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

P4 to Git Change 1775995 by gandryey@gera-w8 on 2019/04/29 13:46:53

Browse Source 
SWDEV-79445 - OCL generic changes and code clean-up
	- Enable P2P extension for PAL path, currently it's staging copy only
	- Fix P2P staging copy

Affected files ...

... //depot/stg/opencl/drivers/opencl/api/hip/hip_memory.cpp#55 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/device.cpp#243 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/device.hpp#336 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/pal/paldevice.cpp#130 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palsettings.cpp#77 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palvirtual.cpp#131 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/pal/palvirtual.hpp#59 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocdevice.cpp#124 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocdevice.hpp#36 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/rocm/rocvirtual.cpp#74 edit
... //depot/stg/opencl/drivers/opencl/runtime/platform/command.cpp#92 edit
This commit is contained in:
foreman
2019-04-29 13:58:26 -04:00
parent 1dddb78021
commit 1e33e8be4d
10 changed files with 210 additions and 45 deletions
Download Patch File Download Diff File Expand all files Collapse all files
rocclr/runtime/device/rocm/rocdevice.cpp
+31 -16
View File
@@ -49,8 +49,6 @@ amd::Device::Compiler* NullDevice::compilerHandle_;
bool roc::Device::isHsaInitialized_ = false;
hsa_agent_t roc::Device::cpu_agent_ = {0};
std::vector<hsa_agent_t> roc::Device::gpu_agents_;
amd::Monitor* roc::Device::p2p_stage_ops_ = nullptr;
std::vector<Memory*> roc::Device::p2p_stages_;
const bool roc::Device::offlineDevice_ = false;
const bool roc::NullDevice::offlineDevice_ = true;
@@ -171,13 +169,14 @@ Device::~Device() {
delete mapCache_;
delete mapCacheOps_;
delete p2p_stage_ops_;
p2p_stage_ops_ = nullptr;
for (auto buf: p2p_stages_) {
delete buf;
if (nullptr != p2p_stage_) {
p2p_stage_->release();
p2p_stage_ = nullptr;
}
if (glb_ctx_ != nullptr) {
glb_ctx_->release();
glb_ctx_ = nullptr;
}
p2p_stages_.clear();
// Destroy temporary buffers for read/write
delete xferRead_;
@@ -713,16 +712,32 @@ bool Device::create(bool sramEccEnabled) {
// Use just 1 entry by default for the map cache
mapCache_->push_back(nullptr);
if (p2p_stage_ops_ == nullptr) {
p2p_stage_ops_ = new amd::Monitor("P2P Staging Lock", true);
if (nullptr == p2p_stage_ops_) {
return false;
if ((p2p_agents_.size() == 0) &&
(glb_ctx_ == nullptr) && (gpu_agents_.size() > 1) &&
// Allow creation for the last device in the list.
(gpu_agents_[gpu_agents_.size() - 1].handle == _bkendDevice.handle)) {
std::vector<amd::Device*> devices;
uint32_t numDevices = amd::Device::numDevices(CL_DEVICE_TYPE_GPU, false);
// Add all PAL devices
for (uint32_t i = 0; i < numDevices; ++i) {
devices.push_back(amd::Device::devices()[i]);
}
for (uint i = 0; i < 2; i++) {
Memory* buf = new Buffer(*this, kP2PStagingSize);
// Add current
devices.push_back(this);
if (devices.size() > 1) {
// Create a dummy context
glb_ctx_ = new amd::Context(devices, info);
if (glb_ctx_ == nullptr) {
return false;
}
amd::Buffer* buf =
new (GlbCtx()) amd::Buffer(GlbCtx(), CL_MEM_ALLOC_HOST_PTR, kP2PStagingSize);
if ((buf != nullptr) && buf->create()) {
p2p_stages_.push_back(buf);
} else {
p2p_stage_ = buf;
}
else {
delete buf;
return false;
}