Skip to content

opti profiler default param #4609

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

Open
wants to merge 1 commit into
base: main
Choose a base branch
from
+28 −1
Open

opti profiler default param #4609

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
14 changes: 14 additions & 0 deletions vllm_ascend/envs.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -176,6 +176,20 @@
# Whether to anbale dynamic EPLB
"DYNAMIC_EPLB":
lambda: os.getenv("DYNAMIC_EPLB", "false").lower(),
# Set torch_npu profiler to profile aicore metrics. There are the following options that can be configured:
# 0: torch_npu.profiler.AiCMetrics.AiCoreNone;
# 1: torch_npu.profiler.AiCMetrics.PipeUtilization;
# 2: torch_npu.profiler.AiCMetrics.ArithmeticUtilization;
# 3: torch_npu.profiler.AiCMetrics.Memory;
# 4: torch_npu.profiler.AiCMetrics.MemoryL0;
# 5: torch_npu.profiler.AiCMetrics.ResourceConflictRatio;
# 6: torch_npu.profiler.AiCMetrics.MemoryUB;
# 7: torch_npu.profiler.AiCMetrics.L2Cache;
# 8: torch_npu.profiler.AiCMetrics.MemoryAccess;
# If not set, it will be torch_npu.profiler.AiCMetrics.PipeUtilization by default.
# The meanings of various options can refer to: https://www.hiascend.com/document/detail/zh/Pytorch/720/apiref/torchnpuCustomsapi/context/torch_npu-profiler-AiCMetrics.md
"VLLM_ASCEND_PROFILER_AIC_METRICS":
lambda: int(os.getenv("VLLM_ASCEND_PROFILER_AIC_METRICS", 1)),
}

# end-env-vars-definition
Expand Down
15 changes: 14 additions & 1 deletion vllm_ascend/worker/worker_v1.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -442,11 +442,24 @@ def _init_profiler(self):
logger.info("Profiling enabled. Traces will be saved to: %s",
torch_profiler_trace_dir)

aic_metrics_list = [
torch_npu.profiler.AiCMetrics.AiCoreNone,
torch_npu.profiler.AiCMetrics.PipeUtilization,
torch_npu.profiler.AiCMetrics.ArithmeticUtilization,
torch_npu.profiler.AiCMetrics.Memory,
torch_npu.profiler.AiCMetrics.MemoryL0,
torch_npu.profiler.AiCMetrics.ResourceConflictRatio,
torch_npu.profiler.AiCMetrics.MemoryUB,
torch_npu.profiler.AiCMetrics.L2Cache,
torch_npu.profiler.AiCMetrics.MemoryAccess
]

experimental_config = torch_npu.profiler._ExperimentalConfig(
export_type=torch_npu.profiler.ExportType.Text,
profiler_level=torch_npu.profiler.ProfilerLevel.Level1,
msprof_tx=False,
aic_metrics=torch_npu.profiler.AiCMetrics.AiCoreNone,
aic_metrics=aic_metrics_list[
envs_ascend.VLLM_ASCEND_PROFILER_AIC_METRICS],
l2_cache=False,
op_attr=False,
data_simplification=False,
Expand Down
Loading