Mirrors
/
hf_text-generation-inference
mirror of https://github.com/huggingface/text-generation-inference.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

add gfx1100 support to AMD pytorch build 

see https://github.com/huggingface/text-generation-inference/issues/2641
This commit is contained in:
Drew Paettie 2024-10-12 22:55:49 -07:00
parent 0c478846c5
commit af546505ad
No known key found for this signature in database
1 changed files with 2 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
Dockerfile_amd
View File

@ -78,7 +78,7 @@ ARG PYTHON_VERSION='3.11.10'
ARG TARGETPLATFORM ARG TARGETPLATFORM
ENV PATH=/opt/conda/bin:$PATH ENV PATH=/opt/conda/bin:$PATH
ARG PYTORCH_ROCM_ARCH="gfx90a;gfx942" ARG PYTORCH_ROCM_ARCH="gfx90a;gfx942;gfx1100"
# TGI seem to require libssl.so.1.1 instead of libssl.so.3 so we can't use ubuntu 22.04. Ubuntu 20.04 has python==3.8, and TGI requires python>=3.9, hence the need for miniconda. # TGI seem to require libssl.so.1.1 instead of libssl.so.3 so we can't use ubuntu 22.04. Ubuntu 20.04 has python==3.8, and TGI requires python>=3.9, hence the need for miniconda.
# Install mamba # Install mamba
@ -170,7 +170,7 @@ RUN --mount=type=bind,from=export_hipblaslt,src=/,target=/install \
fi fi
ARG BUILD_ENVIRONMENT=pytorch-linux-jammy-rocm6.2-py3.11 ARG BUILD_ENVIRONMENT=pytorch-linux-jammy-rocm6.2-py3.11
ARG PYTORCH_ROCM_ARCH="gfx90a;gfx942" ARG PYTORCH_ROCM_ARCH="gfx90a;gfx942;gfx1100"
# A commit to fix the output scaling factor issue in _scaled_mm # A commit to fix the output scaling factor issue in _scaled_mm
# Not yet in 2.5.0-rc1 # Not yet in 2.5.0-rc1