Commit Graph

7 Commits

Author SHA1 Message Date
Jan Vesely 0a5aac3fc4 Provide vstore_half helper to workaround clc restrictions
clang won't accept half precision loads and stores without cl_khr_fp16 since r281904

llvm-svn: 282106
2016-09-21 20:15:55 +00:00
Jan Vesely ad8672727c Implement vstore_half{,n}
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 278962
2016-08-17 20:02:11 +00:00
Aaron Watry f991505d02 vload/vstore: Use casts instead of scalarizing everything in CLC version
This generates bitcode which is indistinguishable from what was
hand-written for int32 types in v[load|store]_impl.ll.

v4: Use vec2+scalar for vec3 load/stores to prevent corruption (per Tom)
v3: Also remove unused generic/lib/shared/v[load|store]_impl.ll
v2: (Per Matt Arsenault) Fix alignment issues with vector load stores

Signed-off-by: Aaron Watry <awatry@gmail.com>
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
CC: Matt Arsenault <Matthew.Arsenault@amd.com>
CC: Tom Stellard <thomas.stellard@amd.com>
llvm-svn: 216069
2014-08-20 13:58:57 +00:00
Aaron Watry 4cb7cf276d libclc: vload/vstore disable assembly and fix offset calculation
This commit gets us back to pure CLC and fixes offset calculations.

The next commit will re-enable the assembly implementation for R600,
fix bugs related to 64-bit address spaces, and also fix the
incorrect assumption that address space identifiers are the same in
all architectures.

llvm-svn: 186415
2013-07-16 14:28:58 +00:00
Tom Stellard d768ac0395 Add __CLC_ prefix to all macro definitions in headers
libclc was defining and undefing GENTYPE and several other macros with
common names in its header files.  This was preventing applications from
defining macros with identical names as command line arguments to the
compiler, because the definitions in the header files were masking the
macros defined as compiler arguements.

Reviewed-by: Aaron Watry <awatry@gmail.com>
llvm-svn: 185838
2013-07-08 17:27:02 +00:00
Tom Stellard 64b3bbae1e libclc: Add assembly versions of vstore for global [u]int4/8/16
The assembly should be generic, but at least currently R600 only supports
32-bit stores of [u]int1/4, and I believe that only global is well-supported.

R600 lowers the 8/16 component stores to multiple 4-component stores.

The unoptimized C versions of the other stuff is left in place.

Patch by: Aaron Watry

llvm-svn: 185009
2013-06-26 18:22:20 +00:00
Tom Stellard 51441f80c5 libclc: Initial vstore implementation
Assumes that the target supports byte-addressable stores.

Completely unoptimized.

Patch by: Aaron Watry

llvm-svn: 185007
2013-06-26 18:22:11 +00:00