[SYCL] Default work-group sizes based on max#952
Conversation
bader
left a comment
There was a problem hiding this comment.
Thanks for working on this fix.
Could you a regression test, please?
There was a problem hiding this comment.
Is this condition valid to represent all "default work-group size" cases?
There was a problem hiding this comment.
WGSize[0] == 0 means user hasn't asked for a specific local size using attribute((reqd_work_group_size(X, Y, Z)) only.
|
@StuartDAdams |
CONFLICT (content): Merge conflict in clang/lib/Frontend/CompilerInvocation.cpp
|
@StuartDAdams, ping. |
|
I have responded to some of the concerns and I agree with @AlexeySachkov that this may not be the most stable solution. I've suggested a different solution and will do regression tests once we reach a conclusion on what to do. |
|
@bader, @romanovvlad can you please review and comment on the latest approach. Please summarize what you think needs to be addressed. |
|
@romanovvlad please review |
9897d54 to
247634f
Compare
4439522 to
f61ebb5
Compare
|
@StuartDAdams, it looks like I merged some conflicting changes recently. Could you update the RP, please? |
Of course, I will do this ASAP. |
Signed-off-by: Stuart Adams <stuart.adams@codeplay.com>
f61ebb5 to
3044baf
Compare
Signed-off-by: Stuart Adams <stuart.adams@codeplay.com>
4be2092 to
646c65c
Compare
…hinx * upstream/sycl: (357 commits) [Support] Implement a simple tabular data management library (intel#1358) [Support] Implement a property set I/O library (intel#1357) [SYCL] Fix buffer constructor using iterators (intel#1386) [SYCL][FPGA] Enable a set of loop attributes (intel#1312) [Driver][SYCL][FPGA] Proper dependency output location when given /Fo<dir> (intel#1346) [SPIR-V] Enabling SPIR-V builtin lookup in device SYCL mode (intel#1384) [SYCL][NFC] Unify setting kernel arguments (intel#1379) [SYCL][Doc] First revision of standard layout relaxation extension (intel#1344) [SYCL] Fixed sub-buffer alloca search (intel#1385) [SYCL][FPGA] Emit multiple IR variants for the IVDep attribute (intel#1383) [SYCL] Add experimental flag to enable front-end optimizations (intel#1376) [SYCL] Remove unexpected double in complex SPIR-V for float support (intel#1381) [SYCL] Default work-group sizes based on max (intel#952) [SYCL][CUDA] Fix usage of multiple backends in the same program (intel#1252) [SPIR-V] Add SPIR-V builtin definitions to the builtin lookup. [SPIR-V] Add macro definition when -fdeclare-spirv-builtins is activated [SYCL] Fix sycl_generic printing [SYCL] Support intel::reqd_work_group_size (intel#1328) [SYCL][NFC] Make the RT::PiPlugin object private (intel#1375) [SPIRV] Add convergent attribute to SPIR-V built-ins (intel#1373) ...
These changes ensures that the default work-group size does not exceed the allowed maximum sizes.
Signed-off-by: Stuart Adams stuart.adams@codeplay.com