Skip to content

test DeepGEMM Serialize concurrent JIT compilation per kernel #1132

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
Gregory-Pereira wants to merge 2 commits into from
Closed

test DeepGEMM Serialize concurrent JIT compilation per kernel #1132

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
4de2f73
building vLLM with stated components
Gregory-Pereira Apr 11, 2026
a4ee37a
add cuobjdump for core request path of deepseek
Gregory-Pereira Apr 11, 2026
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
7 changes: 4 additions & 3 deletions docker/Dockerfile.cuda
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -57,11 +57,12 @@
# branch: https://github.com/neuralmagic/DeepEP/tree/llm-d-release-v0.5.1, but frozen at commit:
ARG DEEPEP_VERSION="38d21b7f9bb6f3b102b1819d09439686eaa87ce8"

ARG DEEPGEMM_REPO="https://github.com/deepseek-ai/DeepGEMM"
ARG DEEPGEMM_VERSION="v2.1.1.post3"
# My branch on top of reported issue version: 477618cd51baffca09c4b0b87e97c03fe827ef03
ARG DEEPGEMM_REPO="https://github.com/Gregory-Pereira/DeepGEMM"
ARG DEEPGEMM_VERSION="8b58b013f019af77edf238e8d6e0ee32f9cbe9cd"

ARG FLASHINFER_REPO="https://github.com/flashinfer-ai/flashinfer.git"
ARG FLASHINFER_VERSION="v0.6.4"
ARG FLASHINFER_VERSION="v0.6.7"

# vLLM build settings - provided via build args from docker/common-versions file
# Override here for platform-specific testing or via build args
Expand Down Expand Up @@ -106,7 +107,7 @@
COPY docker/constraints.txt /tmp/constraints.txt
COPY patches/ /tmp/patches/

ENV LANG="C.UTF-8" \

Check warning on line 110 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d (linux/arm64, vllm-runner-arm, ubuntu, ubuntu24.04, -ubuntu)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$UV_TORCH_BACKEND'
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 110 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d (linux/arm64, vllm-runner-arm, ubuntu, ubuntu24.04, -ubuntu)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$UV_TORCH_BACKEND'
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 110 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d (linux/arm64, vllm-runner-arm, rhel, ubi9)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$UV_TORCH_BACKEND'
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 110 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d (linux/arm64, vllm-runner-arm, rhel, ubi9)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$UV_TORCH_BACKEND'
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 110 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d (linux/amd64, vllm-runner, ubuntu, ubuntu24.04, -ubuntu)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$UV_TORCH_BACKEND'
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 110 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d (linux/amd64, vllm-runner, ubuntu, ubuntu24.04, -ubuntu)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$UV_TORCH_BACKEND'
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 110 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d (linux/amd64, vllm-runner, rhel, ubi9)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$UV_TORCH_BACKEND'
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 110 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d (linux/amd64, vllm-runner, rhel, ubi9)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$UV_TORCH_BACKEND'
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 110 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / aws-build-llm-d (linux/amd64, vllm-runner, rhel, ubi9)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$UV_TORCH_BACKEND'
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 110 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / aws-build-llm-d (linux/amd64, vllm-runner, rhel, ubi9)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$UV_TORCH_BACKEND'
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 110 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d-debug (linux/amd64, vllm-runner, rhel, ubi9, -debug)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$UV_TORCH_BACKEND'
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/
LC_ALL="C.UTF-8" \
UV_HTTP_TIMEOUT="500" \
UV_INDEX_STRATEGY="unsafe-best-match" \
Expand Down Expand Up @@ -163,7 +164,7 @@
NIXL_PREFIX="/opt/nixl" \
EFA_PREFIX="/opt/amazon/efa"

ENV PATH="${NVSHMEM_DIR}/bin:${VIRTUAL_ENV}/bin:${PATH}" \

Check warning on line 167 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d (linux/arm64, vllm-runner-arm, ubuntu, ubuntu24.04, -ubuntu)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$PKG_CONFIG_PATH'
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 167 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d (linux/arm64, vllm-runner-arm, ubuntu, ubuntu24.04, -ubuntu)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$CPATH' (did you mean $PATH?)
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 167 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d (linux/arm64, vllm-runner-arm, ubuntu, ubuntu24.04, -ubuntu)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$PKG_CONFIG_PATH'
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 167 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d (linux/arm64, vllm-runner-arm, ubuntu, ubuntu24.04, -ubuntu)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$CPATH' (did you mean $PATH?)
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 167 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d (linux/arm64, vllm-runner-arm, rhel, ubi9)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$PKG_CONFIG_PATH'
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 167 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d (linux/arm64, vllm-runner-arm, rhel, ubi9)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$CPATH' (did you mean $PATH?)
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 167 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d (linux/arm64, vllm-runner-arm, rhel, ubi9)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$CPATH' (did you mean $PATH?)
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 167 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d (linux/arm64, vllm-runner-arm, rhel, ubi9)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$PKG_CONFIG_PATH'
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 167 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d (linux/amd64, vllm-runner, ubuntu, ubuntu24.04, -ubuntu)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$PKG_CONFIG_PATH'
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 167 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d (linux/amd64, vllm-runner, ubuntu, ubuntu24.04, -ubuntu)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$CPATH' (did you mean $PATH?)
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 167 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d (linux/amd64, vllm-runner, ubuntu, ubuntu24.04, -ubuntu)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$CPATH' (did you mean $PATH?)
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 167 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d (linux/amd64, vllm-runner, ubuntu, ubuntu24.04, -ubuntu)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$PKG_CONFIG_PATH'
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 167 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d (linux/amd64, vllm-runner, rhel, ubi9)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$PKG_CONFIG_PATH'
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 167 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d (linux/amd64, vllm-runner, rhel, ubi9)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$CPATH' (did you mean $PATH?)
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 167 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d (linux/amd64, vllm-runner, rhel, ubi9)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$PKG_CONFIG_PATH'
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 167 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d (linux/amd64, vllm-runner, rhel, ubi9)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$CPATH' (did you mean $PATH?)
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 167 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / aws-build-llm-d (linux/amd64, vllm-runner, rhel, ubi9)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$PKG_CONFIG_PATH'
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 167 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / aws-build-llm-d (linux/amd64, vllm-runner, rhel, ubi9)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$CPATH' (did you mean $PATH?)
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 167 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / aws-build-llm-d (linux/amd64, vllm-runner, rhel, ubi9)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$PKG_CONFIG_PATH'
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 167 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / aws-build-llm-d (linux/amd64, vllm-runner, rhel, ubi9)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$CPATH' (did you mean $PATH?)
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 167 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d-debug (linux/amd64, vllm-runner, rhel, ubi9, -debug)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$PKG_CONFIG_PATH'
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 167 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d-debug (linux/amd64, vllm-runner, rhel, ubi9, -debug)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$CPATH' (did you mean $PATH?)
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/
LIBRARY_PATH="\
${EFA_PREFIX}/lib:${EFA_PREFIX}/lib64:\
${UCX_PREFIX}/lib:${UCX_PREFIX}/lib64:\
Expand Down Expand Up @@ -311,7 +312,7 @@

COPY docker/constraints.txt /tmp/constraints.txt

ENV LANG="C.UTF-8" \

Check warning on line 315 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d (linux/arm64, vllm-runner-arm, ubuntu, ubuntu24.04, -ubuntu)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$UV_TORCH_BACKEND'
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 315 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d (linux/arm64, vllm-runner-arm, rhel, ubi9)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$UV_TORCH_BACKEND'
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 315 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d (linux/amd64, vllm-runner, ubuntu, ubuntu24.04, -ubuntu)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$UV_TORCH_BACKEND'
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 315 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d (linux/amd64, vllm-runner, rhel, ubi9)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$UV_TORCH_BACKEND'
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 315 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / aws-build-llm-d (linux/amd64, vllm-runner, rhel, ubi9)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$UV_TORCH_BACKEND'
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 315 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d-debug (linux/amd64, vllm-runner, rhel, ubi9, -debug)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$UV_TORCH_BACKEND'
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/
LC_ALL="C.UTF-8" \
PYTHON_VERSION="${PYTHON_VERSION}" \
UV_TORCH_BACKEND="${UV_TORCH_BACKEND:-cu${CUDA_MAJOR}${CUDA_MINOR}}" \
Expand All @@ -322,7 +323,7 @@
EFA_PREFIX="/opt/amazon/efa" \
CUDA_HOME="/usr/local/cuda"

ENV LD_LIBRARY_PATH="\

Check warning on line 326 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d (linux/arm64, vllm-runner-arm, ubuntu, ubuntu24.04, -ubuntu)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$CPATH' (did you mean $PATH?)
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 326 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d (linux/arm64, vllm-runner-arm, rhel, ubi9)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$CPATH' (did you mean $PATH?)
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 326 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d (linux/amd64, vllm-runner, ubuntu, ubuntu24.04, -ubuntu)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$CPATH' (did you mean $PATH?)
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 326 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d (linux/amd64, vllm-runner, rhel, ubi9)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$CPATH' (did you mean $PATH?)
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 326 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / aws-build-llm-d (linux/amd64, vllm-runner, rhel, ubi9)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$CPATH' (did you mean $PATH?)
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/

Check warning on line 326 in docker/Dockerfile.cuda

View workflow job for this annotation

GitHub Actions / build-images / cuda-build-llm-d-debug (linux/amd64, vllm-runner, rhel, ubi9, -debug)

Variables should be defined before their use 

UndefinedVar: Usage of undefined variable '$CPATH' (did you mean $PATH?)
More info: https://docs.docker.com/go/dockerfile/rule/undefined-var/
/opt/vllm/lib64/python${PYTHON_VERSION}/site-packages/torch/lib:\
/usr/local/nvidia/lib:/usr/local/nvidia/lib64:\
${CUDA_HOME}/lib64:\
Expand Down
4 changes: 2 additions & 2 deletions docker/common-versions
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,7 +2,7 @@
# shared vLLM Version Configuration
# ============================================================================
VLLM_REPO=https://github.com/vllm-project/vllm.git
VLLM_COMMIT_SHA=95c0f928cdeeaa21c4906e73cee6a156e1b3b995 # maps to v0.17.1 tag
VLLM_COMMIT_SHA=780ba37458362bdc0596c6511e17749d44b145fc # Maps to 0.19.1rc1.dev44+g780ba3745 (git sha: 780ba37), see: https://github.com/vllm-project/vllm/issues/39057

# ============================================================================
# Cuda specific configs
Expand All @@ -15,7 +15,7 @@ VLLM_PREBUILT=0 # 0=editable install, 1=use full prebuilt wheel
VLLM_USE_PRECOMPILED=1 # 1=use precompiled binaries either from VLLM_COMMIT_SHA or VLLM_PRECOMPILED_WHEEL_COMMIT, 0=compile from source
# If building vLLM from a precompiled commit in the vLLM wheel index (only commits off main),
# which commit to use as the base for the copiled bits.
VLLM_PRECOMPILED_WHEEL_COMMIT=95c0f928cdeeaa21c4906e73cee6a156e1b3b995
VLLM_PRECOMPILED_WHEEL_COMMIT=780ba37458362bdc0596c6511e17749d44b145fc # Maps to 0.19.1rc1.dev44+g780ba3745 (git sha: 780ba37), see: https://github.com/vllm-project/vllm/issues/39057

### Cuda Runtime versions
CUDA_MAJOR=12
Expand Down
1 change: 1 addition & 0 deletions docker/packages/cuda/runtime-packages.json
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3,6 +3,7 @@
"cuda-nvcc-${CUDA_MAJOR}-${CUDA_MINOR}": "cuda-nvcc-${CUDA_MAJOR}-${CUDA_MINOR}",
"cuda-nvrtc-${CUDA_MAJOR}-${CUDA_MINOR}": "cuda-nvrtc-${CUDA_MAJOR}-${CUDA_MINOR}",
"cuda-cudart-devel-${CUDA_MAJOR}-${CUDA_MINOR}": "cuda-cudart-dev-${CUDA_MAJOR}-${CUDA_MINOR}",
"cuda-cuobjdump-${CUDA_MAJOR}-${CUDA_MINOR}": "cuda-cuobjdump-${CUDA_MAJOR}-${CUDA_MINOR}",
"python${PYTHON_VERSION}": "python${PYTHON_VERSION}",
"python${PYTHON_VERSION}-pip": "python${PYTHON_VERSION}-venv",
"python${PYTHON_VERSION}-devel": "python${PYTHON_VERSION}-dev",
Expand Down
Loading