[ROCm][CI/Build] ROCm 7.2.1 release version; torch 2.10; triton 3.6 (#38252)

Signed-off-by: Gregory Shtrasberg <Gregory.Shtrasberg@amd.com>

docker/Dockerfile.rocm

@@ -386,6 +386,9 @@ ENV MIOPEN_DEBUG_CONV_GEMM=0
 # will not be imported by other tests
 RUN mkdir src && mv vllm src/vllm
 
+# This is a workaround to ensure pytest exits with the correct status code in CI tests.
+RUN echo "import os\n\ndef pytest_sessionfinish(session, exitstatus):\n    os._exit(int(exitstatus))" > /vllm-workspace/conftest.py
+
 # -----------------------
 # Final vLLM image
 FROM base AS final

docker/Dockerfile.rocm_base

-ARG BASE_IMAGE=rocm/dev-ubuntu-22.04:7.0-complete
-ARG TRITON_BRANCH="57c693b6"
+ARG BASE_IMAGE=rocm/dev-ubuntu-22.04:7.2.1-complete
+ARG TRITON_BRANCH="ba5c1517"
 ARG TRITON_REPO="https://github.com/ROCm/triton.git"
-ARG PYTORCH_BRANCH="89075173"
+ARG PYTORCH_BRANCH="8514f051" # release/2.10 as of 3/17
 ARG PYTORCH_REPO="https://github.com/ROCm/pytorch.git"
 ARG PYTORCH_VISION_BRANCH="v0.24.1"
 ARG PYTORCH_VISION_REPO="https://github.com/pytorch/vision.git"
@@ -114,6 +114,8 @@ ARG TRITON_REPO
 RUN git clone ${TRITON_REPO}
 RUN cd triton \
     && git checkout ${TRITON_BRANCH} \
+    && git config --global user.email "you@example.com" && git config --global user.name "Your Name" \
+    && git cherry-pick 555d04f \
     && if [ ! -f setup.py ]; then cd python; fi \
     && python3 setup.py bdist_wheel --dist-dir=dist \
     && mkdir -p /app/install && cp dist/*.whl /app/install
@@ -142,10 +144,14 @@ ARG PYTORCH_VISION_REPO
 ARG PYTORCH_AUDIO_REPO
 ARG USE_SCCACHE
 
+RUN apt-get update && apt-get install -y pkg-config liblzma-dev
 RUN git clone ${PYTORCH_REPO} pytorch
-RUN cd pytorch && git checkout ${PYTORCH_BRANCH} \
-    && pip install -r requirements.txt && git submodule update --init --recursive \
-    && python3 tools/amd_build/build_amd.py \
+RUN cd pytorch && git checkout ${PYTORCH_BRANCH}
+RUN cd pytorch \
+    && pip install -r requirements.txt && git submodule update --init --recursive
+RUN cd pytorch/third_party/kineto \
+    && git remote add rocm https://github.com/ROCm/kineto && git fetch rocm && git checkout 2d73be3 
+RUN cd pytorch && python3 tools/amd_build/build_amd.py \
     && if [ "$USE_SCCACHE" = "1" ]; then \
            export HIP_CLANG_PATH=/opt/sccache-wrappers \
            && export CMAKE_C_COMPILER_LAUNCHER=sccache \
@@ -239,7 +245,7 @@ RUN pip install pyyaml && cd aiter \
            export HIP_CLANG_PATH=/opt/sccache-wrappers \
            && sccache --show-stats; \
        fi \
-    && GPU_ARCHS=${AITER_ROCM_ARCH} python3 setup.py bdist_wheel --dist-dir=dist \
+    && PREBUILD_KERNELS=1 GPU_ARCHS=${AITER_ROCM_ARCH} python3 setup.py bdist_wheel --dist-dir=dist \
     && if [ "$USE_SCCACHE" = "1" ]; then sccache --show-stats; fi \
     && ls /app/aiter/dist/*.whl
 RUN mkdir -p /app/install && cp /app/aiter/dist/*.whl /app/install