biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[Docs] Simplify CPU x86 Docker build documentation (#33071)

Browse Source 
Signed-off-by: Maryam Tahhan <mtahhan@redhat.com>
This commit is contained in:
Maryam Tahhan
2026-01-28 06:37:09 +00:00
committed by GitHub
parent 22ad649501
commit 2dd359f953
1 changed files with 9 additions and 23 deletions
Download Patch File Download Diff File Expand all files Collapse all files

32
docs/getting_started/installation/cpu.x86.inc.md
View File

@@ -171,10 +171,6 @@ uv pip install dist/*.whl
## Building for your target CPU
vLLM supports building Docker images for x86 CPU platforms with automatic instruction set detection.
### Basic build command
```bash
docker build -f docker/Dockerfile.cpu \
--build-arg VLLM_CPU_DISABLE_AVX512=<false (default)|true> \
@@ -187,29 +183,21 @@ docker build -f docker/Dockerfile.cpu \
--target vllm-openai .
```
!!! note "Instruction set auto-detection"
By default, vLLM will auto-detect CPU instruction sets (AVX512, AVX2, etc.) from the build system's CPU flags. Build arguments like `VLLM_CPU_AVX2`, `VLLM_CPU_AVX512`, `VLLM_CPU_AVX512BF16`, `VLLM_CPU_AVX512VNNI`, and `VLLM_CPU_AMXBF16` are primarily used for **cross-compilation** or for building container images on systems that don't have the target platforms ISA:
!!! note "Auto-detection by default"
By default, CPU instruction sets (AVX512, AVX2, etc.) are automatically detected from the build system's CPU flags. Build arguments like `VLLM_CPU_AVX2`, `VLLM_CPU_AVX512`, `VLLM_CPU_AVX512BF16`, `VLLM_CPU_AVX512VNNI`, and `VLLM_CPU_AMXBF16` are used for cross-compilation:
- Set `VLLM_CPU_{ISA}=true` to force-enable an instruction set (for cross-compilation to target platforms with that ISA)
- Set `VLLM_CPU_{ISA}=false` to rely on auto-detection
- When an ISA build arg is set to `true`, vLLM will build with that instruction set regardless of the build system's CPU capabilities
- `VLLM_CPU_{ISA}=true` - Force-enable the instruction set (build with ISA regardless of build system capabilities)
- `VLLM_CPU_{ISA}=false` - Rely on auto-detection (default)
### Build examples
### Examples
**Example 1: Auto-detection (native build)**
Build on a machine with the same CPU as your target deployment:
**Auto-detection build (default)**
```bash
# Auto-detects all CPU features from the build system
docker build -f docker/Dockerfile.cpu \
--tag vllm-cpu-env \
--target vllm-openai .
docker build -f docker/Dockerfile.cpu --tag vllm-cpu-env --target vllm-openai .
```
**Example 2: Cross-compilation for AVX512 deployment**
Build an AVX512 image on any x86_64 system (even without AVX512):
**Cross-compile for AVX512**
```bash
docker build -f docker/Dockerfile.cpu \
@@ -220,9 +208,7 @@ docker build -f docker/Dockerfile.cpu \
--target vllm-openai .
```
**Example 3: Cross-compilation for AVX2 deployment**
Build an AVX2 image for older CPUs:
**Cross-compile for AVX2**
```bash
docker build -f docker/Dockerfile.cpu \