Sign inSign up

rocm/vllm-dev:quark-ultraquant-lmcache-mi355

Manifest digest

sha256:563d12ab9d6ffb22f4ff1a24ce630ed29829812c2a1f925a94ae350d80ff7bc0

OS/ARCH

linux/amd64

Compressed size

11.83 GB

Last pushed

2 months by dllehramd

Type

Image

Manifest digest

sha256:563d12ab9d6ffb22f4ff1a24ce630ed29829812c2a1f925a94ae350d80ff7bc0

Image Layers

1ARG RELEASE0 B
2ARG LAUNCHPAD_BUILD_ARCH0 B
3LABEL org.opencontainers.image.version=22.040 B
4ADD file ... in / 28.36 MB
5CMD ["/bin/bash"]0 B
7ARG ROCM_VERSION=5.30 B
8ARG AMDGPU_VERSION=5.30 B
9ARG APT_PREF0 B
10RUN |3 ROCM_VERSION=7.2.3 AMDGPU_VERSION=30.30.3 APT_PREF=Package:240 B
11RUN |3 ROCM_VERSION=7.2.3 AMDGPU_VERSION=30.30.3 APT_PREF=Package:6.76 GB
12RUN |3 ROCM_VERSION=7.2.3 AMDGPU_VERSION=30.30.3 APT_PREF=Package:655 B
13ENV PATH=/opt/rocm/llvm/bin:/opt/rocm/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin0 B
14ENV ROCM_PATH=/opt/rocm0 B
15ENV LD_LIBRARY_PATH=/opt/rocm/lib:/usr/local/lib:0 B
16ARG PYTORCH_ROCM_ARCH=gfx90a;gfx942;gfx950;gfx1100;gfx1101;gfx1200;gfx1201;gfx1150;gfx11510 B
17ENV PYTORCH_ROCM_ARCH=gfx90a;gfx942;gfx950;gfx1100;gfx1101;gfx1200;gfx1201;gfx1150;gfx11510 B
18ENV AITER_ROCM_ARCH=gfx942;gfx9500 B
19ENV MORI_GPU_ARCHS=gfx942;gfx9500 B
20ENV HSA_NO_SCRATCH_RECLAIM=10 B
21ARG PYTHON_VERSION=3.120 B
22ENV PYTHON_VERSION=3.120 B
23RUN |2 PYTORCH_ROCM_ARCH=gfx90a;gfx942;gfx950;gfx1100;gfx1101;gfx1200;gfx1201;gfx1150;gfx1151 PYTHON_VERSION=3.12 /bin/sh93 B
24WORKDIR /app32 B
25ENV DEBIAN_FRONTEND=noninteractive0 B
26RUN |2 PYTORCH_ROCM_ARCH=gfx90a;gfx942;gfx950;gfx1100;gfx1101;gfx1200;gfx1201;gfx1150;gfx1151 PYTHON_VERSION=3.12 /bin/sh147.6 MB
27RUN |2 PYTORCH_ROCM_ARCH=gfx90a;gfx942;gfx950;gfx1100;gfx1101;gfx1200;gfx1201;gfx1150;gfx1151 PYTHON_VERSION=3.12 /bin/sh65.96 MB
28RUN |2 PYTORCH_ROCM_ARCH=gfx90a;gfx942;gfx950;gfx1100;gfx1101;gfx1200;gfx1201;gfx1150;gfx1151 PYTHON_VERSION=3.12 /bin/sh3.84 MB
29ARG USE_SCCACHE0 B
30ARG SCCACHE_DOWNLOAD_URL0 B
31ARG SCCACHE_ENDPOINT0 B
32ARG SCCACHE_BUCKET_NAME=vllm-build-sccache0 B
33ARG SCCACHE_REGION_NAME=us-west-20 B
34ARG SCCACHE_S3_NO_CREDENTIALS=00 B
35RUN |8 PYTORCH_ROCM_ARCH=gfx90a;gfx942;gfx950;gfx1100;gfx1101;gfx1200;gfx1201;gfx1150;gfx1151 PYTHON_VERSION=3.12 USE_SCCACHE=32 B
36RUN |8 PYTORCH_ROCM_ARCH=gfx90a;gfx942;gfx950;gfx1100;gfx1101;gfx1200;gfx1201;gfx1150;gfx1151 PYTHON_VERSION=3.12 USE_SCCACHE=32 B
37ARG USE_SCCACHE0 B
38ENV SCCACHE_BUCKET=0 B
39ENV SCCACHE_REGION=0 B
40ENV SCCACHE_S3_NO_CREDENTIALS=0 B
41ENV SCCACHE_IDLE_TIMEOUT=0 B
42RUN |8 PYTORCH_ROCM_ARCH=gfx90a;gfx942;gfx950;gfx1100;gfx1101;gfx1200;gfx1201;gfx1150;gfx1151 PYTHON_VERSION=3.12 USE_SCCACHE=2.2 GB
43ARG BASE_IMAGE=rocm/dev-ubuntu-22.04:7.2.3-complete0 B
44ARG TRITON_BRANCH=ba5c15170 B
45ARG TRITON_REPO=https://github.com/ROCm/triton.git0 B
46ARG PYTORCH_BRANCH=8514f0510 B
47ARG PYTORCH_VISION_BRANCH=v0.24.10 B
48ARG PYTORCH_REPO=https://github.com/ROCm/pytorch.git0 B
49ARG PYTORCH_VISION_REPO=https://github.com/pytorch/vision.git0 B
50ARG PYTORCH_AUDIO_BRANCH=v2.9.00 B
51ARG PYTORCH_AUDIO_REPO=https://github.com/pytorch/audio.git0 B
52ARG FA_BRANCH=0e60e3940 B
53ARG FA_REPO=https://github.com/Dao-AILab/flash-attention.git0 B
54ARG AITER_BRANCH=v0.1.13.post10 B
55ARG AITER_REPO=https://github.com/ROCm/aiter.git0 B
56ARG MORI_BRANCH=v1.1.00 B
57ARG MORI_REPO=https://github.com/ROCm/mori.git0 B
58RUN |23 PYTORCH_ROCM_ARCH=gfx90a;gfx942;gfx950;gfx1100;gfx1101;gfx1200;gfx1201;gfx1150;gfx1151 PYTHON_VERSION=3.12 USE_SCCACHE=396 B
59ARG ARG_PYTORCH_ROCM_ARCH0 B
60ENV PYTORCH_ROCM_ARCH=gfx90a;gfx942;gfx950;gfx1100;gfx1101;gfx1200;gfx1201;gfx1150;gfx11510 B
61RUN |1 ARG_PYTORCH_ROCM_ARCH= /bin/sh -c140.56 MB
62RUN |1 ARG_PYTORCH_ROCM_ARCH= /bin/sh -c3.56 MB
63ARG USE_SCCACHE0 B
64RUN |2 ARG_PYTORCH_ROCM_ARCH= USE_SCCACHE= /bin/sh32 B
65RUN |2 ARG_PYTORCH_ROCM_ARCH= USE_SCCACHE= /bin/sh23.83 MB
66ENV UV_HTTP_TIMEOUT=5000 B
67ENV UV_INDEX_STRATEGY=unsafe-best-match0 B
68ENV UV_LINK_MODE=copy0 B
69ENV CCACHE_DIR=/root/.cache/ccache0 B
70ENV CCACHE_COMPILERCHECK=content0 B
71ARG max_jobs0 B
72ENV MAX_JOBS=0 B
73ARG USE_SCCACHE0 B
74ARG SCCACHE_DOWNLOAD_URL0 B
75ARG SCCACHE_ENDPOINT0 B
76ARG SCCACHE_BUCKET_NAME=vllm-build-sccache0 B
77ARG SCCACHE_REGION_NAME=us-west-20 B
78ARG SCCACHE_S3_NO_CREDENTIALS=00 B
79RUN |8 ARG_PYTORCH_ROCM_ARCH= USE_SCCACHE= max_jobs=32 B
80ARG USE_SCCACHE0 B
81ENV SCCACHE_BUCKET=0 B
82ENV SCCACHE_REGION=0 B
83ENV SCCACHE_S3_NO_CREDENTIALS=0 B
84ENV SCCACHE_IDLE_TIMEOUT=0 B
85ARG COMMON_WORKDIR=/app0 B
86WORKDIR /app32 B
87ARG NIC_BACKEND=all0 B
88ARG AINIC_VERSION=1.117.3-hydra0 B
89ARG UBUNTU_CODENAME=jammy0 B
90RUN |12 ARG_PYTORCH_ROCM_ARCH= USE_SCCACHE= max_jobs=15.64 MB
91RUN |12 ARG_PYTORCH_ROCM_ARCH= USE_SCCACHE= max_jobs=1.36 KB
92RUN |12 ARG_PYTORCH_ROCM_ARCH= USE_SCCACHE= max_jobs=32 B
93ENV SCCACHE_BUCKET=0 B
94ENV SCCACHE_REGION=0 B
95ENV SCCACHE_ENDPOINT=0 B
96ENV SCCACHE_S3_NO_CREDENTIALS=0 B
97ENV SCCACHE_IDLE_TIMEOUT=0 B
98RUN |12 ARG_PYTORCH_ROCM_ARCH= USE_SCCACHE= max_jobs=32 B
99RUN |12 ARG_PYTORCH_ROCM_ARCH= USE_SCCACHE= max_jobs=6.71 MB
100RUN |12 ARG_PYTORCH_ROCM_ARCH= USE_SCCACHE= max_jobs=778.08 MB
101RUN |12 ARG_PYTORCH_ROCM_ARCH= USE_SCCACHE= max_jobs=37.63 MB
102ARG COMMON_WORKDIR=/app0 B
103ARG BASE_IMAGE=rocm/vllm-dev:base0 B
104ARG NIC_BACKEND=all0 B
105ARG AINIC_VERSION=1.117.3-hydra0 B
106COPY /benchmarks /app/vllm/benchmarks # buildkit249.64 KB
107COPY /examples /app/vllm/examples # buildkit257.63 KB
108COPY /docker /app/vllm/docker # buildkit9.13 KB
109ENV HSA_ENABLE_IPC_MODE_LEGACY=10 B
110ENV TOKENIZERS_PARALLELISM=false0 B
111ENV SAFETENSORS_FAST_GPU=10 B
112ENV HIP_FORCE_DEV_KERNARG=10 B
113RUN |13 ARG_PYTORCH_ROCM_ARCH= USE_SCCACHE= max_jobs=168 B
114ENV KINETO_CONFIG=/app/libkineto.conf0 B
115RUN |13 ARG_PYTORCH_ROCM_ARCH= USE_SCCACHE= max_jobs=451 B
116CMD ["/bin/bash"]0 B
117ENV ROCM_PATH=/opt/rocm0 B
118ENV PYTORCH_ROCM_ARCH=gfx942;gfx9500 B
119WORKDIR /workspace99 B
120ARG VLLM_REF=3cc38a1b65dfaee246ebeb9cc203b4d74bbcb6050 B
121ARG LMCACHE_REF=efd5389a0d8b5b2e48033848eb1f694bd41ee19e0 B
122COPY flydsl_41500b0_gfx950_v2.tgz /tmp/flydsl.tgz # buildkit181.87 MB
123RUN |2 VLLM_REF=3cc38a1b65dfaee246ebeb9cc203b4d74bbcb605 LMCACHE_REF=efd5389a0d8b5b2e48033848eb1f694bd41ee19e /bin/sh188.67 MB
124ENV VLLM_FLYDSL_ROOT=/root/FlyDSL0 B
125ENV VLLM_FLYDSL_PKGS=/root/FlyDSL/build-fly/python_packages0 B
126ENV PYTHONPATH=/root/FlyDSL:/root/FlyDSL/build-fly/python_packages0 B
127ARG MAX_JOBS=640 B
128ENV MAX_JOBS=640 B
129ARG VLLM_VERSION=0.19.2rc1.dev111+g3cc38a1b60 B
130COPY vllm_src_3cc38a1b.tgz /tmp/vllm_src.tgz # buildkit30.49 MB
131RUN |4 VLLM_REF=3cc38a1b65dfaee246ebeb9cc203b4d74bbcb605 LMCACHE_REF=efd5389a0d8b5b2e48033848eb1f694bd41ee19e MAX_JOBS=641.04 GB
132RUN |4 VLLM_REF=3cc38a1b65dfaee246ebeb9cc203b4d74bbcb605 LMCACHE_REF=efd5389a0d8b5b2e48033848eb1f694bd41ee19e MAX_JOBS=64222.96 MB
133RUN |4 VLLM_REF=3cc38a1b65dfaee246ebeb9cc203b4d74bbcb605 LMCACHE_REF=efd5389a0d8b5b2e48033848eb1f694bd41ee19e MAX_JOBS=64507.18 KB
134ENV PYTHONHASHSEED=00 B
135ENV LMCACHE_PRE_CACHING_HASH_ALGORITHM=sha2560 B
136RUN |4 VLLM_REF=3cc38a1b65dfaee246ebeb9cc203b4d74bbcb605 LMCACHE_REF=efd5389a0d8b5b2e48033848eb1f694bd41ee19e MAX_JOBS=641.12 MB
137WORKDIR /workspace32 B
138ENTRYPOINT ["vllm" "serve"]0 B

Command

 ARG RELEASE