test: add more llama_v3.3_70b cases in perf test (NVIDIA#4979)

Signed-off-by: ruodil <[email protected]>
Co-authored-by: Larry <[email protected]>

Author: ruodil

tests/integration/defs/perf/test_perf.py

@@ -48,6 +48,7 @@
     "llama_v3.1_8b_instruct_fp4":
     "modelopt-hf-model-hub/Llama-3.1-8B-Instruct-fp4",
     "llama_v3.1_70b": "llama-3.1-model/Meta-Llama-3.1-70B",
+    "llama_v3.3_70b_instruct": "llama-3.3-models/Llama-3.3-70B-Instruct",
     "llama_v3.1_70b_instruct_fp8": "llama-3.1-model/Llama-3.1-70B-Instruct-FP8",
     "llama_v3.3_70b_instruct_fp8":
     "modelopt-hf-model-hub/Llama-3.3-70B-Instruct-fp8",

tests/integration/test_lists/qa/trt_llm_release_perf_test.yml

@@ -275,7 +275,7 @@ trt_llm_release_perf_test:
       - '*h20*'
   tests:
   - perf/test_perf.py::test_perf[llama_v3.1_70b-bench-bfloat16-input_output_len:512,200-quant:fp8-tp:4]
-  - perf/test_perf.py::test_perf[llama_v3.3_70b_instruct_fp8-bench-pytorch-float8-input_output_len:128,128-tp:4]
+  - perf/test_perf.py::test_perf[llama_v3.3_70b_instruct_fp8-bench-float8-input_output_len:128,128-tp:4]
   - perf/test_perf.py::test_perf[llama_v3.3_nemotron_super_49b-bench-bfloat16-maxbs:64-input_output_len:5000,500-reqs:4-con:1-gpus:4]
   - perf/test_perf.py::test_perf[llama_v3.3_nemotron_super_49b-bench-bfloat16-maxbs:64-input_output_len:5000,500-quant:fp8-reqs:4-con:1-gpus:4]
   - perf/test_perf.py::test_perf[llama_v3.3_nemotron_super_49b-bench-bfloat16-maxbs:64-input_output_len:500,2000-reqs:4-con:1-gpus:4]
@@ -304,7 +304,8 @@ trt_llm_release_perf_test:
   - perf/test_perf.py::test_perf[llama_v3.1_70b_instruct-bench-bfloat16-input_output_len:200,2000-reqs:8-con:1-gpus:8] # timeout for h20, move to l2 test
   - perf/test_perf.py::test_perf[llama_v3.1_70b_instruct-bench-bfloat16-input_output_len:2000,200-reqs:64-gpus:8]
   - perf/test_perf.py::test_perf[llama_v3.1_70b_instruct-bench-streaming-bfloat16-input_output_len:2000,200-reqs:64-gpus:8]
-  - perf/test_perf.py::test_perf[llama_v3.3_70b_instruct_fp8-bench-pytorch-float8-input_output_len:128,128-gpus:8]
+  - perf/test_perf.py::test_perf[llama_v3.3_70b_instruct-bench-bfloat16-input_output_len:128,128-gpus:8]
+  - perf/test_perf.py::test_perf[llama_v3.3_70b_instruct-bench-bfloat16-maxbs:16-input_output_len:5000,500-reqs:64-con:250-gpus:8]
   - perf/test_perf.py::test_perf[gpt_20b-bench-float16-maxbs:8-input_output_len:128,128-reqs:80-gpus:8]
   - perf/test_perf.py::test_perf[gpt_20b-bench-float16-maxbs:8-input_output_len:512,32-reqs:80-gpus:8]
 
@@ -361,10 +362,11 @@ trt_llm_release_perf_test:
   tests:
   - perf/test_perf.py::test_perf[llama_v3.3_70b_instruct_fp8-bench-float8-input_output_len:128,128-gpus:8]
   - perf/test_perf.py::test_perf[llama_v3.3_70b_instruct_fp8-bench-float8-input_output_len:512,32-gpus:8]
-  - perf/test_perf.py::test_perf[llama_v3.3_70b_instruct_fp8-bench-pytorch-streaming-float8-input_output_len:512,32-gpus:8]
-  - perf/test_perf.py::test_perf[llama_v3.3_70b_instruct_fp8-bench-pytorch-float8-input_output_len:128,128-gpus:8]
-  - perf/test_perf.py::test_perf[llama_v3.3_70b_instruct_fp8-bench-pytorch-streaming-float8-input_output_len:128,128-gpus:8]
-  - perf/test_perf.py::test_perf[llama_v3.3_70b_instruct_fp8-bench-pytorch-float8-input_output_len:512,32-gpus:8]
+  - perf/test_perf.py::test_perf[llama_v3.3_70b_instruct_fp8-bench-streaming-float8-maxbs:16-input_output_len:512,32-gpus:8]
+  - perf/test_perf.py::test_perf[llama_v3.3_70b_instruct-bench-bfloat16-maxbs:16-maxnt:5000-input_output_len:5000,500-quant:fp8-reqs:8-con:1-gpus:8]
+  - perf/test_perf.py::test_perf[llama_v3.3_70b_instruct-bench-bfloat16-maxbs:16-maxnt:5000-input_output_len:5000,500-quant:fp8-reqs:64-con:250-gpus:8]
+  - perf/test_perf.py::test_perf[llama_v3.3_70b_instruct-bench-bfloat16-maxbs:16-input_output_len:500,2000-quant:fp8-reqs:8-con:1-gpus:8]
+  - perf/test_perf.py::test_perf[llama_v3.3_70b_instruct-bench-bfloat16-maxbs:16-input_output_len:500,2000-quant:fp8-reqs:64-con:250-gpus:8]
   - perf/test_perf.py::test_perf[llama_v3.3_nemotron_super_49b-bench-pytorch-bfloat16-input_output_len:128,128-gpus:8]
 
 
@@ -383,7 +385,7 @@ trt_llm_release_perf_test:
   - perf/test_perf.py::test_perf[llama_v3.1_8b_instruct_fp4-bench-pytorch-float4-input_output_len:128,128]
   - perf/test_perf.py::test_perf[llama_v3.1_70b-bench-pytorch-bfloat16-input_output_len:128,128-tp:2-gpus:2]
   - perf/test_perf.py::test_perf[llama_v3.1_70b_instruct_fp8-bench-pytorch-float8-input_output_len:128,128]
-  - perf/test_perf.py::test_perf[llama_v3.3_nemotron_49b-bench-pytorch-bfloat16-input_output_len:128,128-tp:2-gpus:2]
+  - perf/test_perf.py::test_perf[llama_v3.3_nemotron_super_49b-bench-pytorch-bfloat16-input_output_len:128,128-tp:2-gpus:2]
   - perf/test_perf.py::test_perf[deepseek_v3_lite_nvfp4-bench-pytorch-float4-input_output_len:128,128]
   - perf/test_perf.py::test_perf[deepseek_v3_lite_nvfp4-bench-pytorch-streaming-float4-input_output_len:128,128]
   - perf/test_perf.py::test_perf[deepseek_v3_lite_fp8-bench-pytorch-float8-input_output_len:128,128]