LMCache/LMCache v0.5.7rc3-rocm-torch210

Release v0.5.7rc3 · AMD ROCm 7.2.4 / torch 2.10.0 git3d3aa833 / cp312 / CXX11 ABI=1

on GitHub

ROCm wheel for LMCache v0.5.7rc3, built and tested in
rocm/pytorch:rocm7.2.4_ubuntu24.04_py3.12_pytorch_release_2.10.0
at digest sha256:4449f856653602317e4101a76fce599c7fcd58ccec2e539951fce5f73083179e.

Supported ABI (exact):

• AMD torch wheel source: https://repo.radeon.com/rocm/manylinux/rocm-rel-7.2.4/torch-2.10.0%2Brocm7.2.4.lw.git3d3aa833-cp312-cp312-linux_x86_64.whl
• torch runtime: 2.10.0+rocm7.2.4.git3d3aa833 (git 3d3aa833db84eed6b7f5595cb5f162c2f78300a4)
• ROCm: 7.2.4 (HIP runtime 7.2.53211)
• Python/platform: cp312-cp312-manylinux_2_39_x86_64
• C++ ABI: _GLIBCXX_USE_CXX11_ABI=1

It includes gfx942/gfx950 GPU code objects and all integrations shipped
by the corresponding LMCache release.

Install into the pinned AMD PyTorch container:

VERSION=v0.5.7rc3
pip install lmcache==0.5.7rc3+rocm7.2.4.torch2.10.git3d3aa833.cxx11abi1 --no-deps \
  --find-links https://github.com/LMCache/LMCache/releases/expanded_assets/${VERSION}-rocm-torch210