Simplify!

Signed-off-by: mgoin <michael@neuralmagic.com>

.buildkite/release-pipeline.yaml

@@ -41,37 +41,17 @@ steps:
     env:
       DOCKER_BUILDKIT: "1"
 
-  - block: "Build FlashInfer wheels"
+  - block: "Build FlashInfer wheel"
-    key: block-build-flashinfer-wheels
+    key: block-build-flashinfer-wheel
     depends_on: ~
 
-  - label: "Build FlashInfer wheels - CUDA 12.8"
+  - label: "Build and upload FlashInfer wheel - CUDA 12.8"
-    depends_on: block-build-flashinfer-wheels
+    depends_on: block-build-flashinfer-wheel
-    id: build-flashinfer-wheel-cuda-12-8
+    id: build-upload-flashinfer-wheel
     agents:
       queue: cpu_queue_postmerge
     commands:
-      - "bash .buildkite/scripts/build-flashinfer-wheel.sh 12.8.1"
+      - "bash .buildkite/scripts/build-upload-flashinfer-wheel.sh 12.8.1"
-    env:
-      DOCKER_BUILDKIT: "1"
-
-  - label: "Build FlashInfer wheels - CUDA 12.6"
-    depends_on: block-build-flashinfer-wheels
-    id: build-flashinfer-wheel-cuda-12-6
-    agents:
-      queue: cpu_queue_postmerge
-    commands:
-      - "bash .buildkite/scripts/build-flashinfer-wheel.sh 12.6.3"
-    env:
-      DOCKER_BUILDKIT: "1"
-
-  - label: "Build FlashInfer wheels - CUDA 11.8"
-    depends_on: block-build-flashinfer-wheels
-    id: build-flashinfer-wheel-cuda-11-8
-    agents:
-      queue: cpu_queue_postmerge
-    commands:
-      - "bash .buildkite/scripts/build-flashinfer-wheel.sh 11.8.0"
     env:
       DOCKER_BUILDKIT: "1"
 

.buildkite/scripts/build-flashinfer-wheel.sh

@@ -1,30 +0,0 @@
-#!/usr/bin/env bash
-
-set -ex
-
-CUDA_VERSION="${1:-12.8.1}"
-FLASHINFER_VERSION="${FLASHINFER_VERSION:-v0.2.9rc2}"
-
-echo "Building FlashInfer wheel for CUDA ${CUDA_VERSION} using vLLM Dockerfile"
-
-# Build the FlashInfer wheel using the existing Dockerfile stage
-DOCKER_BUILDKIT=1 docker build \
-  --build-arg max_jobs=16 \
-  --build-arg USE_SCCACHE=1 \
-  --build-arg CUDA_VERSION="${CUDA_VERSION}" \
-  --build-arg FLASHINFER_GIT_REF="${FLASHINFER_VERSION}" \
-  --tag flashinfer-wheel-builder:${CUDA_VERSION} \
-  --target flashinfer-wheel-builder \
-  --progress plain \
-  -f docker/Dockerfile .
-
-# Extract the wheel
-mkdir -p artifacts/dist
-docker run --rm -v $(pwd)/artifacts:/output_host flashinfer-wheel-builder:${CUDA_VERSION} \
-  bash -c 'cp /output/*.whl /output_host/dist/ && chmod -R a+rw /output_host'
-
-# Upload the wheel
-bash .buildkite/scripts/upload-flashinfer-wheels.sh
-
-echo "FlashInfer wheel built and uploaded successfully for CUDA ${CUDA_VERSION}"
-ls -la artifacts/dist/

.buildkite/scripts/build-upload-flashinfer-wheel.sh

@@ -0,0 +1,58 @@
+#!/usr/bin/env bash
+
+set -ex
+
+CUDA_VERSION="${1:-12.8.1}"
+FLASHINFER_VERSION="${FLASHINFER_VERSION:-v0.2.9rc2}"
+
+echo "Building FlashInfer wheel for CUDA ${CUDA_VERSION} using vLLM Dockerfile"
+
+# Build the FlashInfer wheel using the existing Dockerfile stage
+DOCKER_BUILDKIT=1 docker build \
+  --build-arg max_jobs=16 \
+  --build-arg USE_SCCACHE=1 \
+  --build-arg CUDA_VERSION="${CUDA_VERSION}" \
+  --build-arg FLASHINFER_GIT_REF="${FLASHINFER_VERSION}" \
+  --tag flashinfer-wheel-builder:${CUDA_VERSION} \
+  --target flashinfer-wheel-builder \
+  --progress plain \
+  -f docker/Dockerfile .
+
+# Extract the wheel
+mkdir -p artifacts/dist
+docker run --rm -v $(pwd)/artifacts:/output_host flashinfer-wheel-builder:${CUDA_VERSION} \
+  bash -c 'cp /output/*.whl /output_host/dist/ && chmod -R a+rw /output_host'
+
+# Upload the wheel to S3
+echo "Uploading FlashInfer wheel to S3..."
+wheel_files=(artifacts/dist/*.whl)
+
+# Check that exactly one wheel is found
+if [[ ${#wheel_files[@]} -ne 1 ]]; then
+  echo "Error: Expected exactly one wheel file in artifacts/dist/, but found ${#wheel_files[@]}"
+  exit 1
+fi
+
+# Get the single wheel file
+wheel="${wheel_files[0]}"
+echo "Processing FlashInfer wheel: $wheel"
+
+# Rename 'linux' to 'manylinux1' in the wheel filename for compatibility
+new_wheel="${wheel/linux/manylinux1}"
+if [[ "$wheel" != "$new_wheel" ]]; then
+  mv -- "$wheel" "$new_wheel"
+  wheel="$new_wheel"
+  echo "Renamed wheel to: $wheel"
+fi
+
+# Extract the version from the wheel
+version=$(unzip -p "$wheel" '**/METADATA' | grep '^Version: ' | cut -d' ' -f2)
+wheel_name=$(basename "$wheel")
+echo "FlashInfer version: $version"
+
+# Upload the wheel to S3 under flashinfer directory
+aws s3 cp "$wheel" "s3://vllm-wheels/flashinfer/"
+
+echo "✅ FlashInfer wheel built and uploaded successfully for CUDA ${CUDA_VERSION}"
+echo "📦 Wheel: $wheel_name (version $version)"
+ls -la artifacts/dist/

.buildkite/scripts/upload-flashinfer-wheels.sh

@@ -1,65 +0,0 @@
-#!/usr/bin/env bash
-
-set -ex
-
-# Assume wheels are in artifacts/dist/*.whl
-wheel_files=(artifacts/dist/*.whl)
-
-# Check that exactly one wheel is found
-if [[ ${#wheel_files[@]} -ne 1 ]]; then
-  echo "Error: Expected exactly one wheel file in artifacts/dist/, but found ${#wheel_files[@]}"
-  exit 1
-fi
-
-# Get the single wheel file
-wheel="${wheel_files[0]}"
-
-echo "Processing FlashInfer wheel: $wheel"
-
-# Rename 'linux' to 'manylinux1' in the wheel filename for compatibility
-new_wheel="${wheel/linux/manylinux1}"
-if [[ "$wheel" != "$new_wheel" ]]; then
-  mv -- "$wheel" "$new_wheel"
-  wheel="$new_wheel"
-  echo "Renamed wheel to: $wheel"
-fi
-
-# Extract the version from the wheel
-version=$(unzip -p "$wheel" '**/METADATA' | grep '^Version: ' | cut -d' ' -f2)
-wheel_name=$(basename "$wheel")
-echo "FlashInfer version: $version"
-
-# Upload the wheel to S3 under flashinfer directory
-aws s3 cp "$wheel" "s3://vllm-wheels/flashinfer/"
-
-# Download existing index if it exists, then rebuild it with all wheels
-echo "Rebuilding index with all available wheels..."
-aws s3 ls s3://vllm-wheels/flashinfer/ --recursive | grep '\.whl$' | awk '{print $4}' | sed 's|flashinfer/||' > wheel_list.txt
-
-# Generate complete index.html for the package (following pip index pattern)
-cat > flashinfer_index.html << 'EOF'
-<!DOCTYPE html>
-<html>
-<head><title>Links for flashinfer-python</title></head>
-<body>
-<h1>Links for flashinfer-python</h1>
-EOF
-
-# Add each wheel to the index
-while IFS= read -r wheel_file; do
-    if [[ -n "$wheel_file" ]]; then
-        echo "<a href=\"$wheel_file\">$wheel_file</a><br/>" >> flashinfer_index.html
-    fi
-done < wheel_list.txt
-
-cat >> flashinfer_index.html << 'EOF'
-</body>
-</html>
-EOF
-
-aws s3 cp flashinfer_index.html "s3://vllm-wheels/flashinfer/index.html"
-
-# Clean up
-rm -f flashinfer_index.html wheel_list.txt
-
-echo "Successfully uploaded FlashInfer wheel $wheel_name (version $version)"

docker/Dockerfile

@@ -274,10 +274,10 @@ ARG CUDA_VERSION
 ARG FLASHINFER_GIT_REPO="https://github.com/flashinfer-ai/flashinfer.git"
 ARG FLASHINFER_GIT_REF="v0.2.9rc2"
 
-COPY tools/build-flashinfer.sh /tmp/build-flashinfer.sh
+COPY tools/flashinfer-build.sh /tmp/flashinfer-build.sh
 RUN --mount=type=cache,target=/root/.cache/uv \
     . /etc/environment && \
-    BUILD_WHEEL=true /tmp/build-flashinfer.sh
+    BUILD_WHEEL=true /tmp/flashinfer-build.sh
 
 #################### EXTENSION Build IMAGE ####################
 
@@ -406,10 +406,10 @@ ARG FLASHINFER_GIT_REPO="https://github.com/flashinfer-ai/flashinfer.git"
 # Keep this in sync with https://github.com/vllm-project/vllm/blob/main/requirements/cuda.txt
 # We use `--force-reinstall --no-deps` to avoid issues with the existing FlashInfer wheel.
 ARG FLASHINFER_GIT_REF="v0.2.9rc2"
-COPY tools/build-flashinfer.sh /tmp/build-flashinfer.sh
+COPY tools/flashinfer-build.sh /tmp/flashinfer-build.sh
 RUN --mount=type=cache,target=/root/.cache/uv \
     . /etc/environment && \
-    /tmp/build-flashinfer.sh
+    /tmp/flashinfer-build.sh
 COPY examples examples
 COPY benchmarks benchmarks
 COPY ./vllm/collect_env.py .