server-202510130816-ffce306

feat(bodhiapp): add GPU-accelerated base images infrastructure and re…

server-202510122107-28b6645

[Amir] added help and target to build the llama-server to check for v…

Base Images v202510130500-ffce306

llama.cpp Base Images Release v202510130500-ffce306

Released on: ****
Git commit: ffce306

Available Images

All images include embedded version metadata and are ready for BodhiApp integration.

Build Results

✅ CPU Runtime (Platform: linux/amd64, linux/arm64)

docker pull ghcr.io/bodhisearch/llama.cpp:cpu-202510130500-ffce306

✅ CUDA Runtime (Platform: linux/amd64) - Includes CPU fallback

docker pull ghcr.io/bodhisearch/llama.cpp:cuda-202510130500-ffce306

✅ ROCm Runtime (Platform: linux/amd64)

docker pull ghcr.io/bodhisearch/llama.cpp:rocm-202510130500-ffce306

✅ Vulkan Runtime (Platform: linux/amd64)

docker pull ghcr.io/bodhisearch/llama.cpp:vulkan-202510130500-ffce306

✅ MUSA Runtime (Platform: linux/amd64) - Moore Threads GPU

docker pull ghcr.io/bodhisearch/llama.cpp:musa-202510130500-ffce306

✅ Intel Runtime (Platform: linux/amd64) - Intel GPU with SYCL

docker pull ghcr.io/bodhisearch/llama.cpp:intel-202510130500-ffce306

✅ CANN Runtime (Platform: linux/amd64, linux/arm64) - Huawei Ascend NPU

docker pull ghcr.io/bodhisearch/llama.cpp:cann-202510130500-ffce306

Usage with BodhiApp

# Use as base image in BodhiApp
ARG BASE_VARIANT=cpu
FROM ghcr.io/bodhisearch/llama.cpp:${BASE_VARIANT}-202510130500-ffce306 AS runtime-base

# Your BodhiApp build continues...
# llama-server binary available at /app/bin/llama-server

Version Information

Each image contains version metadata accessible at runtime:

# Via Docker labels
docker inspect ghcr.io/bodhisearch/llama.cpp:cpu-202510130500-ffce306

# Via version file in container
docker run --rm ghcr.io/bodhisearch/llama.cpp:cpu-202510130500-ffce306 cat /app/version.json

Changelog

• Simplified Architecture: Focus on llama-server binary only
• Embedded Metadata: Version, commit, timestamp info in images
• Multi-platform Support: ARM64 support for CPU and Vulkan variants
• BodhiApp Integration: Clean inheritance pattern with predictable paths
• Tag-based Versioning: Timestamp-based versions for chronological ordering

Base Images v202510122231-a8daf2c

llama.cpp Base Images Release v202510122231-a8daf2c

Released on: ****
Git commit: a8daf2c

Available Images

All images include embedded version metadata and are ready for BodhiApp integration.

Build Results

❌ CPU Runtime - Build failed

❌ CUDA Runtime - Build failed

❌ ROCm Runtime - Build failed

✅ Vulkan Runtime (Platform: linux/amd64)

docker pull ghcr.io/bodhisearch/llama.cpp:vulkan-202510122231-a8daf2c

❌ MUSA Runtime - Build failed

❌ Intel Runtime - Build failed

✅ CANN Runtime (Platform: linux/amd64, linux/arm64) - Huawei Ascend NPU

docker pull ghcr.io/bodhisearch/llama.cpp:cann-202510122231-a8daf2c

Usage with BodhiApp

# Use as base image in BodhiApp
ARG BASE_VARIANT=cpu
FROM ghcr.io/bodhisearch/llama.cpp:${BASE_VARIANT}-202510122231-a8daf2c AS runtime-base

# Your BodhiApp build continues...
# llama-server binary available at /app/bin/llama-server

Version Information

Each image contains version metadata accessible at runtime:

# Via Docker labels
docker inspect ghcr.io/bodhisearch/llama.cpp:cpu-202510122231-a8daf2c

# Via version file in container
docker run --rm ghcr.io/bodhisearch/llama.cpp:cpu-202510122231-a8daf2c cat /app/version.json

Changelog

• Simplified Architecture: Focus on llama-server binary only
• Embedded Metadata: Version, commit, timestamp info in images
• Multi-platform Support: ARM64 support for CPU and Vulkan variants
• BodhiApp Integration: Clean inheritance pattern with predictable paths
• Tag-based Versioning: Timestamp-based versions for chronological ordering

Base Images v202510122137-c8c3671

llama.cpp Base Images Release v202510122137-c8c3671

Released on: ****
Git commit: c8c3671

Available Images

All images include embedded version metadata and are ready for BodhiApp integration.

Build Results

✅ CPU Runtime (Platform: linux/amd64, linux/arm64)

docker pull ghcr.io/bodhisearch/llama.cpp:cpu-202510122137-c8c3671

✅ CUDA Runtime (Platform: linux/amd64) - Includes CPU fallback

docker pull ghcr.io/bodhisearch/llama.cpp:cuda-202510122137-c8c3671

✅ ROCm Runtime (Platform: linux/amd64)

docker pull ghcr.io/bodhisearch/llama.cpp:rocm-202510122137-c8c3671

❌ Vulkan Runtime - Build failed

✅ MUSA Runtime (Platform: linux/amd64) - Moore Threads GPU

docker pull ghcr.io/bodhisearch/llama.cpp:musa-202510122137-c8c3671

✅ Intel Runtime (Platform: linux/amd64) - Intel GPU with SYCL

docker pull ghcr.io/bodhisearch/llama.cpp:intel-202510122137-c8c3671

❌ CANN Runtime - Build failed

Usage with BodhiApp

# Use as base image in BodhiApp
ARG BASE_VARIANT=cpu
FROM ghcr.io/bodhisearch/llama.cpp:${BASE_VARIANT}-202510122137-c8c3671 AS runtime-base

# Your BodhiApp build continues...
# llama-server binary available at /app/bin/llama-server

Version Information

Each image contains version metadata accessible at runtime:

# Via Docker labels
docker inspect ghcr.io/bodhisearch/llama.cpp:cpu-202510122137-c8c3671

# Via version file in container
docker run --rm ghcr.io/bodhisearch/llama.cpp:cpu-202510122137-c8c3671 cat /app/version.json

Changelog

• Simplified Architecture: Focus on llama-server binary only
• Embedded Metadata: Version, commit, timestamp info in images
• Multi-platform Support: ARM64 support for CPU and Vulkan variants
• BodhiApp Integration: Clean inheritance pattern with predictable paths
• Tag-based Versioning: Timestamp-based versions for chronological ordering

server-2509231057-7e2749d

[Amir] Enhance BodhiApp release management with new Makefile and docu…

server-202509231239-08fee9c

[Amir] Simplify BodhiApp release management with reduced complexity a…

server-20250923

[Amir] Update Makefile and add BodhiApp release management files

- Commented out deprecated Makefile build error message and default target for clarity.
- Introduced Makefile.bodhiapp for managing BodhiApp-specific release processes, including commands for binary and Docker image releases.
- Added README-bodhiapp.md to document the BodhiApp release management workflow and versioning strategy.
- Updated GitHub Actions workflow for llama-server to support version extraction from tags and manual triggers.

Base Images v202509231556-54d4354

llama.cpp Base Images Release v202509231556-54d4354

Released on: ****
Git commit: 54d4354

Available Images

All images include embedded version metadata and are ready for BodhiApp integration.

Build Results

✅ CPU Runtime (Platform: linux/amd64)

docker pull ghcr.io/bodhisearch/llama.cpp:cpu-202509231556-54d4354

✅ CUDA Runtime (Platform: linux/amd64) - Includes CPU fallback

docker pull ghcr.io/bodhisearch/llama.cpp:cuda-202509231556-54d4354

✅ ROCm Runtime (Platform: linux/amd64)

docker pull ghcr.io/bodhisearch/llama.cpp:rocm-202509231556-54d4354

✅ Vulkan Runtime (Platform: linux/amd64)

docker pull ghcr.io/bodhisearch/llama.cpp:vulkan-202509231556-54d4354

Usage with BodhiApp

# Use as base image in BodhiApp
ARG BASE_VARIANT=cpu
FROM ghcr.io/bodhisearch/llama.cpp:${BASE_VARIANT}-202509231556-54d4354 AS runtime-base

# Your BodhiApp build continues...
# llama-server binary available at /app/bin/llama-server

Version Information

Each image contains version metadata accessible at runtime:

# Via Docker labels
docker inspect ghcr.io/bodhisearch/llama.cpp:cpu-202509231556-54d4354

# Via version file in container
docker run --rm ghcr.io/bodhisearch/llama.cpp:cpu-202509231556-54d4354 cat /app/version.json

Changelog

• Simplified Architecture: Focus on llama-server binary only
• Embedded Metadata: Version, commit, timestamp info in images
• Multi-platform Support: ARM64 support for CPU and Vulkan variants
• BodhiApp Integration: Clean inheritance pattern with predictable paths
• Tag-based Versioning: Timestamp-based versions for chronological ordering

Base Images v202509231240-08fee9c

llama.cpp Base Images Release v202509231240-08fee9c

Released on: ****
Git commit: 08fee9c

Available Images

All images include embedded version metadata and are ready for BodhiApp integration.

Build Results

✅ CPU Runtime (Platform: linux/amd64)

docker pull ghcr.io/bodhisearch/llama.cpp:cpu-202509231240-08fee9c

✅ CUDA Runtime (Platform: linux/amd64) - Includes CPU fallback

docker pull ghcr.io/bodhisearch/llama.cpp:cuda-202509231240-08fee9c

✅ ROCm Runtime (Platform: linux/amd64)

docker pull ghcr.io/bodhisearch/llama.cpp:rocm-202509231240-08fee9c

❌ Vulkan Runtime - Build failed

Usage with BodhiApp

# Use as base image in BodhiApp
ARG BASE_VARIANT=cpu
FROM ghcr.io/bodhisearch/llama.cpp:${BASE_VARIANT}-202509231240-08fee9c AS runtime-base

# Your BodhiApp build continues...
# llama-server binary available at /app/bin/llama-server

Version Information

Each image contains version metadata accessible at runtime:

# Via Docker labels
docker inspect ghcr.io/bodhisearch/llama.cpp:cpu-202509231240-08fee9c

# Via version file in container
docker run --rm ghcr.io/bodhisearch/llama.cpp:cpu-202509231240-08fee9c cat /app/version.json

Changelog

• Simplified Architecture: Focus on llama-server binary only
• Embedded Metadata: Version, commit, timestamp info in images
• Multi-platform Support: ARM64 support for CPU and Vulkan variants
• BodhiApp Integration: Clean inheritance pattern with predictable paths
• Tag-based Versioning: Timestamp-based versions for chronological ordering