi_3_mini_128k_instruct
#trt backend
- perf/test_perf.py::test_perf[phi_3_mini_128k_instruct-bench-float16-maxbs:128-input_output_len:1000,1000-quant:fp8-tp:2]