Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[SYCL][Joint Matrix] Test combinations are queried Part 4 #14019

Merged
merged 5 commits into from
Jun 6, 2024
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
13 changes: 5 additions & 8 deletions sycl/test-e2e/Matrix/SG32/joint_matrix_bf16_fill_k_cache.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -5,21 +5,18 @@
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
// REQUIRES: matrix
// SG size = 32 is not currently supported for SYCL Joint Matrix by IGC on DG2
// UNSUPPORTED: gpu-intel-dg2
// REQUIRES: aspect-ext_intel_matrix
// REQUIRES-INTEL-DRIVER: lin: 27501, win: 101.4943

// RUN: %{build} -o %t_gpu.out -ffp-model=precise
// RUN: %if gpu %{ %{run} %t_gpu.out %}

// RUN: %{build} -ffp-model=precise -o %t_cpu.out -DtM=16 -DtK=32 -DNCACHE1=32 -DKCACHE1=32
// RUN: %if cpu %{ %{run} %t_cpu.out %}
// RUN: %{build} -o %t.out -ffp-model=precise
// RUN: %{run} %t.out

// -ffp-model=precise is added to not depend on compiler defaults.

#include "../common.hpp"
#include <cstddef>

#define SG_SZ 32
constexpr size_t TN = 16;

#include "../joint_matrix_bf16_fill_k_cache_impl.hpp"
Original file line number Diff line number Diff line change
Expand Up @@ -5,7 +5,9 @@
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
// REQUIRES: matrix, gpu
// SG size = 32 is not currently supported for SYCL Joint Matrix by IGC on DG2
// UNSUPPORTED: gpu-intel-dg2
// REQUIRES: aspect-ext_intel_matrix, gpu
// REQUIRES-INTEL-DRIVER: lin: 27501, win: 101.4943

// RUN: %{build} -o %t.out -DINIT_LIST -ffp-model=precise
Expand All @@ -14,9 +16,7 @@
// -ffp-model=precise is added to not depend on compiler defaults.

#include "../common.hpp"
#include <cstddef>

#define SG_SZ 32
constexpr size_t TN = 16;

#include "../joint_matrix_bf16_fill_k_cache_impl.hpp"
Original file line number Diff line number Diff line change
Expand Up @@ -5,23 +5,20 @@
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
// REQUIRES: matrix
// SG size = 32 is not currently supported for SYCL Joint Matrix by IGC on DG2
// UNSUPPORTED: gpu-intel-dg2
// REQUIRES: aspect-ext_intel_matrix
// REQUIRES-INTEL-DRIVER: lin: 27501, win: 101.4943

// RUN: %{build} -mllvm -inline-threshold=5000 -ffp-model=precise -o %t_gpu.out -DMANUAL_UNROLL
// RUN: %if gpu %{ %{run} %t_gpu.out %}

// RUN: %{build} -mllvm -inline-threshold=5000 -ffp-model=precise -o %t_cpu.out -DMANUAL_UNROLL -DtM=16 -DtK=32 -DNCACHE1=32 -DKCACHE1=32
// RUN: %if cpu %{ %{run} %t_cpu.out %}
// RUN: %{build} -mllvm -inline-threshold=5000 -ffp-model=precise -o %t.out -DMANUAL_UNROLL
// RUN: %{run} %t.out

// -mllvm -inline-threshold added as a workaround,
// since IGC doesn't support some variants of IR for Joint Matrix currently
// -ffp-model=precise is added to not depend on compiler defaults.

#include "../common.hpp"
#include <cstddef>

#define SG_SZ 32
constexpr size_t TN = 16;

#include "../joint_matrix_bf16_fill_k_cache_impl.hpp"
Original file line number Diff line number Diff line change
Expand Up @@ -5,7 +5,9 @@
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
// REQUIRES: matrix, gpu
// SG size = 32 is not currently supported for SYCL Joint Matrix by IGC on DG2
// UNSUPPORTED: gpu-intel-dg2
// REQUIRES: aspect-ext_intel_matrix, gpu
// REQUIRES-INTEL-DRIVER: lin: 27501, win: 101.4943

// RUN: %{build} -mllvm -inline-threshold=5000 -ffp-model=precise -o %t_gpu.out -DINIT_LIST -DMANUAL_UNROLL
Expand All @@ -16,9 +18,7 @@
// -ffp-model=precise is added to not depend on compiler defaults.

#include "../common.hpp"
#include <cstddef>

#define SG_SZ 32
constexpr size_t TN = 16;

#include "../joint_matrix_bf16_fill_k_cache_impl.hpp"
20 changes: 0 additions & 20 deletions sycl/test-e2e/Matrix/XMX8/joint_matrix_bf16_fill_k_cache.cpp

This file was deleted.

This file was deleted.

This file was deleted.

This file was deleted.

19 changes: 0 additions & 19 deletions sycl/test-e2e/Matrix/XMX8/joint_matrix_colA_rowB_colC.cpp

This file was deleted.

20 changes: 0 additions & 20 deletions sycl/test-e2e/Matrix/XMX8/joint_matrix_out_bounds.cpp

This file was deleted.

20 changes: 0 additions & 20 deletions sycl/test-e2e/Matrix/XMX8/joint_matrix_unaligned_k.cpp

This file was deleted.

13 changes: 3 additions & 10 deletions sycl/test-e2e/Matrix/joint_matrix_bf16_fill_k_cache.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -5,19 +5,12 @@
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
// REQUIRES: matrix
// REQUIRES: aspect-ext_intel_matrix

// RUN: %{build} -o %t_gpu.out -ffp-model=precise
// RUN: %if gpu %{ %{run} %t_gpu.out %}

// RUN: %{build} -ffp-model=precise -o %t_cpu.out -DtM=16 -DtK=32 -DNCACHE1=32 -DKCACHE1=32
// RUN: %if cpu %{ %{run} %t_cpu.out %}
// RUN: %{build} -o %t.out -ffp-model=precise
// RUN: %{run} %t.out

// -ffp-model=precise is added to not depend on compiler defaults.

#include "common.hpp"
#include <cstddef>

constexpr size_t TN = 16;

#include "joint_matrix_bf16_fill_k_cache_impl.hpp"
5 changes: 0 additions & 5 deletions sycl/test-e2e/Matrix/joint_matrix_bf16_fill_k_cache_OOB.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -16,9 +16,4 @@
// -ffp-model=precise is added to not depend on compiler defaults.

#include "common.hpp"
#include <cstddef>

#define SG_SZ 16
constexpr size_t TN = 16;

#include "joint_matrix_bf16_fill_k_cache_impl.hpp"
Loading
Loading