varshav f4e3a4b1c3 AVX2 Support for BF16 Kernels - Bug fixes ...

- Added early return checks for A/B transpose cases and Column major
  support, as it is not currently supported.
- Enabled the JIT kernels for the Zen4 architecture.

AMD Internal: [SWLCSG - 3281]

Change-Id: Ie671676c51c739dd18709892414fd34d26a540df

2025-02-11 12:40:43 +05:30

..

config

Deprecate S16 LPGEMM APIs.

2025-02-07 11:43:28 +00:00

frame

Implemented reference unreorder bf16 function

2025-02-11 02:04:42 +00:00

JIT

Code cleanup: Copyright notices (2)

2025-02-07 05:41:44 -05:00

kernels

AVX2 Support for BF16 Kernels - Bug fixes

2025-02-11 12:40:43 +05:30

aocl_batch_gemm_bf16bf16f32of32.c

Fixed compilation errors for gcc versions < 11.2

2025-01-21 07:13:31 -05:00

aocl_batch_gemm_bf16s4f32of32.c

Fixed compilation errors for gcc versions < 11.2

2025-01-21 07:13:31 -05:00

aocl_batch_gemm_f32f32f32of32.c

Implemented batch_matmul for f32 & int8 datatypes

2025-01-10 04:10:53 -05:00

aocl_batch_gemm_s8s8s32os32.c

Added new Int8 batch_gemm APIs

2025-02-10 11:19:02 +00:00

aocl_batch_gemm_u8s8s32os32.c

Added new Int8 batch_gemm APIs

2025-02-10 11:19:02 +00:00

aocl_bf16_type.h

Code cleanup: Copyright notices

2024-08-05 15:35:08 -04:00

aocl_eltwise_ops_interface_apis.h

Element wise post-op APIs are upgraded with new post-ops

2025-02-10 01:06:39 -05:00

aocl_eltwise_ops.c

Element wise post-op APIs are upgraded with new post-ops

2025-02-10 01:06:39 -05:00

aocl_gemm_bf16_utils.c

Implemented reference unreorder bf16 function

2025-02-11 02:04:42 +00:00

aocl_gemm_bf16bf16f32obf16.c

Tiny GEMM path for BF16 LPGEMM API.

2025-02-07 04:37:11 -05:00

aocl_gemm_bf16bf16f32of32.c

AVX2 Support for BF16 Kernels - Bug fixes

2025-02-11 12:40:43 +05:30

aocl_gemm_bf16s4f32of32.c

Code cleanup: Copyright notices (2)

2025-02-07 05:41:44 -05:00

aocl_gemm_check.h

Fix compiler warning messages

2025-02-07 05:03:49 -05:00

aocl_gemm_f32f32f32of32_utils.c

Adding support for AOCL_ENABLE_INSTRUCTIONS for f32 LPGEMM API.

2024-10-30 08:52:22 +00:00

aocl_gemm_f32f32f32of32.c

Tiny GEMM path for BF16 LPGEMM API.

2025-02-07 04:37:11 -05:00

aocl_gemm_interface_apis.h

Implemented reference unreorder bf16 function

2025-02-11 02:04:42 +00:00

aocl_gemm_post_ops.h

Multi-data type buffer and scale support for matrix add|mul post-ops in s32 API.

2025-01-10 02:11:12 -05:00

aocl_gemm_s8s8s32obf16.c

Added F32 bias type support, F32, BF16 output type support in int8 APIs

2025-01-26 11:38:30 -05:00

aocl_gemm_s8s8s32of32.c

Added F32 bias type support, F32, BF16 output type support in int8 APIs

2025-01-26 11:38:30 -05:00

aocl_gemm_s8s8s32os8.c

Light-weight logging framewok for LPGEMM.

2025-01-03 11:28:57 +00:00

aocl_gemm_s8s8s32os32_utils.c

Added LPGEMV(n=1) kernels for s8s8s32os32|s8 and s8s8s16os16|s8 APIs

2024-08-09 06:17:52 -04:00

aocl_gemm_s8s8s32os32.c

Light-weight logging framewok for LPGEMM.

2025-01-03 11:28:57 +00:00

aocl_gemm_s8s8s32ou8.c

Updated all post-ops in s8s8s32 API to operate in float precision

2025-02-06 07:31:28 -05:00

aocl_gemm_u8s4s32os32_utils.c

Int4 B matrix reordering support in LPGEMM.

2024-06-24 07:55:34 -04:00

aocl_gemm_u8s8s32obf16.c

Added F32 bias type support, F32, BF16 output type support in int8 APIs

2025-01-26 11:38:30 -05:00

aocl_gemm_u8s8s32of32.c

Added F32 bias type support, F32, BF16 output type support in int8 APIs

2025-01-26 11:38:30 -05:00

aocl_gemm_u8s8s32os8.c

Returning early for col-major inputs in u8s8s32os32|s8 APIs

2025-01-05 23:46:06 +05:30

aocl_gemm_u8s8s32os32_utils.c

Added transB in all AVX512 based int8 API's

2024-05-23 03:46:13 +05:30

aocl_gemm_u8s8s32os32.c

Returning early for col-major inputs in u8s8s32os32|s8 APIs

2025-01-05 23:46:06 +05:30

aocl_gemm_u8s8s32ou8.c

Updated all post-ops in u8s8s32 API to operate in float precision

2025-01-29 04:21:17 -05:00

aocl_gemm.h

Deprecate S16 LPGEMM APIs.

2025-02-07 11:43:28 +00:00

aocl_util_interface_apis.h

Fixed compilation issue with clang 18 on windows

2024-12-12 06:37:06 -05:00

aocl_util_l1_ops.c

Code cleanup: Copyright notices

2024-08-05 15:35:08 -04:00