[Feature] OTEL tracing during loading (#31162)
This commit is contained in:
@@ -19,6 +19,7 @@ from vllm.model_executor.layers.fused_moe.triton_deep_gemm_moe import (
|
||||
)
|
||||
from vllm.model_executor.layers.linear import LinearBase
|
||||
from vllm.model_executor.layers.quantization.fp8 import Fp8LinearMethod
|
||||
from vllm.tracing import instrument
|
||||
from vllm.utils.deep_gemm import (
|
||||
fp8_gemm_nt,
|
||||
get_mk_alignment_for_contiguous_layout,
|
||||
@@ -358,6 +359,7 @@ def _count_warmup_iterations(model: torch.nn.Module, max_tokens: int) -> int:
|
||||
return total
|
||||
|
||||
|
||||
@instrument(span_name="DeepGemm warmup")
|
||||
def deep_gemm_warmup(model: torch.nn.Module, max_tokens: int):
|
||||
total = _count_warmup_iterations(model, max_tokens)
|
||||
if total == 0:
|
||||
|
||||
Reference in New Issue
Block a user