Merge pull request #2625 from ROCm/r2.14-add-ub22

[r2.14-rocm-enhanced] add ub22; unify Python install scripts
ROCm · Aug 14, 2024 · 4e7b1c1 · 4e7b1c1
2 parents 36a4c41 + 685d825
commit 4e7b1c1
Show file tree

Hide file tree

Showing 14 changed files with 310 additions and 236 deletions.
diff --git a/tensorflow/tools/tf_sig_build_dockerfiles/Dockerfile.rocm.manylinux2014 b/tensorflow/tools/tf_sig_build_dockerfiles/Dockerfile.rocm.manylinux2014
@@ -8,7 +8,7 @@ COPY setup.packages.rocm.cs7.sh setup.packages.rocm.cs7.sh
 COPY builder.packages.rocm.cs7.txt builder.packages.rocm.cs7.txt
 RUN /setup.packages.rocm.cs7.sh /builder.packages.rocm.cs7.txt
 
-ARG GPU_DEVICE_TARGETS="gfx900 gfx906 gfx908 gfx90a gfx940 gfx941 gfx942 gfx1030 gfx1100"
+ARG GPU_DEVICE_TARGETS="gfx908 gfx90a gfx940 gfx941 gfx942 gfx1030 gfx1100"
 ENV GPU_DEVICE_TARGETS=${GPU_DEVICE_TARGETS}
 
 # Install ROCM
@@ -18,9 +18,9 @@ ARG ROCM_PATH=/opt/rocm-${ROCM_VERSION}
 ENV ROCM_PATH=${ROCM_PATH}
 COPY ${CUSTOM_INSTALL} /${CUSTOM_INSTALL}
 COPY setup.packages.rocm.cs7.sh /setup.packages.rocm.cs7.sh
-COPY setup.rocm.cs7.sh /setup.rocm.cs7.sh
+COPY setup.rocm.sh /setup.rocm.sh
 COPY devel.packages.rocm.cs7.txt /devel.packages.rocm.cs7.txt
-RUN /setup.rocm.cs7.sh $ROCM_VERSION
+RUN /setup.rocm.sh $ROCM_VERSION el7
 
 # Install various tools.
 # - bats: bash unit testing framework

diff --git a/tensorflow/tools/tf_sig_build_dockerfiles/Dockerfile.rocm.manylinux_2_28 b/tensorflow/tools/tf_sig_build_dockerfiles/Dockerfile.rocm.manylinux_2_28
@@ -17,7 +17,7 @@ COPY setup.packages.rocm.el8.sh setup.packages.rocm.el8.sh
 COPY builder.packages.rocm.el8.txt builder.packages.rocm.el8.txt
 RUN /setup.packages.rocm.el8.sh /builder.packages.rocm.el8.txt
 
-ARG GPU_DEVICE_TARGETS="gfx900 gfx906 gfx908 gfx90a gfx940 gfx941 gfx942 gfx1030 gfx1100"
+ARG GPU_DEVICE_TARGETS="gfx908 gfx90a gfx940 gfx941 gfx942 gfx1030 gfx1100"
 ENV GPU_DEVICE_TARGETS=${GPU_DEVICE_TARGETS}
 
 # Install ROCM
@@ -26,9 +26,9 @@ ARG CUSTOM_INSTALL
 ARG ROCM_PATH=/opt/rocm-${ROCM_VERSION}
 ENV ROCM_PATH=${ROCM_PATH}
 COPY ${CUSTOM_INSTALL} /${CUSTOM_INSTALL}
-COPY setup.rocm.el8.sh /setup.rocm.el8.sh
+COPY setup.rocm.sh /setup.rocm.sh
 COPY devel.packages.rocm.el8.txt /devel.packages.rocm.el8.txt
-RUN /setup.rocm.el8.sh $ROCM_VERSION
+RUN /setup.rocm.sh $ROCM_VERSION el8
 
 # Install various tools.
 # - bats: bash unit testing framework

diff --git a/tensorflow/tools/tf_sig_build_dockerfiles/Dockerfile.rocm.rt b/tensorflow/tools/tf_sig_build_dockerfiles/Dockerfile.rocm.rt
diff --git a/tensorflow/tools/tf_sig_build_dockerfiles/Dockerfile.rocm.rt b/tensorflow/tools/tf_sig_build_dockerfiles/Dockerfile.rocm.rt
@@ -0,0 +1 @@
+Dockerfile.rocm.rt.ub20
diff --git a/tensorflow/tools/tf_sig_build_dockerfiles/Dockerfile.rocm.rt.ub20 b/tensorflow/tools/tf_sig_build_dockerfiles/Dockerfile.rocm.rt.ub20
@@ -0,0 +1,55 @@
+################################################################################
+ARG DISTRO_IMG
+FROM ${DISTRO_IMG:-'ubuntu:20.04'} as runtime
+################################################################################
+
+# Install dependencies
+COPY setup.packages.sh /setup.packages.sh
+COPY runtime.packages.txt /runtime.packages.txt
+COPY sles.runtime.packages.txt /sles.runtime.packages.txt
+RUN /setup.packages.sh /runtime.packages.txt
+
+# Install ROCM
+ARG TF_PKGS_DIR=tmp/packages
+ARG TENSORFLOW_PACKAGE=tf_nightly_rocm
+ARG ROCM_VERSION=6.1.2
+ARG CUSTOM_INSTALL
+ARG ROCM_PATH=/opt/rocm-${ROCM_VERSION}
+ENV ROCM_PATH=${ROCM_PATH}
+COPY ${TF_PKGS_DIR}/${TENSORFLOW_PACKAGE} /${TF_PKGS_DIR}/${TENSORFLOW_PACKAGE}
+COPY ${CUSTOM_INSTALL} /${CUSTOM_INSTALL}
+COPY setup.rocm.sh /setup.rocm.sh
+COPY devel.packages.rocm.txt /devel.packages.rocm.txt
+COPY sles.devel.packages.rocm.txt /sles.devel.packages.rocm.txt
+RUN /setup.rocm.sh $ROCM_VERSION focal
+
+# All lines past this point are reset when $CACHEBUSTER is set. We need this
+# for Python specifically because we install some nightly packages which are
+# likely to change daily.
+ARG CACHEBUSTER=0
+RUN echo $CACHEBUSTER
+
+# Setup Python environment. PYTHON_VERSION is e.g. "python3.8"
+ARG PYTHON_VERSION
+COPY setup.python.sh /setup.python.sh
+COPY devel.requirements.txt /devel.requirements.txt
+RUN /setup.python.sh $PYTHON_VERSION devel.requirements.txt true
+
+# Setup ENV variables for tensorflow pip build
+ENV TF_NEED_ROCM=1
+ENV TF_ROCM_GCC=1
+ENV ROCM_TOOLKIT_PATH=${ROCM_PATH}
+
+RUN pip install --no-cache-dir /${TF_PKGS_DIR}/${TENSORFLOW_PACKAGE}
+RUN echo 'ALL ALL=NOPASSWD:ALL' | tee /etc/sudoers.d/sudo-nopasswd
+
+ARG TF_TESTING_FL
+ENV TF_TESTING_FL=${TF_TESTING_FL}
+ARG DWLD_TF_SRC_CMD
+RUN if [ -n "${DWLD_TF_SRC_CMD}" ]; then eval "${DWLD_TF_SRC_CMD}"; fi
+RUN wget https://github.com/bazelbuild/bazelisk/releases/download/v1.11.0/bazelisk-linux-amd64 -O /usr/local/bin/bazel && \
+    chmod +x /usr/local/bin/bazel
+RUN git clone https://github.com/tensorflow/models.git
+RUN git clone https://github.com/tensorflow/examples.git
+RUN git clone https://github.com/tensorflow/autograph.git
+RUN git clone https://github.com/tensorflow/benchmarks.git
diff --git a/tensorflow/tools/tf_sig_build_dockerfiles/Dockerfile.rocm.rt.ub22 b/tensorflow/tools/tf_sig_build_dockerfiles/Dockerfile.rocm.rt.ub22
@@ -0,0 +1,55 @@
+################################################################################
+ARG DISTRO_IMG
+FROM ${DISTRO_IMG:-'ubuntu:22.04'} as runtime
+################################################################################
+
+# Install dependencies
+COPY setup.packages.sh /setup.packages.sh
+COPY runtime.packages.txt /runtime.packages.txt
+COPY sles.runtime.packages.txt /sles.runtime.packages.txt
+RUN /setup.packages.sh /runtime.packages.txt
+
+# Install ROCM
+ARG TF_PKGS_DIR=tmp/packages
+ARG TENSORFLOW_PACKAGE=tf_nightly_rocm
+ARG ROCM_VERSION=6.1.2
+ARG CUSTOM_INSTALL
+ARG ROCM_PATH=/opt/rocm-${ROCM_VERSION}
+ENV ROCM_PATH=${ROCM_PATH}
+COPY ${TF_PKGS_DIR}/${TENSORFLOW_PACKAGE} /${TF_PKGS_DIR}/${TENSORFLOW_PACKAGE}
+COPY ${CUSTOM_INSTALL} /${CUSTOM_INSTALL}
+COPY setup.rocm.sh /setup.rocm.sh
+COPY devel.packages.rocm.txt /devel.packages.rocm.txt
+COPY sles.devel.packages.rocm.txt /sles.devel.packages.rocm.txt
+RUN /setup.rocm.sh $ROCM_VERSION jammy
+
+# All lines past this point are reset when $CACHEBUSTER is set. We need this
+# for Python specifically because we install some nightly packages which are
+# likely to change daily.
+ARG CACHEBUSTER=0
+RUN echo $CACHEBUSTER
+
+# Setup Python environment. PYTHON_VERSION is e.g. "python3.8"
+ARG PYTHON_VERSION
+COPY setup.python.sh /setup.python.sh
+COPY devel.requirements.txt /devel.requirements.txt
+RUN /setup.python.sh $PYTHON_VERSION devel.requirements.txt true
+
+# Setup ENV variables for tensorflow pip build
+ENV TF_NEED_ROCM=1
+ENV TF_ROCM_GCC=1
+ENV ROCM_TOOLKIT_PATH=${ROCM_PATH}
+
+RUN pip install --no-cache-dir /${TF_PKGS_DIR}/${TENSORFLOW_PACKAGE}
+RUN echo 'ALL ALL=NOPASSWD:ALL' | tee /etc/sudoers.d/sudo-nopasswd
+
+ARG TF_TESTING_FL
+ENV TF_TESTING_FL=${TF_TESTING_FL}
+ARG DWLD_TF_SRC_CMD
+RUN if [ -n "${DWLD_TF_SRC_CMD}" ]; then eval "${DWLD_TF_SRC_CMD}"; fi
+RUN wget https://github.com/bazelbuild/bazelisk/releases/download/v1.11.0/bazelisk-linux-amd64 -O /usr/local/bin/bazel && \
+    chmod +x /usr/local/bin/bazel
+RUN git clone https://github.com/tensorflow/models.git
+RUN git clone https://github.com/tensorflow/examples.git
+RUN git clone https://github.com/tensorflow/autograph.git
+RUN git clone https://github.com/tensorflow/benchmarks.git
diff --git a/tensorflow/tools/tf_sig_build_dockerfiles/Dockerfile.rocm.ub20 b/tensorflow/tools/tf_sig_build_dockerfiles/Dockerfile.rocm.ub20
@@ -2,7 +2,7 @@
 FROM ubuntu:20.04
 ################################################################################
 
-ARG GPU_DEVICE_TARGETS="gfx900 gfx906 gfx908 gfx90a gfx940 gfx941 gfx942 gfx1030 gfx1100"
+ARG GPU_DEVICE_TARGETS="gfx908 gfx90a gfx940 gfx941 gfx942 gfx1030 gfx1100"
 ENV GPU_DEVICE_TARGETS=${GPU_DEVICE_TARGETS}
 
 # Install build dependencies
@@ -18,7 +18,7 @@ ENV ROCM_PATH=${ROCM_PATH}
 COPY ${CUSTOM_INSTALL} /${CUSTOM_INSTALL}
 COPY setup.rocm.sh /setup.rocm.sh
 COPY devel.packages.rocm.txt /devel.packages.rocm.txt
-RUN /setup.rocm.sh $ROCM_VERSION
+RUN /setup.rocm.sh $ROCM_VERSION focal
 
 # Install various tools.
 # - bats: bash unit testing framework

diff --git a/tensorflow/tools/tf_sig_build_dockerfiles/Dockerfile.rocm.ub22 b/tensorflow/tools/tf_sig_build_dockerfiles/Dockerfile.rocm.ub22
@@ -0,0 +1,72 @@
+################################################################################
+FROM ubuntu:22.04
+################################################################################
+
+ARG GPU_DEVICE_TARGETS="gfx908 gfx90a gfx940 gfx941 gfx942 gfx1030 gfx1100"
+ENV GPU_DEVICE_TARGETS=${GPU_DEVICE_TARGETS}
+
+# Install build dependencies
+COPY setup.packages.sh setup.packages.sh
+COPY builder.packages.txt builder.packages.txt
+RUN /setup.packages.sh /builder.packages.txt
+
+# Install ROCM
+ARG ROCM_VERSION=6.1.2
+ARG CUSTOM_INSTALL
+ARG ROCM_PATH=/opt/rocm-${ROCM_VERSION}
+ENV ROCM_PATH=${ROCM_PATH}
+COPY ${CUSTOM_INSTALL} /${CUSTOM_INSTALL}
+COPY setup.rocm.sh /setup.rocm.sh
+COPY devel.packages.rocm.txt /devel.packages.rocm.txt
+RUN /setup.rocm.sh $ROCM_VERSION jammy
+
+# Install various tools.
+# - bats: bash unit testing framework
+# - bazelisk: always use the correct bazel version
+# - buildifier: clean bazel build deps
+# - buildozer: clean bazel build deps
+# - gcloud SDK: communicate with Google Cloud Platform (GCP) for RBE, CI
+RUN git clone --branch v1.7.0 https://github.com/bats-core/bats-core.git && bats-core/install.sh /usr/local && rm -rf bats-core
+RUN wget https://github.com/bazelbuild/bazelisk/releases/download/v1.11.0/bazelisk-linux-amd64 -O /usr/local/bin/bazel && chmod +x /usr/local/bin/bazel
+RUN wget https://github.com/bazelbuild/buildtools/releases/download/3.5.0/buildifier -O /usr/local/bin/buildifier && chmod +x /usr/local/bin/buildifier
+RUN wget https://github.com/bazelbuild/buildtools/releases/download/3.5.0/buildozer -O /usr/local/bin/buildozer && chmod +x /usr/local/bin/buildozer
+RUN curl -sSL https://sdk.cloud.google.com > /tmp/gcloud && bash /tmp/gcloud --install-dir=~/usr/local/bin --disable-prompts
+
+
+# All lines past this point are reset when $CACHEBUSTER is set. We need this
+# for Python specifically because we install some nightly packages which are
+# likely to change daily.
+ARG CACHEBUSTER=0
+RUN echo $CACHEBUSTER
+
+# Setup Python environment. PYTHON_VERSION is e.g. "python3.8"
+ARG PYTHON_VERSION
+COPY setup.python.sh /setup.python.sh
+COPY devel.requirements.txt /devel.requirements.txt
+RUN /setup.python.sh $PYTHON_VERSION devel.requirements.txt
+
+ARG TF_WHEEL_URL
+RUN if [ -n "${TF_WHEEL_URL}" ]; then pip install "${TF_WHEEL_URL}"; fi
+ARG DWLD_TF_SRC_CMD
+RUN if [ -n "${DWLD_TF_SRC_CMD}" ]; then eval "${DWLD_TF_SRC_CMD}"; fi
+
+# Setup build and environment
+COPY devel.usertools /usertools
+COPY devel.bashrc /root/.bashrc
+
+# Setup ENV variables for tensorflow pip build
+ENV TF_NEED_ROCM=1
+ENV TF_ROCM_GCC=1
+ENV ROCM_TOOLKIT_PATH=${ROCM_PATH}
+
+# Don't use the bazel cache when a new docker image is created.
+RUN echo build --action_env=DOCKER_CACHEBUSTER=$(date +%s%N)$RANDOM >> /etc/bazel.bazelrc
+RUN echo build --host_action_env=DOCKER_HOST_CACHEBUSTER=$(date +%s%N)$RANDOM >> /etc/bazel.bazelrc
+
+ARG TF_TESTING_FL
+ENV TF_TESTING_FL=${TF_TESTING_FL}
+ARG DWLD_TF_SRC_CMD
+RUN if [ -n "${DWLD_TF_SRC_CMD}" ]; then eval "${DWLD_TF_SRC_CMD}"; fi
+ARG CLONE_TEST_REPO
+COPY ${CLONE_TEST_REPO} /${CLONE_TEST_REPO}
+RUN if [ -n "${CLONE_TEST_REPO}" ]; then bash /${CLONE_TEST_REPO}; fi
diff --git a/tensorflow/tools/tf_sig_build_dockerfiles/devel.usertools/code_check_full_ROCM.bats b/tensorflow/tools/tf_sig_build_dockerfiles/devel.usertools/code_check_full_ROCM.bats
@@ -15,7 +15,6 @@
 # limitations under the License.
 # ==============================================================================
 setup_file() {
-    cd /tf/tensorflow
     bazel version  # Start the bazel server
 }
 
@@ -151,14 +150,11 @@ _test_lib$
 //tensorflow:no_tensorflow_py_deps
 //tensorflow/tools/pip_package:win_pip_package_marker
 //tensorflow/core:image_testdata
-//tensorflow/core/lib/lmdb:lmdb_testdata
-//tensorflow/core/lib/lmdb/testdata:lmdb_testdata
 //tensorflow/core/kernels/cloud:bigquery_reader_ops
 //tensorflow/python:extra_py_tests_deps
 //tensorflow/python:mixed_precision
 //tensorflow/python:tf_optimizer
 //tensorflow/python:compare_test_proto_py
-//tensorflow/python/framework:test_ops_2
 //tensorflow/python/framework:test_file_system.so
 //tensorflow/python/debug:grpc_tensorflow_server.par
 //tensorflow/python/feature_column:vocabulary_testdata
@@ -302,7 +298,7 @@ EOF
 # anything with a Windows-only toolchain, and bazel errors if trying to build
 # that directory.
 @test "bazel nobuild passes on all of TF except TF Lite and win toolchains" {
-    bazel build --experimental_cc_shared_library --nobuild --keep_going -- //tensorflow/... -//tensorflow/lite/... -//tensorflow/tools/toolchains/win/... -//tensorflow/tools/toolchains/win_1803/... -//tensorflow/compiler/xla/stream_executor/cuda/...
+    bazel build --experimental_cc_shared_library --nobuild --keep_going -- //tensorflow/... -//tensorflow/lite/... -//tensorflow/tools/toolchains/win/... -//tensorflow/tools/toolchains/win_1803/...
 }
 
 

diff --git a/tensorflow/tools/tf_sig_build_dockerfiles/devel.usertools/gpu.bazelrc b/tensorflow/tools/tf_sig_build_dockerfiles/devel.usertools/gpu.bazelrc
@@ -103,7 +103,6 @@ test:nonpip_filters_multi_gpu --test_tag_filters=-no_gpu,-no_rocm
 test:nonpip_filters_multi_gpu --build_tag_filters=-no_gpu,-no_rocm
 test:nonpip_filters_multi_gpu --test_lang_filters=py --flaky_test_attempts=2 --test_size_filters=small,medium,large --test_env=TF_PER_DEVICE_MEMORY_LIMIT_MB=2048
 test:nonpip_multi_gpu --config=nonpip_filters_multi_gpu -- \
-//tensorflow/core/common_runtime/gpu:gpu_device_unified_memory_test_2gpu \
 //tensorflow/core/nccl:nccl_manager_test_2gpu \
 //tensorflow/python/distribute/integration_test:mwms_peer_failure_test_2gpu \
 //tensorflow/python/distribute:checkpoint_utils_test_2gpu \

diff --git a/tensorflow/tools/tf_sig_build_dockerfiles/devel.usertools/gpu_gcc.bazelrc b/tensorflow/tools/tf_sig_build_dockerfiles/devel.usertools/gpu_gcc.bazelrc
@@ -86,7 +86,6 @@ test:nonpip_filters_multi_gpu --test_tag_filters=-no_gpu,-no_rocm
 test:nonpip_filters_multi_gpu --build_tag_filters=-no_gpu,-no_rocm
 test:nonpip_filters_multi_gpu --test_lang_filters=py --flaky_test_attempts=2 --test_size_filters=small,medium,large --test_env=TF_PER_DEVICE_MEMORY_LIMIT_MB=2048
 test:nonpip_multi_gpu --config=nonpip_filters_multi_gpu -- \
-//tensorflow/core/common_runtime/gpu:gpu_device_unified_memory_test_2gpu \
 //tensorflow/core/nccl:nccl_manager_test_2gpu \
 //tensorflow/python/distribute/integration_test:mwms_peer_failure_test_2gpu \
 //tensorflow/python/distribute:checkpoint_utils_test_2gpu \