Mojo function

store_fragment_to_smem

store_fragment_to_smem[vec_dtype: DType, vec_size: Int, //, swizzle: Swizzle, stageN: Int, transpose_c: Bool = False, c_swizzle: TensorMapSwizzle = TensorMapSwizzle.SWIZZLE_128B](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 = UInt32(0))

Store fragment to SMEM via st.matrix instruction for bf16 output type and st.shared instruction for FP8 output type.

Was this page helpful?

Thank you! We'll create more content like this.

Thank you for helping us improve!

😔 What went wrong?

Some code doesn’t work

It includes inaccurate information

It's missing information I need

It was difficult to understand

Other

Submit