Skip to main content

Mojo function

multistage_dual_gemm_kernel

multistage_dual_gemm_kernel[c_type: DType, c_layout: Layout, a_type: DType, a_layout: Layout, b_type: DType, b_layout: Layout, transpose_b: Bool, config: MatmulConfig[a_type, b_type, c_type, transpose_b], binary_lambda_fn: fn[DType, Int](SIMD[$0, $1], SIMD[$0, $1]) -> SIMD[$0, $1], elementwise_lambda_fn: OptionalReg[fn[DType, Int, Int](IndexList[2], SIMD[$0, $1]) capturing -> None] = OptionalReg[fn[DType, Int, Int](IndexList[2], SIMD[$0, $1]) capturing -> None]({:i1 0, 1})](c: LayoutTensor[c_type, c_layout, MutableAnyOrigin], a: LayoutTensor[a_type, a_layout, MutableAnyOrigin], b0: LayoutTensor[b_type, b_layout, MutableAnyOrigin], b1: LayoutTensor[b_type, b_layout, MutableAnyOrigin])

Was this page helpful?