[KT] Get rid of excessive code in onesweep #1001
Closed
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Signed-off-by: Dmitriy Sobolev <dmitriy.sobolev@intel.com>
Signed-off-by: Dmitriy Sobolev <dmitriy.sobolev@intel.com>
Signed-off-by: Dmitriy Sobolev <dmitriy.sobolev@intel.com>
Signed-off-by: Dmitriy Sobolev <dmitriy.sobolev@intel.com>
Signed-off-by: Dmitriy Sobolev <dmitriy.sobolev@intel.com>
Signed-off-by: Dmitriy Sobolev <dmitriy.sobolev@intel.com>
Signed-off-by: Dmitriy Sobolev <dmitriy.sobolev@intel.com>
Signed-off-by: Dmitriy Sobolev <dmitriy.sobolev@intel.com>
Signed-off-by: Dmitriy Sobolev <dmitriy.sobolev@intel.com>
Signed-off-by: Dmitriy Sobolev <dmitriy.sobolev@intel.com>
Signed-off-by: Dmitriy Sobolev <dmitriy.sobolev@intel.com>
Signed-off-by: Dmitriy Sobolev <dmitriy.sobolev@intel.com>
Signed-off-by: Dmitriy Sobolev <dmitriy.sobolev@intel.com>
Use verification functions from the testing harness Signed-off-by: Dmitriy Sobolev <dmitriy.sobolev@intel.com>
Signed-off-by: Dmitriy Sobolev <dmitriy.sobolev@intel.com>
Signed-off-by: Dmitriy Sobolev <dmitriy.sobolev@intel.com>
Signed-off-by: Dmitriy Sobolev <dmitriy.sobolev@intel.com>
Signed-off-by: Dmitriy Sobolev <dmitriy.sobolev@intel.com>
…t_utils.h - fix enable_if usage
…e to build this test only dpcpp backend
* Add missed headers * Remove unnecessary 'inline' specifiers * Avoid unnecessary SFINAE * Align the function semantics (parameter order, types, and names) * Take offsets in the # of data elements, convert internally to bytes as needed Co-authored-by: Sergey Kopienko <sergey.kopienko@intel.com>
Also we made change in one_wg_kernel function:
- we replace block of code
#pragma unroll
for (uint32_t s = 0; s<PROCESS_SIZE; s+=64){
keys.template select<64, 1>(s) = lsc_slm_block_load<KeyT, 64>(slm_reorder_this_thread+s*sizeof(KeyT));
}
to
keys = utils::BlockLoad<KeyT, PROCESS_SIZE>(slm_reorder_this_thread);
to avoid compile error:
../include/sycl/ext/intel/experimental/esimd/memory.hpp:381:3: error: static assertion failed due to requirement '_DS == lsc_data_size::u32 || _DS == lsc_data_size::u64': Transposed load is supported only for data size u32 or u64
…output into console in test_all_view function
Signed-off-by: Dmitriy Sobolev <dmitriy.sobolev@intel.com>
Fix event dependency in onesweep Signed-off-by: Dmitriy Sobolev <dmitriy.sobolev@intel.com>
Do not corrupt memory outside of sync_buffer borders Signed-off-by: Dmitriy Sobolev <dmitriy.sobolev@intel.com>
Signed-off-by: Dmitriy Sobolev <dmitriy.sobolev@intel.com>
Signed-off-by: Dmitriy Sobolev <dmitriy.sobolev@intel.com>
Signed-off-by: Dmitriy Sobolev <dmitriy.sobolev@intel.com>
SergeyKopienko
force-pushed
the
dev/dmitriy-sobolev/esimd-radix-sort
branch
from
3251a7a
to
77639f5
Compare
* [KT] Test 8-bit types Thin out types to test due to long compilation time Get rid of 'k' suffix in constant names Disable sort of 8-bit types with onesweep --------- Signed-off-by: Dmitriy Sobolev <dmitriy.sobolev@intel.com>
dmitriy-sobolev
force-pushed
the
dev/dmitriy-sobolev/excessive-code
branch
from
0b3a6ce
to
a21ce36
Compare
Signed-off-by: Dmitriy Sobolev <dmitriy.sobolev@intel.com>
dmitriy-sobolev
force-pushed
the
dev/dmitriy-sobolev/excessive-code
branch
from
a21ce36
to
ccb7253
Compare
dmitriy-sobolev
force-pushed
the
dev/dmitriy-sobolev/esimd-radix-sort
branch
from
bfa6e39
to
90400aa
Compare
SergeyKopienko
force-pushed
the
dev/dmitriy-sobolev/esimd-radix-sort
branch
from
90400aa
to
da624f3
Compare
SergeyKopienko
force-pushed
the
dev/dmitriy-sobolev/esimd-radix-sort
branch
5 times, most recently
from
62a9672
to
9f9c1f7
Compare
|
This will be handled as a part of #1024 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Measurements show that this does not affect performance, at least on Intel® Data Center GPU Max. So the changes might be treated as refactoring.
Nontrivial changes:
ResetBinCountersis removed.RankSLMdoes the same, but without usingfencewhich is unnecessary as well.