- User Since
- May 4 2020, 11:17 AM (49 w, 17 h)
Thu, Apr 1
Wed, Mar 31
Mon, Mar 29
Fri, Mar 26
Wed, Mar 24
Tue, Mar 23
Adding enum for address space constant.
Changed the pass registration to happen before the Attributor pass to avoid function specialization. Added a PhaseOrdering test that won't pass without the module pass.
Forgot to format.
Fri, Mar 19
Remove command line argument and more unused runtime functions from clang.
Thu, Mar 18
Fixing test and formatting
Wed, Mar 17
Changing the test to simply check if we are in a non-SPMD kernel function. A more advanced approach can be used in the future.
Fixing tests and changing function interface back.
Tue, Mar 16
Adding context, forgot to specify the range.
Mar 12 2021
Changing this optimization to replace the globalization calls with shared memory. Removing them will be done by the attributor using HeapToStack once we add the allocation calls and improve the attributor.
Changed the RTL to have an argument that indicates if there is only one active caller for a team. This makes it easier to optimize.
Mar 11 2021
Mar 10 2021
Updating tests and adding the new function to the exports.
Mar 3 2021
Mar 2 2021
Mar 1 2021
Fixing tests is WIP
Feb 19 2021
Removing check for assert to keep the pass agnostic to the front-end and cleaning up the test file.
Feb 17 2021
Adding an extra check for multiple codes. I'll land it later today.
Adding test additional test case that checks for operands defined outside the
loop, and makes sure multiple remarks are not emitted for the same floating
point extension instruction.
Feb 16 2021
Pruning test file metadata.
Feb 11 2021
Feb 10 2021
Feb 3 2021
Feb 2 2021
Feb 1 2021
Jan 27 2021
Adding test. Changing store type check.
Making suggested changes. I'll add a test case later.
If anyone knows a better place to put this let me know. I couldn't find any existing infrastructure to run checks on the loop after its been vectorized.
Jan 26 2021
Jan 25 2021
Jan 22 2021
Jan 21 2021
Added a flag selector to the INFO macro.
Jan 20 2021
Adding a test case for the mapper names.
Changing the handling of OMP_TGT_MAPTYPE_PTR_AND_OBJ so it will not pass the same name for the base object. This doesn't pass in the correct name, simply makes it show up as unknown if queried.
Jan 15 2021
Jan 13 2021
Jan 12 2021
I tested it with a clean build using GCC 8.3 and CUDA 10.1, the resulting compiler successfuly built an offloading application that ran on the CUDA device. LGTM.
Jan 7 2021
Jan 6 2021
Changing the solution. The problem seems to be that after adjusting the debug info, the driver would change the debug kind if remarks were enabled. Now it adjusts the debug information after performing that change. This means that some diagnostics won't work with optimizations but it's necessary to compile correctly.
Jan 5 2021
Jan 4 2021
Dec 28 2020
Fixing a bug with passing a nullptr and making the kernel arguments print a different message for which type of region was executed.
Changing environment variables to initialize using std::call_once. Added a message when exiting an OpenMP data kernel.