Skip to content
GitLab
Menu
Projects
Groups
Snippets
Loading...
Help
Help
Support
Community forum
Keyboard shortcuts
?
Submit feedback
Contribute to GitLab
Sign in
Toggle navigation
Menu
Open sidebar
OpenDAS
vllm_cscc
Commits
79899b63
Unverified
Commit
79899b63
authored
Aug 16, 2025
by
nvjullin
Committed by
GitHub
Aug 15, 2025
Browse files
[Bugfix] Added more env vars to hash (#22449)
Signed-off-by:
Julien Lin
<
jullin@nvidia.com
>
parent
6e670778
Changes
1
Hide whitespace changes
Inline
Side-by-side
Showing
1 changed file
with
36 additions
and
10 deletions
+36
-10
vllm/envs.py
vllm/envs.py
+36
-10
No files found.
vllm/envs.py
View file @
79899b63
...
@@ -1199,14 +1199,6 @@ def compute_hash() -> str:
...
@@ -1199,14 +1199,6 @@ def compute_hash() -> str:
affect the choice of different kernels or attention backends should
affect the choice of different kernels or attention backends should
also be included in the factors list.
also be included in the factors list.
"""
"""
factors
:
list
[
Any
]
=
[]
# summarize environment variables
def
factorize
(
name
:
str
):
if
__getattr__
(
name
):
factors
.
append
(
__getattr__
(
name
))
else
:
factors
.
append
(
"None"
)
# The values of envs may affects the computation graph.
# The values of envs may affects the computation graph.
# TODO(DefTruth): hash all environment variables?
# TODO(DefTruth): hash all environment variables?
...
@@ -1221,11 +1213,45 @@ def compute_hash() -> str:
...
@@ -1221,11 +1213,45 @@ def compute_hash() -> str:
"VLLM_DP_SIZE"
,
"VLLM_DP_SIZE"
,
"VLLM_USE_STANDALONE_COMPILE"
,
"VLLM_USE_STANDALONE_COMPILE"
,
"VLLM_FUSED_MOE_CHUNK_SIZE"
,
"VLLM_FUSED_MOE_CHUNK_SIZE"
,
"VLLM_FLASHINFER_MOE_BACKEND"
,
"VLLM_V1_USE_PREFILL_DECODE_ATTENTION"
,
"VLLM_USE_AITER_UNIFIED_ATTENTION"
,
"VLLM_ATTENTION_BACKEND"
,
"VLLM_USE_FLASHINFER_SAMPLER"
,
"VLLM_FLASHINFER_FORCE_TENSOR_CORES"
,
"VLLM_DISABLED_KERNELS"
,
"VLLM_USE_DEEP_GEMM"
,
"VLLM_USE_TRTLLM_FP4_GEMM"
,
"VLLM_USE_TRTLLM_FP4_GEMM"
,
"VLLM_USE_FLASHINFER_MOE_FP8"
,
"VLLM_USE_FLASHINFER_MOE_FP4"
,
"VLLM_USE_FLASHINFER_MOE_MXFP4_MXFP8"
,
"VLLM_USE_FLASHINFER_MOE_MXFP4_BF16"
,
"VLLM_USE_CUDNN_PREFILL"
,
"VLLM_USE_TRTLLM_ATTENTION"
,
"VLLM_ROCM_USE_AITER"
,
"VLLM_ROCM_USE_AITER_PAGED_ATTN"
,
"VLLM_ROCM_USE_AITER_LINEAR"
,
"VLLM_ROCM_USE_AITER_MOE"
,
"VLLM_ROCM_USE_AITER_RMSNORM"
,
"VLLM_ROCM_USE_AITER_MLA"
,
"VLLM_ROCM_USE_AITER_MHA"
,
"VLLM_ROCM_USE_SKINNY_GEMM"
,
"VLLM_ROCM_FP8_PADDING"
,
"VLLM_ROCM_MOE_PADDING"
,
"VLLM_ROCM_CUSTOM_PAGED_ATTN"
,
"VLLM_ROCM_QUICK_REDUCE_QUANTIZATION"
,
"VLLM_ROCM_QUICK_REDUCE_CAST_BF16_TO_FP16"
,
"VLLM_ROCM_QUICK_REDUCE_MAX_SIZE_BYTES_MB"
,
]
]
for
key
in
environment_variables_to_hash
:
for
key
in
environment_variables_to_hash
:
if
key
in
environment_variables
:
# if this goes out of sync with environment_variables,
factorize
(
key
)
# it's not a user error, it's a bug
assert
key
in
environment_variables
,
\
"Please update environment_variables_to_hash in envs.py"
factors
=
[
environment_variables
[
key
]()
for
key
in
environment_variables_to_hash
]
hash_str
=
hashlib
.
md5
(
str
(
factors
).
encode
(),
hash_str
=
hashlib
.
md5
(
str
(
factors
).
encode
(),
usedforsecurity
=
False
).
hexdigest
()
usedforsecurity
=
False
).
hexdigest
()
...
...
Write
Preview
Markdown
is supported
0%
Try again
or
attach a new file
.
Attach a file
Cancel
You are about to add
0
people
to the discussion. Proceed with caution.
Finish editing this message first!
Cancel
Please
register
or
sign in
to comment