From ed155ace2edfff0c416a2aee53800bf87d0f1876 Mon Sep 17 00:00:00 2001 From: foreman Date: Wed, 31 Aug 2016 18:20:16 -0400 Subject: [PATCH] P4 to Git Change 1309426 by lmoriche@lmoriche_opencl_dev on 2016/08/31 18:05:23 SWDEV-94610 - roundUp the kernarg segment size to the hidden arguments alignment (sizeof(size_t)). Fixes ocltst::oclperf::OCLPerfLDSReadSpeed. Affected files ... ... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocprogram.cpp#21 edit --- rocclr/runtime/device/rocm/rocprogram.cpp | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/rocclr/runtime/device/rocm/rocprogram.cpp b/rocclr/runtime/device/rocm/rocprogram.cpp index f61690152e..7993a07adb 100644 --- a/rocclr/runtime/device/rocm/rocprogram.cpp +++ b/rocclr/runtime/device/rocm/rocprogram.cpp @@ -1117,7 +1117,7 @@ namespace roc { // TODO: remove the workaround // add 24 bytes for global offsets as workaround for LC reporting // excluded the hidden arguments - kernargSegmentByteSize + numHiddenKernelArgs * sizeof(size_t), + amd::alignUp(kernargSegmentByteSize, sizeof(size_t)) + numHiddenKernelArgs * sizeof(size_t), kernargSegmentAlignment, numHiddenKernelArgs );