Please use GitHub pull requests for new patches. Avoid migrating existing patches. Phabricator shutdown timeline
Changeset View
Changeset View
Standalone View
Standalone View
openmp/libomptarget/plugins-nextgen/amdgpu/src/rtl.cpp
| Show First 20 Lines • Show All 1,839 Lines • ▼ Show 20 Lines | struct AMDGPUDeviceTy : public GenericDeviceTy, AMDGenericDeviceTy { | ||||||||
| } | } | ||||||||
| /// Unpin the host buffer. | /// Unpin the host buffer. | ||||||||
| Error dataUnlockImpl(void *HstPtr) override { | Error dataUnlockImpl(void *HstPtr) override { | ||||||||
| hsa_status_t Status = hsa_amd_memory_unlock(HstPtr); | hsa_status_t Status = hsa_amd_memory_unlock(HstPtr); | ||||||||
| return Plugin::check(Status, "Error in hsa_amd_memory_unlock: %s\n"); | return Plugin::check(Status, "Error in hsa_amd_memory_unlock: %s\n"); | ||||||||
| } | } | ||||||||
| /// Check through the HSA runtime whether the \p HstPtr buffer is pinned. | |||||||||
| Expected<bool> isPinnedPtrImpl(void *HstPtr, void *&BaseHstPtr, | |||||||||
| void *&BaseDevAccessiblePtr, | |||||||||
| size_t &BaseSize) const override { | |||||||||
| hsa_amd_pointer_info_t Info; | |||||||||
| Info.size = sizeof(hsa_amd_pointer_info_t); | |||||||||
| hsa_status_t Status = | |||||||||
jdoerfertUnsubmitted Done ReplyInline Actions
jdoerfert: | |||||||||
It's an output parameter. I'll add the proper comment. kevinsala: It's an output parameter. I'll add the proper comment. | |||||||||
| hsa_amd_pointer_info(HstPtr, &Info, /* Allocator */ nullptr, | |||||||||
| /* Number of accessible agents (out) */ nullptr, | |||||||||
| /* Accessible agents */ nullptr); | |||||||||
| if (auto Err = Plugin::check(Status, "Error in hsa_amd_pointer_info: %s")) | |||||||||
| return Err; | |||||||||
| // The buffer may be locked or allocated through HSA allocators. Assume that | |||||||||
| // the buffer is host pinned if the runtime reports a HSA type. | |||||||||
| if (Info.type != HSA_EXT_POINTER_TYPE_LOCKED && | |||||||||
| Info.type != HSA_EXT_POINTER_TYPE_HSA) | |||||||||
| return false; | |||||||||
| assert(Info.hostBaseAddress && "Invalid host pinned address"); | |||||||||
| assert(Info.agentBaseAddress && "Invalid agent pinned address"); | |||||||||
| assert(Info.sizeInBytes > 0 && "Invalid pinned allocation size"); | |||||||||
| // Save the allocation info in the output parameters. | |||||||||
| BaseHstPtr = Info.hostBaseAddress; | |||||||||
| BaseDevAccessiblePtr = Info.agentBaseAddress; | |||||||||
| BaseSize = Info.sizeInBytes; | |||||||||
| return true; | |||||||||
| } | |||||||||
| /// Submit data to the device (host to device transfer). | /// Submit data to the device (host to device transfer). | ||||||||
| Error dataSubmitImpl(void *TgtPtr, const void *HstPtr, int64_t Size, | Error dataSubmitImpl(void *TgtPtr, const void *HstPtr, int64_t Size, | ||||||||
| AsyncInfoWrapperTy &AsyncInfoWrapper) override { | AsyncInfoWrapperTy &AsyncInfoWrapper) override { | ||||||||
| // Use one-step asynchronous operation when host memory is already pinned. | // Use one-step asynchronous operation when host memory is already pinned. | ||||||||
| if (void *PinnedPtr = | if (void *PinnedPtr = | ||||||||
| PinnedAllocs.getDeviceAccessiblePtrFromPinnedBuffer(HstPtr)) { | PinnedAllocs.getDeviceAccessiblePtrFromPinnedBuffer(HstPtr)) { | ||||||||
| AMDGPUStreamTy &Stream = getStream(AsyncInfoWrapper); | AMDGPUStreamTy &Stream = getStream(AsyncInfoWrapper); | ||||||||
| return Stream.pushPinnedMemoryCopyAsync(TgtPtr, PinnedPtr, Size); | return Stream.pushPinnedMemoryCopyAsync(TgtPtr, PinnedPtr, Size); | ||||||||
| ▲ Show 20 Lines • Show All 778 Lines • Show Last 20 Lines | |||||||||