Upgrade default compiler to ROCm6.3 (#1723)

* upgrade to rocm6.3 compiler

* Proposed solution to convnd test failures in ROCm 6.3

---------
Co-authored-by: Andriy Roshchenko <andriy.roshchenko@amd.com>

Dockerfile

 FROM ubuntu:20.04
 ARG DEBIAN_FRONTEND=noninteractive
-ARG ROCMVERSION=6.2
+ARG ROCMVERSION=6.3
 ARG compiler_version=""
 ARG compiler_commit=""
 ARG CK_SCCACHE=""
@@ -13,17 +13,12 @@ RUN set -xe && \
     apt-get update && apt-get install -y --allow-unauthenticated apt-utils wget gnupg2 curl && \
     curl -fsSL https://repo.radeon.com/rocm/rocm.gpg.key | gpg --dearmor -o /etc/apt/trusted.gpg.d/rocm-keyring.gpg
 
-RUN if [ "$ROCMVERSION" != "6.3" ]; then \
-        sh -c "wget https://repo.radeon.com/amdgpu-install/$ROCMVERSION/ubuntu/focal/amdgpu-install_6.2.60200-1_all.deb  --no-check-certificate" && \
-        apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y --allow-unauthenticated ./amdgpu-install_6.2.60200-1_all.deb && \
+RUN if [ "$ROCMVERSION" != "6.4" ]; then \
+        sh -c "wget https://repo.radeon.com/amdgpu-install/$ROCMVERSION/ubuntu/focal/amdgpu-install_6.3.60300-1_all.deb  --no-check-certificate" && \
+        apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y --allow-unauthenticated ./amdgpu-install_6.3.60300-1_all.deb && \
         wget -qO - http://repo.radeon.com/rocm/rocm.gpg.key | apt-key add - && \
         sh -c "echo deb [arch=amd64 signed-by=/etc/apt/trusted.gpg.d/rocm-keyring.gpg] $DEB_ROCM_REPO focal main > /etc/apt/sources.list.d/rocm.list" && \
         sh -c 'echo deb [arch=amd64 signed-by=/etc/apt/trusted.gpg.d/rocm-keyring.gpg] https://repo.radeon.com/amdgpu/$ROCMVERSION/ubuntu focal main > /etc/apt/sources.list.d/amdgpu.list'; \
-    elif [ "$ROCMVERSION" = "6.3" ] && [ "$compiler_version" = "rc1" ]; then \
-        sh -c "wget http://artifactory-cdn.amd.com/artifactory/list/amdgpu-deb/amdgpu-install-internal_6.3-20.04-1_all.deb --no-check-certificate" && \
-        apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install dialog libpopt0 rsync && DEBIAN_FRONTEND=noninteractive apt-get install ./amdgpu-install-internal_6.3-20.04-1_all.deb && \
-        sh -c 'echo deb [arch=amd64 trusted=yes] http://compute-artifactory.amd.com/artifactory/list/rocm-release-archive-20.04-deb/ 6.3 rel-20 > /etc/apt/sources.list.d/rocm-build.list' && \
-        amdgpu-repo --amdgpu-build=2074281; \
     fi
 
 RUN sh -c "echo deb http://mirrors.kernel.org/ubuntu focal main universe | tee -a /etc/apt/sources.list" && \

Dockerfile.compiler

-ARG BASE_DOCKER="rocm/composable_kernel:ck_ub20.04_rocm6.2"
+ARG BASE_DOCKER="rocm/composable_kernel:ck_ub20.04_rocm6.3"
 FROM $BASE_DOCKER
 ARG compiler_version=""
 ARG compiler_commit=""

Jenkinsfile

@@ -38,7 +38,8 @@ def getBaseDockerImageName(){
         img = "${params.USE_CUSTOM_DOCKER}"
     }
     else{
-    if (params.ROCMVERSION != "6.3"){
+        def ROCM_numeric = "${params.ROCMVERSION}" as float
+        if ( ROCM_numeric < 6.4 ){
             img = "${env.CK_DOCKERHUB}:ck_ub20.04_rocm${params.ROCMVERSION}"
             }
         else{
@@ -739,8 +740,8 @@ def process_results(Map conf=[:]){
 }
 
 //launch develop branch daily at 23:00 UT in FULL_QA mode and at 19:00 UT with latest staging compiler version
-CRON_SETTINGS = BRANCH_NAME == "develop" ? '''0 23 * * * % RUN_FULL_QA=true;ROCMVERSION=6.2;RUN_CK_TILE_FMHA_TESTS=true;RUN_CK_TILE_GEMM_TESTS=true
-                                              0 21 * * * % ROCMVERSION=6.2;hipTensor_test=true;RUN_CODEGEN_TESTS=true
+CRON_SETTINGS = BRANCH_NAME == "develop" ? '''0 23 * * * % RUN_FULL_QA=true;ROCMVERSION=6.3;RUN_CK_TILE_FMHA_TESTS=true;RUN_CK_TILE_GEMM_TESTS=true
+                                              0 21 * * * % ROCMVERSION=6.3;hipTensor_test=true;RUN_CODEGEN_TESTS=true
                                               0 19 * * * % BUILD_DOCKER=true;DL_KERNELS=true;COMPILER_VERSION=amd-staging;BUILD_COMPILER=/llvm-project/build/bin/clang++;BUILD_GFX12=true;USE_SCCACHE=false;NINJA_BUILD_TRACE=true
                                               0 17 * * * % BUILD_DOCKER=true;DL_KERNELS=true;COMPILER_VERSION=amd-mainline;BUILD_COMPILER=/llvm-project/build/bin/clang++;BUILD_GFX12=true;USE_SCCACHE=false;NINJA_BUILD_TRACE=true
                                               0 15 * * * % BUILD_INSTANCES_ONLY=true;RUN_PERFORMANCE_TESTS=false;USE_SCCACHE=false
@@ -765,8 +766,8 @@ pipeline {
             description: 'If you want to use a custom docker image, please specify it here (default: leave blank).')
         string(
             name: 'ROCMVERSION', 
-            defaultValue: '6.2', 
-            description: 'Specify which ROCM version to use: 6.2 (default).')
+            defaultValue: '6.3',
+            description: 'Specify which ROCM version to use: 6.3 (default).')
         string(
             name: 'COMPILER_VERSION', 
             defaultValue: '', 

example/62_convnd_activ/convscale/convnd_fwd_convscale_common.hpp

@@ -172,12 +172,13 @@ bool run_grouped_conv_fwd(bool do_verification,
     {
     case 0: break;
     case 1:
-        in.GenerateTensorValue(GeneratorTensor_2<InDataType>{-5, 5});
-        wei.GenerateTensorValue(GeneratorTensor_2<WeiDataType>{-5, 5});
+        // values generated: -5, -4, -3, -2, -1, 0, 1, 2, 3, 4, 5
+        in.GenerateTensorValue(GeneratorTensor_2<InDataType>{-5, 6});
+        wei.GenerateTensorValue(GeneratorTensor_3<WeiDataType>{-1.0, 1.0});
         break;
     default:
-        in.GenerateTensorValue(GeneratorTensor_3<InDataType>{0.0, 1.0});
-        wei.GenerateTensorValue(GeneratorTensor_3<WeiDataType>{-0.5, 0.5});
+        in.GenerateTensorValue(GeneratorTensor_3<InDataType>{-5.0, 5.0});
+        wei.GenerateTensorValue(GeneratorTensor_3<WeiDataType>{-1.0, 1.0});
     }
 
     DeviceMem in_device_buf(sizeof(InDataType) * in.mDesc.GetElementSpaceSize());