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 [ROCm/clr commit: ed155ace2e]
2016-08-31 18:20:16 -04:00
parent 59fbb048ea
commit ac10a39e22
1 changed files with 1 additions and 1 deletions
@@ -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
            );