P4 to Git Change 1746912 by cpaquot@cpaquot-ocl-lc-lnx on 2019/02/21 14:16:35

SWDEV-178453 - [HIP] Add extra parameter for sharedMemBytes

Affected files ...

... //depot/stg/opencl/drivers/opencl/api/hip/hip_module.cpp#20 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palvirtual.cpp#129 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palvirtual.hpp#58 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocvirtual.cpp#72 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocvirtual.hpp#21 edit
... //depot/stg/opencl/drivers/opencl/runtime/platform/command.cpp#91 edit
... //depot/stg/opencl/drivers/opencl/runtime/platform/command.hpp#91 edit


[ROCm/clr commit: a343072c71]
This commit is contained in:
foreman
2019-02-21 17:42:08 -05:00
parent 10520b6487
commit d9bdadb673
6 changed files with 17 additions and 11 deletions
@@ -226,8 +226,8 @@ void Command::enqueue() {
const Context& Command::context() const { return queue_->context(); }
NDRangeKernelCommand::NDRangeKernelCommand(HostQueue& queue, const EventWaitList& eventWaitList,
Kernel& kernel, const NDRangeContainer& sizes)
: Command(queue, CL_COMMAND_NDRANGE_KERNEL, eventWaitList), kernel_(kernel), sizes_(sizes) {
Kernel& kernel, const NDRangeContainer& sizes, uint32_t sharedMemBytes)
: Command(queue, CL_COMMAND_NDRANGE_KERNEL, eventWaitList), kernel_(kernel), sizes_(sizes), sharedMemBytes_(sharedMemBytes) {
auto& device = queue.device();
auto devKernel = const_cast<device::Kernel*>(kernel.getDeviceKernel(device));
profilingInfo_.setCallback(devKernel->getProfilingCallback(
@@ -752,11 +752,12 @@ class NDRangeKernelCommand : public Command {
Kernel& kernel_;
NDRangeContainer sizes_;
address parameters_;
uint32_t sharedMemBytes_;
public:
//! Construct an ExecuteKernel command
NDRangeKernelCommand(HostQueue& queue, const EventWaitList& eventWaitList, Kernel& kernel,
const NDRangeContainer& sizes);
const NDRangeContainer& sizes, uint32_t sharedMemBytes = 0);
virtual void submit(device::VirtualDevice& device) { device.submitKernel(*this); }
@@ -772,6 +773,9 @@ class NDRangeKernelCommand : public Command {
//! Return the kernel NDRange.
const NDRangeContainer& sizes() const { return sizes_; }
//! Return the shared memory size
uint32_t sharedMemBytes() const { return sharedMemBytes_; }
//! Set the local work size.
void setLocalWorkSize(const NDRange& local) { sizes_.local() = local; }