Fix QA, allow switching compiler versions, fix google test compilation error. (#348)

* allow selecting compiler version

* fix typo

* add Wno-deprecated flag for google tests

* change git repo, fix qa log files names

* change the git clone syntax

* use Omkar's git credentials

* try to use jenkins as git user

* try using illsilin username for gerrit repo with ssh key

* try new gerrit authorization

* change ssh key syntax

* try another way of passing ssh key to docker

* add mount ssh in dockerfile

* create .ssh folder

* move ssh-keyscan to later

* get rid of npm call

* build first docker image on master

* check the contents of the .ssh folder

* try replacing omkars creds with gerrit creds

* use open repo, clean up changes

* get rid of ssh default argument

Dockerfile

@@ -16,7 +16,6 @@ RUN sh -c "echo deb [arch=amd64] $DEB_ROCM_REPO ubuntu main > /etc/apt/sources.l
 RUN wget --no-check-certificate -qO - https://apt.kitware.com/keys/kitware-archive-latest.asc 2>/dev/null | apt-key add -
 RUN sh -c "echo deb https://apt.kitware.com/ubuntu/ bionic main | tee -a /etc/apt/sources.list"
 
-# ADD requirements.txt requirements.txt
 # Install dependencies
 RUN apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y --allow-unauthenticated \
     apt-utils \
@@ -86,8 +85,8 @@ WORKDIR /
 ENV compiler_version=$compiler_version
 RUN sh -c "echo compiler version = '$compiler_version'"
 
-RUN if [ "$compiler_version" = "9110" ]; then \
-        git clone -b ck-9110 https://github.com/RadeonOpenCompute/llvm-project.git && \
+RUN --mount=type=ssh if [ "$compiler_version" != "release" ]; then \
+        git clone -b "$compiler_version" https://github.com/RadeonOpenCompute/llvm-project.git && \
         cd llvm-project && mkdir build && cd build && \
         cmake -DCMAKE_INSTALL_PREFIX=/opt/rocm/llvm -DCMAKE_BUILD_TYPE=Release -DLLVM_ENABLE_ASSERTIONS=1 -DLLVM_TARGETS_TO_BUILD="AMDGPU;X86" -DLLVM_ENABLE_PROJECTS="clang;lld;compiler-rt" ../llvm && \
         make -j 8 ; \

Jenkinsfile

@@ -92,7 +92,7 @@ def buildHipClangJob(Map conf=[:]){
         env.HSA_ENABLE_SDMA=0
         checkout scm
 
-        def image = "composable_kernels"
+        def image = "composable_kernels_${params.COMPILER_VERSION}"
         def prefixpath = conf.get("prefixpath", "/opt/rocm")
         def gpu_arch = conf.get("gpu_arch", "gfx908")
 
@@ -102,14 +102,10 @@ def buildHipClangJob(Map conf=[:]){
         if (conf.get("enforce_xnack_on", false)) {
             dockerOpts = dockerOpts + " --env HSA_XNACK=1"
         }
-        def dockerArgs
-        if (params.USE_9110){
-            dockerArgs = "--build-arg PREFIX=${prefixpath} --build-arg GPU_ARCH='${gpu_arch}' --build-arg compiler_version='9110' "
+        def dockerArgs = "--build-arg PREFIX=${prefixpath} --build-arg GPU_ARCH='${gpu_arch}' --build-arg compiler_version='${params.COMPILER_VERSION}' "
+        if (params.COMPILER_VERSION != "release"){
             dockerOpts = dockerOpts + " --env HIP_CLANG_PATH='/llvm-project/build/bin' "
         }
-        else{
-            dockerArgs = "--build-arg PREFIX=${prefixpath} --build-arg GPU_ARCH='${gpu_arch}' --build-arg compiler_version='release' "
-        }
 
         def variant = env.STAGE_NAME
 
@@ -185,7 +181,8 @@ def runCKProfiler(Map conf=[:]){
         env.HSA_ENABLE_SDMA=0
         checkout scm
 
-        def image = "composable_kernels"
+
+        def image = "composable_kernels_${params.COMPILER_VERSION}"
         def prefixpath = conf.get("prefixpath", "/opt/rocm")
         def gpu_arch = conf.get("gpu_arch", "gfx908")
 
@@ -195,14 +192,10 @@ def runCKProfiler(Map conf=[:]){
         if (conf.get("enforce_xnack_on", false)) {
             dockerOpts = dockerOpts + " --env HSA_XNACK=1"
         }
-        def dockerArgs
-        if (params.USE_9110){
-            dockerArgs = "--build-arg PREFIX=${prefixpath} --build-arg GPU_ARCH='${gpu_arch}' --build-arg compiler_version='9110' "
+        def dockerArgs = "--build-arg PREFIX=${prefixpath} --build-arg GPU_ARCH='${gpu_arch}' --build-arg compiler_version='${params.COMPILER_VERSION}' "
+        if (params.COMPILER_VERSION != "release"){
             dockerOpts = dockerOpts + " --env HIP_CLANG_PATH='/llvm-project/build/bin' "
         }
-        else{
-            dockerArgs = "--build-arg PREFIX=${prefixpath} --build-arg GPU_ARCH='${gpu_arch}' --build-arg compiler_version='release' "
-        }
 
         def variant = env.STAGE_NAME
         def retimage
@@ -248,20 +241,15 @@ def runCKProfiler(Map conf=[:]){
 					dir("script"){
                         if (params.RUN_FULL_QA){
                             def qa_log = "qa_${gpu_arch}.log"
-                            if (params.USE_9110){
-                                sh "./run_full_performance_tests.sh 1 QA_9110 ${gpu_arch} ${env.BRANCH_NAME} ${NODE_NAME}"
-                            }
-                            else{
-                                sh "./run_full_performance_tests.sh 1 QA_release ${gpu_arch} ${env.BRANCH_NAME} ${NODE_NAME}"
-                            }
+                            sh "./run_full_performance_tests.sh 1 QA_${params.COMPILER_VERSION} ${gpu_arch} ${env.BRANCH_NAME} ${NODE_NAME}"
                             archiveArtifacts "perf_gemm_${gpu_arch}.log"
                             archiveArtifacts "perf_resnet50_N256_${gpu_arch}.log"
                             archiveArtifacts "perf_resnet50_N4_${gpu_arch}.log"
                             archiveArtifacts "perf_batched_gemm_${gpu_arch}.log"
                             archiveArtifacts "perf_grouped_gemm_${gpu_arch}.log"
-                            archiveArtifacts "perf_fwd_conv_${gpu_arch}.log"
-                            archiveArtifacts "perf_bwd_conv_${gpu_arch}.log"
-                            archiveArtifacts "perf_fusion_${gpu_arch}.log"
+                            archiveArtifacts "perf_conv_fwd_${gpu_arch}.log"
+                            archiveArtifacts "perf_conv_bwd_${gpu_arch}.log"
+                            archiveArtifacts "perf_gemm_bilinear_${gpu_arch}.log"
                             archiveArtifacts "perf_reduction_${gpu_arch}.log"
                            // stash perf files to master
                             stash name: "perf_gemm_${gpu_arch}.log"
@@ -269,19 +257,14 @@ def runCKProfiler(Map conf=[:]){
                             stash name: "perf_resnet50_N4_${gpu_arch}.log"
                             stash name: "perf_batched_gemm_${gpu_arch}.log"
                             stash name: "perf_grouped_gemm_${gpu_arch}.log"
-                            stash name: "perf_fwd_conv_${gpu_arch}.log"
-                            stash name: "perf_bwd_conv_${gpu_arch}.log"
-                            stash name: "perf_fusion_${gpu_arch}.log"
+                            stash name: "perf_conv_fwd_${gpu_arch}.log"
+                            stash name: "perf_conv_bwd_${gpu_arch}.log"
+                            stash name: "perf_gemm_bilinear_${gpu_arch}.log"
                             stash name: "perf_reduction_${gpu_arch}.log"
                             //we will process results on the master node
                         }
                         else{
-                            if (params.USE_9110){
-                                sh "./run_performance_tests.sh 0 CI_9110 ${gpu_arch} ${env.BRANCH_NAME} ${NODE_NAME}"
-                            }
-                            else{
-                                sh "./run_performance_tests.sh 0 CI_release ${gpu_arch} ${env.BRANCH_NAME} ${NODE_NAME}"
-                            }
+                            sh "./run_performance_tests.sh 0 CI_${params.COMPILER_VERSION} ${gpu_arch} ${env.BRANCH_NAME} ${NODE_NAME}"
                             archiveArtifacts "perf_gemm_${gpu_arch}.log"
                             archiveArtifacts "perf_resnet50_N256_${gpu_arch}.log"
                             archiveArtifacts "perf_resnet50_N4_${gpu_arch}.log"
@@ -318,7 +301,7 @@ def runPerfTest(Map conf=[:]){
 def process_results(Map conf=[:]){
     env.HSA_ENABLE_SDMA=0
     checkout scm
-    def image = "composable_kernels"
+    def image = "composable_kernels_${params.COMPILER_VERSION}"
     def prefixpath = "/opt/rocm"
     def gpu_arch = conf.get("gpu_arch", "gfx908")
 
@@ -353,9 +336,9 @@ def process_results(Map conf=[:]){
                         unstash "perf_resnet50_N4_${gpu_arch}.log"
                         unstash "perf_batched_gemm_${gpu_arch}.log"
                         unstash "perf_grouped_gemm_${gpu_arch}.log"
-                        unstash "perf_fwd_conv_${gpu_arch}.log"
-                        unstash "perf_bwd_conv_${gpu_arch}.log"
-                        unstash "perf_fusion_${gpu_arch}.log"
+                        unstash "perf_conv_fwd_${gpu_arch}.log"
+                        unstash "perf_conv_bwd${gpu_arch}.log"
+                        unstash "perf_gemm_bilinear_${gpu_arch}.log"
                         unstash "perf_reduction_${gpu_arch}.log"
                         sh "./process_qa_data.sh ${gpu_arch}"
                     }
@@ -378,7 +361,7 @@ def process_results(Map conf=[:]){
 }
 
 //launch develop branch daily at 23:00 in FULL_QA mode
-CRON_SETTINGS = BRANCH_NAME == "develop" ? '''0 23 * * * % RUN_FULL_QA=true;USE_9110=true''' : ""
+CRON_SETTINGS = BRANCH_NAME == "develop" ? '''0 23 * * * % RUN_FULL_QA=true''' : ""
 
 pipeline {
     agent none
@@ -389,10 +372,10 @@ pipeline {
         parallelsAlwaysFailFast()
     }
     parameters {
-        booleanParam(
-            name: "USE_9110",
-            defaultValue: true,
-            description: "Select compiler version: 9110 (default) or release")
+        string(
+            name: 'COMPILER_VERSION', 
+            defaultValue: 'ck-9110', 
+            description: 'Specify which version of compiler to use: ck-9110 (default), release, or amd-mainline-open.')
         booleanParam(
             name: "RUN_FULL_QA",
             defaultValue: false,
@@ -406,6 +389,8 @@ pipeline {
         dbsshuser = "${dbsshuser}"
         dbsshpassword = "${dbsshpassword}"
         status_wrapper_creds = "${status_wrapper_creds}"
+        gerrit_cred="${gerrit_cred}"
+        DOCKER_BUILDKIT = "1"
     }
     stages{
         stage("Static checks") {

cmake/googletest.cmake

@@ -20,6 +20,7 @@ list(APPEND GTEST_CMAKE_CXX_FLAGS
      -Wno-unused-member-function
      -Wno-comma
      -Wno-old-style-cast
+     -Wno-deprecated
 )
 message(STATUS "Suppressing googltest warnings with flags: ${GTEST_CMAKE_CXX_FLAGS}")
 

script/process_perf_data.sh

@@ -12,5 +12,5 @@ pip3 install sqlalchemy pymysql pandas sshtunnel
 #process results
 gpu_arch=$1
 python3 process_perf_data.py perf_gemm_"$gpu_arch".log
-python3 process_perf_data.py perf_resnet50_N265_"$gpu_arch".log
-python3 process_perf_data.py perf_resnet50_N4_"$gpu_arch".log
+python3 process_perf_data.py perf_resnet50_N256_"$gpu_arch".log
+python3 process_perf_data.py perf_resnet50_N4_"$gpu_arch".log
\ No newline at end of file

script/process_qa_data.sh

@@ -12,11 +12,11 @@ pip3 install sqlalchemy pymysql pandas sshtunnel
 #process results
 gpu_arch=$1
 python3 process_perf_data.py perf_gemm_"$gpu_arch".log
-python3 process_perf_data.py perf_resnet50_N265_"$gpu_arch".log
+python3 process_perf_data.py perf_resnet50_N256_"$gpu_arch".log
 python3 process_perf_data.py perf_resnet50_N4_"$gpu_arch".log
 python3 process_perf_data.py perf_batched_gemm_"$gpu_arch".log
 python3 process_perf_data.py perf_grouped_gemm_"$gpu_arch".log
 python3 process_perf_data.py perf_conv_fwd_"$gpu_arch".log
 python3 process_perf_data.py perf_conv_bwd_data_"$gpu_arch".log
 python3 process_perf_data.py perf_gemm_bilinear_"$gpu_arch".log
-python3 process_perf_data.py perf_reduction_"$gpu_arch".log
+python3 process_perf_data.py perf_reduction_"$gpu_arch".log
\ No newline at end of file