Skip to content

Update func and perf test cases for DGX-Spark #4

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
JennyLiu-nv wants to merge 14 commits into from
Closed

Update func and perf test cases for DGX-Spark #4

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
14 commits
Select commit Hold shift + click to select a range
83a7243
update NVILA-15B-FP16 match keywords
Jul 1, 2025
f997eb2
update perf case - only support pytorch backend
Jul 2, 2025
f038bf8
add model Mixtral-7B-Instruct-v0.3
Jul 2, 2025
5cd1740
add perf test cases
Jul 2, 2025
063e586
update func test
Jul 3, 2025
dc95d34
delete blanks
Jul 3, 2025
e6b226f
Fix whitespace and end-of-file issues in test list
Jul 3, 2025
30f3397
Update performance test configuration
Jul 3, 2025
589085e
Add performance test configurations with maxbs:1-maxnt:3000-input_out…
Jul 10, 2025
aadd70e
change perf test order
Jul 10, 2025
b6db2e0
change perf test order
Jul 10, 2025
4cb725a
change perf test order
Jul 10, 2025
5d774f5
delete some bigger input cases
Jul 10, 2025
2f794ef
waive some cases by bug
Jul 11, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 3 additions & 1 deletion tests/integration/defs/perf/test_perf.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -82,6 +82,7 @@
"deepseek_r1_distill_qwen_32b": "DeepSeek-R1/DeepSeek-R1-Distill-Qwen-32B",
"mixtral_8x22b_v0.1": "Mixtral-8x22B-v0.1",
"mistral_7b_v0.1": "mistral-7b-v0.1",
"mistral_7b_v0.3": "Mistral-7B-Instruct-v0.3",
"deepseek_r1_fp8": "DeepSeek-R1/DeepSeek-R1",
"deepseek_r1_nvfp4": "DeepSeek-R1/DeepSeek-R1-FP4",
"deepseek_v3_lite_fp8": "DeepSeek-V3-Lite/fp8",
Expand Down Expand Up @@ -137,6 +138,7 @@
"mistral_7b_v0.1_hf": "mistralai/Mistral-7B-v0.1",
"flan_t5_base_hf": "google/flan-t5-small",
"phi_4_mini_instruct_hf": "microsoft/Phi-4-mini-instruct",
"nvila_15b": "nvidia/NVILA-15B",
}
LORA_MODEL_PATH = {
"llama_v2_13b": "llama-models-v2/chinese-llama-2-lora-13b",
Expand Down Expand Up @@ -1379,8 +1381,8 @@ def get_commands(self):
data_cmd = self.get_prepare_data_command(
engine_dir, input_len, output_len)
data_cmds.append(data_cmd)

# Construct MPI command.
print(f"##########JJJJJJ data_cmd: {data_cmd}")
mpi_cmd = []
if num_gpus > 1 and num_gpus <= 8 and not self._config.runtime == "bench":
if cpu_socket_count_gt_1():
Expand Down
17 changes: 17 additions & 0 deletions tests/integration/defs/test_e2e.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1607,6 +1607,9 @@ def test_ptp_quickstart(llm_root, llm_venv):
pytest.param('Mixtral-8x7B-BF16',
'Mixtral-8x7B-Instruct-v0.1',
marks=skip_pre_blackwell),
pytest.param('Mistral-7B-Instruct-v0.3',
'Mistral-7B-Instruct-v0.3',
marks=skip_pre_blackwell),
pytest.param('Mistral-Nemo-12b-Base',
'Mistral-Nemo-Base-2407',
marks=skip_pre_blackwell),
Expand Down Expand Up @@ -2012,6 +2015,20 @@ def test_ptp_quickstart_multimodal(llm_root, llm_venv, model_name, model_path,
],
],
},
"NVILA-15B-FP16": {
"image": [
["stormy", "ocean", "waves", "clouds", "gray", "sky"],
["rock", "formation", "sunny", "sky", "clouds"],
["road", "busy", "car", "black", "blue"],
],
"video": [
["woman", "street", "night", "walking", "camera"],
[
"stunning", "earth", "space", "planet", "curvature", "dark",
"bright", "contrast", "illuminate"
],
],
},
"llava-v1.6-mistral-7b": {
"image": [
[
Expand Down
1 change: 1 addition & 0 deletions tests/integration/test_lists/qa/llm_release_digits_func.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -12,6 +12,7 @@ test_e2e.py::test_ptp_quickstart_advanced[Llama3.3-70B-FP8-modelopt-hf-model-hub
test_e2e.py::test_ptp_quickstart_advanced[Llama3.3-70B-FP4-modelopt-hf-model-hub/Llama-3.3-70B-Instruct-fp4]
test_e2e.py::test_ptp_quickstart_advanced[Nemotron-Super-49B-v1-BF16-nemotron-nas/Llama-3_3-Nemotron-Super-49B-v1]
test_e2e.py::test_ptp_quickstart_advanced[Mixtral-8x7B-BF16-Mixtral-8x7B-Instruct-v0.1]
test_e2e.py::test_ptp_quickstart_advanced[Mistral-7B-Instruct-v0.3-Mistral-7B-Instruct-v0.3]
test_e2e.py::test_ptp_quickstart_advanced[Mistral-Nemo-12b-Base-Mistral-Nemo-Base-2407]
test_e2e.py::test_ptp_quickstart_advanced[DeepSeek-R1-Distill-Qwen-32B-DeepSeek-R1/DeepSeek-R1-Distill-Qwen-32B]

Expand Down
56 changes: 33 additions & 23 deletions tests/integration/test_lists/qa/llm_release_digits_perf.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,28 +1,38 @@
perf/test_perf.py::test_perf[llama_v3.1_8b_instruct-bench-streaming-bfloat16-input_output_len:128,128]
perf/test_perf.py::test_perf[llama_v3.1_8b_instruct-bench-streaming-bfloat16-input_output_len:512,128]
perf/test_perf.py::test_perf[llama_v3.1_8b_instruct-bench-bfloat16-input_output_len:128,128]
perf/test_perf.py::test_perf[llama_v3.1_8b_instruct-bench-bfloat16-input_output_len:512,128]
perf/test_perf.py::test_perf[llama_v3.1_8b_instruct-bench-pytorch-bfloat16-maxbs:1-maxnt:3000-input_output_len:2048,128-reqs:32-con:1] #passed
perf/test_perf.py::test_perf[llama_v3.1_8b_instruct_fp8-bench-pytorch-float8-maxbs:1-maxnt:3000-input_output_len:2048,128-reqs:32-con:1] #passed
perf/test_perf.py::test_perf[mistral_7b_v0.3-bench-pytorch-float16-maxbs:1-maxnt:3000-input_output_len:2048,128-reqs:32-con:1] #passed
perf/test_perf.py::test_perf[llama_v3.3_nemotron_super_49b_fp8-bench-pytorch-float8-maxbs:1-maxnt:3000-input_output_len:2048,128-reqs:32-con:1] SKIP (https://nvbugspro.nvidia.com/bug/5387445)
perf/test_perf.py::test_perf[llama_v3.3_nemotron_super_49b-bench-pytorch-bfloat16-maxbs:1-maxnt:320-input_output_len:128,128-reqs:32-con:1] SKIP (https://nvbugspro.nvidia.com/bug/5387445)
perf/test_perf.py::test_perf[llama_v3.3_70b_instruct-bench-pytorch-float16-maxbs:1-maxnt:320-input_output_len:128,128-reqs:32-con:1] SKIP (https://nvbugspro.nvidia.com/bug/5387445)
perf/test_perf.py::test_perf[llama_v3.3_70b_instruct_fp8-bench-pytorch-float8-maxbs:1-maxnt:3000-input_output_len:2048,128-reqs:32-con:1] SKIP (https://nvbugspro.nvidia.com/bug/5387445)
perf/test_perf.py::test_perf[mixtral_8x7b_v0.1_instruct-bench-pytorch-float16-maxbs:1-maxnt:320-input_output_len:128,128-reqs:32-con:1]
perf/test_perf.py::test_perf[mixtral_8x7b_v0.1_instruct_fp8-bench-pytorch-float8-maxbs:1-maxnt:3000-input_output_len:2048,128-reqs:32-con:1]
perf/test_perf.py::test_perf[mistral_nemo_12b_base-bench-pytorch-float16-maxbs:1-maxnt:3000-input_output_len:2048,128-reqs:32-con:1]
perf/test_perf.py::test_perf[deepseek_r1_distill_qwen_32b-bench-pytorch-float16-maxbs:1-maxnt:320-input_output_len:128,128-reqs:32-con:1]
perf/test_perf.py::test_perf[llama_v3.3_70b_instruct_fp8-bench-pytorch-float8-maxbs:1-maxnt:320-input_output_len:128,128-reqs:32-con:1]
perf/test_perf.py::test_perf[mixtral_8x7b_v0.1_instruct_fp8-bench-pytorch-float8-maxbs:1-maxnt:320-input_output_len:128,128-reqs:32-con:1]
perf/test_perf.py::test_perf[mistral_nemo_12b_base-bench-pytorch-float16-maxbs:1-maxnt:320-input_output_len:128,128-reqs:32-con:1]
perf/test_perf.py::test_perf[llama_v3.1_8b_instruct_fp4-bench-pytorch-float4-maxbs:1-maxnt:3000-input_output_len:2048,128-reqs:32-con:1]
perf/test_perf.py::test_perf[mixtral_8x7b_v0.1_instruct_fp4-bench-pytorch-float4-maxbs:1-maxnt:3000-input_output_len:2048,128-reqs:32-con:1]
perf/test_perf.py::test_perf[llama_v3.3_70b_instruct_fp4-bench-pytorch-float4-maxbs:1-maxnt:3000-input_output_len:2048,128-reqs:32-con:1]
perf/test_perf.py::test_perf[llama_v3.1_8b_instruct_fp8-bench-pytorch-float8-input_output_len:512,32-kv_cache_dtype:fp8]
perf/test_perf.py::test_perf[llama_v3.1_8b_instruct-bench-pytorch-bfloat16-input_output_len:128,128]
perf/test_perf.py::test_perf[llama_v3.1_8b_instruct-bench-pytorch-bfloat16-input_output_len:512,128]
perf/test_perf.py::test_perf[llama_v3.1_8b_instruct-bench-pytorch-streaming-bfloat16-input_output_len:512,128]
perf/test_perf.py::test_perf[llama_v3.1_8b_instruct-bench-pytorch-streaming-bfloat16-input_output_len:128,128]
perf/test_perf.py::test_perf[llama_v3.1_8b_instruct_fp8-bench-pytorch-float8-input_output_len:128,128]
perf/test_perf.py::test_perf[llama_v3.1_8b_instruct_fp8-bench-pytorch-float8-input_output_len:512,128]
perf/test_perf.py::test_perf[llama_v3.1_8b_instruct_fp8-bench-pytorch-float8-input_output_len:128,2048]
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_8b_instruct_fp4-bench-pytorch-float4-input_output_len:512,128]
perf/test_perf.py::test_perf[llama_v3.1_8b_instruct_fp4-bench-pytorch-float4-input_output_len:128,2048]

perf/test_perf.py::test_perf[llama_v3.3_nemotron_super_49b-bench-pytorch-bfloat16-input_output_len:128,128]
perf/test_perf.py::test_perf[llama_v3.1_8b_instruct_fp8-bench-pytorch-float8-input_output_len:128,128-reqs:100-con:2]
perf/test_perf.py::test_perf[mistral_7b_v0.3-bench-pytorch-float16-input_output_len:128,128]
perf/test_perf.py::test_perf[mistral_7b_v0.3-bench-pytorch-float16-input_output_len:512,32]
perf/test_perf.py::test_perf[llama_v3.3_nemotron_super_49b_fp8-bench-pytorch-float8-input_output_len:128,128]
perf/test_perf.py::test_perf[llama_v3.3_nemotron_super_49b_fp8-bench-pytorch-float8-input_output_len:512,128]

perf/test_perf.py::test_perf[llama_v3.3_nemotron_super_49b-bench-pytorch-bfloat16-input_output_len:128,128]
perf/test_perf.py::test_perf[llama_v3.3_70b_instruct_fp8-bench-pytorch-float8-input_output_len:128,128]
perf/test_perf.py::test_perf[llama_v3.3_70b_instruct_fp8-bench-pytorch-float8-input_output_len:512,128]
perf/test_perf.py::test_perf[llama_v3.3_70b_instruct_fp4-bench-pytorch-float4-input_output_len:128,128]
perf/test_perf.py::test_perf[llama_v3.3_70b_instruct_fp4-bench-pytorch-float4-input_output_len:512,128]

perf/test_perf.py::test_perf[mixtral_8x7b_v0.1_instruct-bench-pytorch-float16-input_output_len:128,128]
perf/test_perf.py::test_perf[mixtral_8x7b_v0.1_instruct_fp8-bench-pytorch-float8-input_output_len:128,128]
perf/test_perf.py::test_perf[mixtral_8x7b_v0.1_instruct_fp4-bench-pytorch-float4-input_output_len:512,128]
perf/test_perf.py::test_perf[llama_v3.3_70b_instruct_fp8-bench-pytorch-streaming-float8-input_output_len:512,32]
perf/test_perf.py::test_perf[llama_v3.1_8b_instruct_fp4-bench-pytorch-float4-input_output_len:512,32-kv_cache_dtype:fp8]
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[mixtral_8x7b_v0.1_instruct_fp4-bench-pytorch-float4-input_output_len:128,128]
perf/test_perf.py::test_perf[mixtral_8x7b_v0.1_instruct_fp4-bench-pytorch-float4-input_output_len:512,128]

perf/test_perf.py::test_perf[mistral_nemo_12b_base-bench-pytorch-float16-input_output_len:128,128]
perf/test_perf.py::test_perf[deepseek_r1_distill_qwen_32b-bench-pytorch-float16-input_output_len:128,128]
perf/test_perf.py::test_perf[mixtral_8x7b_v0.1_instruct_fp4-bench-pytorch-float4-input_output_len:128,128-kv_cache_dtype:fp8]
perf/test_perf.py::test_perf[llama_v3.3_70b_instruct_fp4-bench-pytorch-float4-input_output_len:128,128]
perf/test_perf.py::test_perf[llama_v3.1_8b_instruct_fp4-bench-pytorch-float4-maxbs:1-maxnt:3000-input_output_len:2048,128-reqs:32-con:1]
perf/test_perf.py::test_perf[nvila_15b-bench-pytorch-bfloat16-input_output_len:128,128]