Skip to main content

Mojo function

fp8_frag_to_smem

fp8_frag_to_smem[swizzle_mode: TensorMapSwizzle, stageN: Int, transpose_c: Bool, vec_dtype: DType, vec_size: Int](vec: InlineArray[Scalar[vec_dtype], vec_size], dst: TileTensor[dst.dtype, dst.LayoutType, dst.origin, address_space=AddressSpace.SHARED, linear_idx_type=dst.linear_idx_type, element_size=dst.element_size], warp_offset: UInt32 = 0)

Store fragment to SMEM via st.shared instruction.

Was this page helpful?