Fix Computation for large batch sizes #73
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
…tware#16) * Updating README-sycl.md to capture the 3.5 modifications * Update README-sycl.md Co-authored-by: aacostadiaz <[email protected]> * Remove the sgemm_nt_1_sycl PoC (codeplaysoftware#15) * Remove sgemm_nt_1 PoC * Fix build issues * Fix code style format * Remove ENABLE_NVPTX flag * Update include/cute/util/debug.hpp Co-authored-by: Mehdi Goli <[email protected]> * Cosmetic --------- Co-authored-by: Mehdi Goli <[email protected]> * Applying the comments --------- Co-authored-by: aacostadiaz <[email protected]>
AD2605
requested review from
aacostadiaz and
muhammad-tanvir-1211
and removed request for
aacostadiaz
* Migrate Cute components to SYCL
* Add cmake configuration * Update examples/cute/tutorial/CMakeLists.txt Co-authored-by: Mehdi Goli <[email protected]> --------- Co-authored-by: Mehdi Goli <[email protected]>
* Update README-sycl.md Fixing CUDA version
Fix typo in Macro Co-authored-by: Mehdi Goli <[email protected]> * Cosmetic --------- Co-authored-by: Mehdi Goli <[email protected]> * Applying the comments --------- Co-authored-by: aacostadiaz <[email protected]> * Revert "Updating README-sycl.md to capture the 3.5 modifications (codeplaysoftware#16)" (codeplaysoftware#17) This reverts commit a726bd3. * fix typo in macro --------- Co-authored-by: Mehdi Goli <[email protected]> Co-authored-by: aacostadiaz <[email protected]>
Co-authored-by: Mehdi Goli <[email protected]>
Fix the calls to BlockDim* and GridDim* through SYCL. The current changes give incorrect output if you run a CUDA kernel through SYCL on NVIDIA A100. Co-authored-by: Mehdi Goli <[email protected]> --------- Co-authored-by: Mehdi Goli <[email protected]>
- Add block 2d load encapsulation
combination : block_y X block_x X (8 x 16) where block_y in {1, 2, 4} and block_x in {1, 2, 4}
and structure name follows the rule data_type x {8x16} x {block_y} x {block_x}_{V, N} and N means not-transposed , V means VNNI-packed
- include some sycl type
- include openCL extension intrinsic
---------
Co-authored-by: Mehdi Goli <[email protected]>
Co-authored-by: Atharva Dubey <[email protected]>
Co-authored-by: Muhammad Tanvir <[email protected]>
Removes the CUDA toolkit dependency when SYCL is enabled. --------- Co-authored-by: Mehdi Goli <[email protected]> Co-authored-by: Atharva Dubey <[email protected]>
Fix some issues when trying to build and run the 14_ampere example for SYCL. --------- Co-authored-by: Muhammad Tanvir <[email protected]>
Enables Cute Tests via SYCL Path --------- Co-authored-by: Mehdi Goli <[email protected]> Co-authored-by: aacostadiaz <[email protected]>
Add an Intel PVC pipeline to compute GEMM.
Add example for intel PVC GEMM
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.
Fixes computation on batch sizes greater than 16
#64 and #63 are already merged in this PR