Tags: maximdimakov/llvm
Tags
[SYCL][NFC] Set appropriate targets for SYCL mode tests when needed (i… …ntel#2750) Some tests require updating to fix target architecture differences. The tests are expecting 64-bit targets, but under some conditions the default architecture would not match (e.g. running a 32-bit target compiler)
[SYCL] Fix link after directory structure change (intel#2741)
[SYCL][NFC] Extend ABI tests to cover device code (intel#2725) * [SYCL][NFC] Extend ABI tests to cover device code
[SYCL][Test] Remove leftovers for FPGA archives (intel#2735) As with [intel/llvm@750c979], remove the already-existing archive file to ensure correct behavior of the LIT in local environments. Signed-off-by: Artem Gindinson <artem.gindinson@intel.com>
[SYCL][PI][L0] Fix a problem with kernels and programs destruction wh… …ile they can be used (intel#2710) Requirements from summary of level zero spec: 1 – cannot do zeKernelDestroy until all commands using the kernel have completed execution as measured by their event or fence being signaled. 2 – The application must destroy all kernel and build log handles created from the module before destroying the module itself (zeModuleDestroy) So, here is a design I think will work. 1 – We want to do a piKernelRetain sometime before kernel is appended to command-list in piEnqueueKernelLaunch. The intent is that we will do a piKernelRelease once the Event that we create to be associated with the Enqueued Kernel is signaled/released. The handle of kernel to be released will be stored in the event. 2 – Event currently contains fields for CommandType and CommandData, and for the CommandType that piEnqueueKernelLaunch uses, CommandData is currently unused. So, for this case, we will set CommandData to be the Kernel, so then when the event is signaled as done, then we can use the kernel we stored in CommandData, we can do piKernelRelease(CommandData). 3 – I think in piKernelRetain, we also need to call piProgramRetain on the program that the kernel is part of, and we need piKernelRelease to call piProgramRelease. This is necessary to make sure we do not destroy the program before destroying all the kernels that may be in use from the program. 4 – in the destructor for pi_program, we need to destroy the build log before we destroy the program itself.
[Driver][SYCL] Refactor -fsycl-device-only behaviors (intel#2713) Update -fsycl-device-only to go through the offload path, effectively removing the host dependencies when it is used. This allows for a cleaner representation of SYCL device specific compilation needs.
[SYCL] Clarify --system-ocl key in GetStartedGuide (intel#2712)
[NFC][SYCL] Use visitor to emit forward declarations (intel#2670)
[Driver][SYCL] Remove real_path workaround for integration header (in… …tel#2689) We applied a workaround for integration header usage on Windows that corrected the directory name in case it did not match expectations of the file system (lowercase/uppercase like things). The workaround is not needed anymore as community has addressed the non-portable path warning directly for Windows.
PreviousNext