Hi,
This small series makes use of new LLVM 8 buffer intrinsics.
No CTS regressions on GFX8 with LLVM 7, 8 and master.
v3: use different names (ie. struct vs raw) for IDXEN
V2: fix use of IDXEN for GFX9
Please review,
Thanks!
Samuel Pitoiset (11):
ac: make use of ac_get_store_intr_attribs() where possible
ac/nir: set attrib flags for SSBO and image store operations
ac: add ac_build_buffer_store_format() helper
ac/nir: remove one useless check in visit_store_ssbo()
ac/nir: use new LLVM 8 intrinsics for SSBO atomic operations
ac/nir: use ac_build_buffer_load() for SSBO load operations
ac/nir: use ac_build_buffer_store_dword() for SSBO store operations
ac: use new LLVM 8 intrinsics in ac_build_buffer_load()
ac: add ac_build_{struct,raw}_tbuffer_store() helpers
ac: use new LLVM 8 intrinsic when storing 16-bit values
ac: use new LLVM 8 intrinsics in ac_build_buffer_store_dword()
src/amd/common/ac_llvm_build.c | 325 +++++++++++++++++++++++++++-----
src/amd/common/ac_llvm_build.h | 49 +++++
src/amd/common/ac_nir_to_llvm.c | 184 +++++++-----------
3 files changed, 402 insertions(+), 156 deletions(-)
--
2.21.0
_______________________________________________
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev
