Skip to content

🐛 use eager compile #488

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

Merged
merged 1 commit into from
Sep 27, 2025
Merged

🐛 use eager compile #488

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
10 changes: 9 additions & 1 deletion vllm_spyre/envs.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -23,6 +23,7 @@
VLLM_SPYRE_WORKER_LOG_REDIRECT_DIR: str = ""
VLLM_SPYRE_GLOO_TIMEOUT_MINUTES: int = 60
VLLM_SPYRE_REQUIRE_PRECOMPILED_DECODERS: bool = False
VLLM_SPYRE_SIMPLE_COMPILE_BACKEND: str = "eager"

logger = init_logger(__name__)

Expand Down Expand Up @@ -143,7 +144,14 @@ def _backend_backwards_compat() -> str:
# disable compilation for decoders
"VLLM_SPYRE_REQUIRE_PRECOMPILED_DECODERS":
lambda: bool(int(os.getenv("VLLM_SPYRE_REQUIRE_PRECOMPILED_DECODERS", "0"))
)
),

# Simple compile backend for some dynamically compiled operations, like
# gathering logprobs in the sampler.
# Defaults to eager, iductor can be used if python headers and a compiler
# are available.
"VLLM_SPYRE_SIMPLE_COMPILE_BACKEND":
lambda: os.getenv("VLLM_SPYRE_SIMPLE_COMPILE_BACKEND", "eager"),
}
# --8<-- [end:env-vars-definition]

Expand Down
4 changes: 2 additions & 2 deletions vllm_spyre/platform.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -65,9 +65,9 @@ class SpyrePlatform(Platform):
_num_spyre_blocks_override: int = -1 # override num of KV cache blocks
_config: VllmConfig = None

# TODO: see if this needs to be set
# Backend for dynamic compilation ops
# See vllm batched_count_greater_than method
# simple_compile_backend: str = "eager"
simple_compile_backend: str = envs_spyre.VLLM_SPYRE_SIMPLE_COMPILE_BACKEND

# Needed by vllm/model_executor/layers/pooler.py:562
current_stream = lambda _: _StreamPlaceholder()
Expand Down