/home/docs/checkouts/readthedocs.org/user_builds/advanced-micro-devices-composable-kernel/checkouts/develop/include/ck/tensor_operation/gpu/warp/xdlops_gemm.hpp File Reference

• develop
• include
• ck
• tensor_operation
• gpu
• warp

Classes | Namespaces | Enumerations

xdlops_gemm.hpp File Reference

#include "ck/utility/common_header.hpp"
#include "ck/utility/math.hpp"
#include "ck/utility/amd_xdlops.hpp"
#include "ck/utility/amd_wmma.hpp"

Go to the source code of this file.

Classes
struct	ck::mfma_type< MfmaInstr::mfma_f32_32x32x1f32 >
struct	ck::mfma_type< MfmaInstr::mfma_f32_32x32x2f32 >
struct	ck::mfma_type< MfmaInstr::mfma_f32_16x16x4f32 >
struct	ck::mfma_type< MfmaInstr::mfma_f32_16x16x1f32 >
struct	ck::mfma_type< MfmaInstr::mfma_f32_4x4x1f32 >
struct	ck::mfma_type< MfmaInstr::mfma_f32_32x32x4f16 >
struct	ck::mfma_type< MfmaInstr::mfma_f32_32x32x8f16 >
struct	ck::mfma_type< MfmaInstr::mfma_f32_32x32x16f16 >
struct	ck::mfma_type< MfmaInstr::mfma_f32_16x16x32f16 >
struct	ck::mfma_type< MfmaInstr::mfma_f32_16x16x16f16 >
struct	ck::mfma_type< MfmaInstr::mfma_f32_16x16x4f16 >
struct	ck::mfma_type< MfmaInstr::mfma_f32_4x4x4f16 >
struct	ck::mfma_type< MfmaInstr::mfma_f32_32x32x16bf16 >
struct	ck::mfma_type< MfmaInstr::mfma_f32_32x32x8bf16_1k >
struct	ck::mfma_type< MfmaInstr::mfma_f32_16x16x32bf16 >
struct	ck::mfma_type< MfmaInstr::mfma_f32_16x16x16bf16_1k >
struct	ck::mfma_type< MfmaInstr::mfma_f32_32x32x4bf16 >
struct	ck::mfma_type< MfmaInstr::mfma_f32_16x16x8bf16 >
struct	ck::mfma_type< MfmaInstr::mfma_i32_32x32x8i8 >
struct	ck::mfma_type< MfmaInstr::mfma_i32_16x16x16i8 >
struct	ck::mfma_type< MfmaInstr::mfma_i32_32x32x16i8 >
struct	ck::mfma_type< MfmaInstr::mfma_i32_16x16x32i8 >
struct	ck::mfma_type< MfmaInstr::mfma_i32_32x32x32i8 >
struct	ck::mfma_type< MfmaInstr::mfma_i32_16x16x64i8 >
struct	ck::mfma_type< MfmaInstr::mfma_f64_16x16x4f64 >
struct	ck::mfma_type< MfmaInstr::mfma_f32_32x32x16f8f8 >
struct	ck::mfma_type< MfmaInstr::mfma_f32_16x16x32f8f8 >
struct	ck::mfma_type< MfmaInstr::mfma_f32_32x32x16bf8bf8 >
struct	ck::mfma_type< MfmaInstr::mfma_f32_16x16x32bf8bf8 >
struct	ck::mfma_type< MfmaInstr::mfma_f32_32x32x16f8bf8 >
struct	ck::mfma_type< MfmaInstr::mfma_f32_16x16x32f8bf8 >
struct	ck::mfma_type< MfmaInstr::mfma_f32_32x32x16bf8f8 >
struct	ck::mfma_type< MfmaInstr::mfma_f32_16x16x32bf8f8 >
struct	ck::mfma_type< MfmaInstr::mfma_f32_32x32x64f8f6f4 >
struct	ck::mfma_type< MfmaInstr::mfma_f32_16x16x128f8f6f4 >
struct	ck::mfma_type< MfmaInstr::mfma_scale_f32_32x32x64f8f6f4 >
struct	ck::mfma_type< MfmaInstr::mfma_scale_f32_16x16x128f8f6f4 >
struct	ck::mfma_type< MfmaInstr::mfma_f32_16x16x8xf32 >
struct	ck::mfma_type< MfmaInstr::mfma_f32_32x32x4xf32 >
struct	ck::mfma_type_gfx11_base
struct	ck::mfma_type< MfmaInstr::wmma_f32_16x16x16_f16 >
struct	ck::mfma_type< MfmaInstr::wmma_f32_16x16x16_bf16 >
struct	ck::mfma_type< MfmaInstr::wmma_i32_16x16x16_iu8 >
struct	ck::mfma_type< MfmaInstr::wmma_unsupport_16x16_gfx11 >
struct	ck::mfma_type_gfx12_base
struct	ck::mfma_type< MfmaInstr::wmma_f32_16x16x16_f16_gfx12 >
struct	ck::mfma_type< MfmaInstr::wmma_f32_16x16x16_bf16_gfx12 >
struct	ck::mfma_type< MfmaInstr::wmma_i32_16x16x16_iu8_gfx12 >
struct	ck::mfma_type< MfmaInstr::wmma_f32_16x16x16_f8f8_gfx12 >
struct	ck::mfma_type< MfmaInstr::wmma_f32_16x16x16_f8bf8_gfx12 >
struct	ck::mfma_type< MfmaInstr::wmma_f32_16x16x16_bf8f8_gfx12 >
struct	ck::mfma_type< MfmaInstr::wmma_f32_16x16x16_bf8bf8_gfx12 >
struct	ck::mfma_type< MfmaInstr::wmma_unsupport_16x16_gfx12 >
class	ck::MfmaSelector< base_type, MPerXdlops, NPerXdlops, additional_type, is_single_rate_mfma, is_scale_mfma >
	Selects the appropriate MFMA instruction type and configuration for given data types and tile sizes on AMD GPUs. More...
struct	ck::XdlopsGemm< base_type, MPerXdlops, NPerXdlops, KPack, additional_type, TransposeC, is_scale_mfma >

Namespaces
	ck

Enumerations
enum class	ck::MfmaInstr {   ck::mfma_f32_32x32x1f32 = 0 ,   ck::mfma_f32_16x16x1f32 ,   ck::mfma_f32_4x4x1f32 ,   ck::mfma_f32_32x32x2f32 ,   ck::mfma_f32_16x16x4f32 ,   ck::mfma_f32_32x32x4f16 ,   ck::mfma_f32_16x16x4f16 ,   ck::mfma_f32_4x4x4f16 ,   ck::mfma_f32_32x32x8f16 ,   ck::mfma_f32_16x16x16f16 ,   ck::mfma_f32_32x32x8bf16_1k ,   ck::mfma_f32_16x16x16bf16_1k ,   ck::mfma_f32_32x32x4bf16 ,   ck::mfma_f32_16x16x8bf16 ,   ck::mfma_i32_32x32x8i8 ,   ck::mfma_i32_16x16x16i8 ,   ck::mfma_i32_32x32x16i8 ,   ck::mfma_i32_16x16x32i8 ,   ck::mfma_f64_16x16x4f64 ,   ck::mfma_f32_32x32x16f8f8 ,   ck::mfma_f32_16x16x32f8f8 ,   ck::mfma_f32_32x32x16bf8bf8 ,   ck::mfma_f32_16x16x32bf8bf8 ,   ck::mfma_f32_32x32x16f8bf8 ,   ck::mfma_f32_16x16x32f8bf8 ,   ck::mfma_f32_32x32x16bf8f8 ,   ck::mfma_f32_16x16x32bf8f8 ,   ck::mfma_f32_32x32x16f16 ,   ck::mfma_f32_16x16x32f16 ,   ck::mfma_f32_32x32x16bf16 ,   ck::mfma_f32_16x16x32bf16 ,   ck::mfma_i32_32x32x32i8 ,   ck::mfma_i32_16x16x64i8 ,   ck::mfma_f32_32x32x64f8f6f4 ,   ck::mfma_f32_16x16x128f8f6f4 ,   ck::mfma_scale_f32_32x32x64f8f6f4 ,   ck::mfma_scale_f32_16x16x128f8f6f4 ,   ck::mfma_f32_16x16x8xf32 ,   ck::mfma_f32_32x32x4xf32 ,   ck::wmma_f32_16x16x16_f16 ,   ck::wmma_f32_16x16x16_bf16 ,   ck::wmma_i32_16x16x16_iu8 ,   ck::wmma_unsupport_16x16_gfx11 ,   ck::wmma_f32_16x16x16_f16_gfx12 ,   ck::wmma_f32_16x16x16_bf16_gfx12 ,   ck::wmma_i32_16x16x16_iu8_gfx12 ,   ck::wmma_f32_16x16x16_f8f8_gfx12 ,   ck::wmma_f32_16x16x16_f8bf8_gfx12 ,   ck::wmma_f32_16x16x16_bf8f8_gfx12 ,   ck::wmma_f32_16x16x16_bf8bf8_gfx12 ,   ck::wmma_unsupport_16x16_gfx12 }