AI/rocm-systems
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

P4 to Git Change 1531530 by gandryey@gera-w8 on 2018/03/23 13:45:11

Browse Source 
SWDEV-79445 - OCL generic changes and code clean-up
	- Add explicit type specialization for AQL copy types to avoid memcpy() calls in the most of cases
	- Increase size of sysmem copy for constants buffers to account the internal arguments

Affected files ...

... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palconstbuf.cpp#5 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palkernel.cpp#46 edit
This commit is contained in:
foreman
2018-03-23 14:12:49 -04:00
parent 6002477091
commit 62a00d336e
2 changed files with 47 additions and 17 deletions
Download Patch File Download Diff File Expand all files Collapse all files
rocclr/runtime/device/pal/palconstbuf.cpp
+3 -2
View File
@@ -33,9 +33,10 @@ ManagedBuffer::~ManagedBuffer() {
// ================================================================================================
bool ManagedBuffer::create(Resource::MemoryType type, bool constbuf) {
if (constbuf) {
// Create sysmem copy for the constant buffer
// Create sysmem copy for the constant buffer.
// Allocate extra memory to account the internal arguments
sysMemCopy_ = reinterpret_cast<address>(amd::AlignedMemory::allocate(
gpu_.dev().info().maxParameterSize_, 256));
2 * gpu_.dev().info().maxParameterSize_, 256));
if (sysMemCopy_ == nullptr) {
LogPrintfError("We couldn't allocate sysmem copy for constant buffer, size(%d)!", size_);
return false;