strix-llama.cpp (sha256:0f08b450cdcc8cd46d32f57bd7d20b6dd402ff717f1005ca5d395992c6f42b44)
Installation
docker pull git.clowdertech.com/clowdertech/strix-llama.cpp@sha256:0f08b450cdcc8cd46d32f57bd7d20b6dd402ff717f1005ca5d395992c6f42b44sha256:0f08b450cdcc8cd46d32f57bd7d20b6dd402ff717f1005ca5d395992c6f42b44About this package
GSQHalo.cpp: llama.cpp for GSQ-quantized Qwen3.8-Flash-Next on AMD Strix Halo, with a persistent KV cache on SSD
Image layers
| KIWI 11.0.2 |
| ARG ROCM_VERSION=7.14 |
| ARG ROCM_REPO=https://repo.amd.com/rocm/packages-multi-arch/rhel10/x86_64 |
| ARG COMMIT=unknown |
| ARG JOBS=16 |
| RUN |4 ROCM_VERSION=7.14 ROCM_REPO=https://repo.amd.com/rocm/packages-multi-arch/rhel10/x86_64 COMMIT=unknown JOBS=16 /bin/sh -c printf '%s\n' '[rocm]' "name=ROCm ${ROCM_VERSION}" "baseurl=${ROCM_REPO}" 'enabled=1' 'priority=50' 'gpgcheck=1' 'gpgkey=https://repo.amd.com/rocm/packages-multi-arch/gpg/rocm.gpg' > /etc/yum.repos.d/rocm.repo && dnf -y --nodocs --setopt=install_weak_deps=False install make gcc gcc-c++ cmake ninja-build libcurl-devel rdma-core-devel git-core python3 procps-ng amdrocm-core-devel${ROCM_VERSION}-gfx1151 && dnf clean all && rm -rf /var/cache/dnf/* # buildkit |
| ENV ROCM_PATH=/opt/rocm HIP_PATH=/opt/rocm PATH=/opt/rocm/bin:/opt/rocm/core/bin:/opt/rocm/core/lib/llvm/bin:/usr/local/bin:/usr/bin LD_LIBRARY_PATH=/opt/rocm/core/lib/rocm_sysdeps/lib:/opt/rocm/core/lib ROCBLAS_USE_HIPBLASLT=1 |
| LABEL org.opencontainers.image.source=https://github.com/Aristo94/GSQHalo.cpp org.opencontainers.image.revision=unknown org.opencontainers.image.licenses=MIT org.opencontainers.image.description=GSQHalo.cpp: llama.cpp for GSQ-quantized Qwen3.8-Flash-Next on AMD Strix Halo, with a persistent KV cache on SSD |
| WORKDIR /opt/llama.cpp |
| COPY . /opt/llama.cpp # buildkit |
| RUN |4 ROCM_VERSION=7.14 ROCM_REPO=https://repo.amd.com/rocm/packages-multi-arch/rhel10/x86_64 COMMIT=unknown JOBS=16 /bin/sh -c cmake -S . -B build -DGGML_HIP=ON -DAMDGPU_TARGETS=gfx1151 -DCMAKE_BUILD_TYPE=Release -DROCM_PATH=/opt/rocm -DHIP_PLATFORM=amd -DLLAMA_BUILD_TESTS=ON && cmake --build build --config Release -- -j${JOBS} && cmake --install build --config Release && echo /usr/local/lib64 > /etc/ld.so.conf.d/usr-local-lib64.conf && ldconfig && echo ${COMMIT} > /opt/llama.cpp/BUILD_COMMIT # buildkit |
| CMD ["/bin/bash"] |
Labels
| Key | Value |
|---|---|
| io.buildah.version | 1.43.4 |
| org.opencontainers.image.description | GSQHalo.cpp: llama.cpp for GSQ-quantized Qwen3.8-Flash-Next on AMD Strix Halo, with a persistent KV cache on SSD |
| org.opencontainers.image.licenses | MIT |
| org.opencontainers.image.revision | unknown |
| org.opencontainers.image.source | https://github.com/Aristo94/GSQHalo.cpp |
| org.opencontainers.image.title | fedora |
| org.opencontainers.image.url | https://fedoraproject.org/ |
| org.opencontainers.image.vendor | Fedora Project |
| org.opencontainers.image.version | 44 |
Details
2026-10-08 22:50:54 +00:00
Versions (3)
View all
Container
1
OCI / Docker
linux/amd64
MIT
2.1 GiB