unified-runtime
unified-runtime copied to clipboard
oneapi-src
LLVM: https://github.com/intel/llvm/pull/13450 - Check GPU type by device name - Added Shadow memory allocation and MemToShadow for DG2
Add debug logging to the Level Zero structs used when a command-buffer kernel command is updated. DPC++ PR https://github.com/intel/llvm/pull/13892
(This is currently a WIP) TODO: Add description and explanation of the changes in this refactor. **TODO:** Revisit this making a simpler helper `getUSMHostOrDevicePtr` function internal to `enqueue.cpp`, turning the...
Two main changes to how `Kernel/ProgramSetSpecializationConstants` are handled: * They may now output either `INVALID_VALUE` or the new `INVALID_SPEC_ID` when the provided list is invalid. * The OpenCL and level...
This is mostly just a copy of the CUDA version of this implementation.
https://github.com/intel/llvm/pull/13876
A context might only contain a subset of the devices in a platform. Each `ur_mem_handle_t_` contains some vectors of length `ctx.NumDevices`. `Device->getIndex()` will return the index of a device in...
This PR fixes some new coverity warnings reported in the Native CPU adapter: * `mdNode` is defined with `const auto&` to avoid copies * hProgram uses `unique_ptr` to avoid leaks...
The Level Zero adapter implementing `urCommandBufferUpdateKernelLaunchExp` is doing a blocking host wait with `zeEventHostSynchronize` on the executing of the command-buffer. However, there is no guarantee that the command-buffer has been...
This PR fixes some tests on DG2. * update level-zero loader tag to v1.17.0 which contains definition of ZE_IMAGE_BINDLESS_EXP_FLAG_SAMPLED_IMAGE * Use L0 ImageCopy ext API for copying to/from non-usm *...
