11 Commits

Author	SHA1	Message	Date
Jens Steube	1b84a9e53b	Add missing backports from code base v6.2.2 ... Fix context to thread management Fix missing code in selftest.c, autotune.c, hashes.c, dispatch.c and backend.c Use IS_HIP depending code makes it easier for future optimization related to inline assembly calls - instead of using IS_CUDA || IS_HIP See TODO markers for more optimizations / next steps	2021-07-11 12:38:59 +02:00
Jens Steube	a22f8149fc	Merge branch 'HIP' into hip	2021-07-10 21:34:09 +02:00
reger-men	ea7b74389f	First draft HIP Version	2021-07-09 03:50:40 +00:00
Jens Steube	bb402b784a	Update module_unstable_warning for benchmark short selection on macOS for CPU and GPU; Allow use of GPU without --force testwise	2021-05-10 14:36:41 +02:00
Jens Steube	62fc3601bb	Wrap atomic functions with hc_ prefix to have better platform control	2021-04-20 17:47:44 +02:00
Jens Steube	316095c151	Some more ROCm performance tuning	2019-06-20 10:04:31 +02:00
Jens Steube	027af75a39	Fix rotate function names	2019-05-08 20:42:46 +02:00
Jens Steube	d0bd33c9d1	Rename CONSTANT_AS to CONSTANT_VK	2019-05-06 14:34:16 +02:00
Jens Steube	9faba41848	Use nvrtc to compile PTX (resulting PTX not yet used)	2019-04-26 13:28:44 +02:00
Jens Steube	4045e60021	Add nvrtc wrapper for later use	2019-04-26 10:03:16 +02:00
Jens Steube	4b986de5fb	Prepare native CUDA hybrid integration	2019-04-25 14:45:17 +02:00