Mojo function

block_scaled_matmul

block_scaled_matmul[c_type: DType, a_type: DType, b_type: DType, scales_dtype: DType, //, *, SF_VECTOR_SIZE: Int, transpose_b: Bool = True, target: StringSlice[StaticConstantOrigin] = "cpu", elementwise_compute_lambda_fn: OptionalReg[fn[dtype: DType, width: Int, *, alignment: Int = 1](IndexList[2], SIMD[dtype, width]) capturing -> SIMD[dtype, width]] = None](c_device: NDBuffer[c_type, 2, MutAnyOrigin, shape], a_device: NDBuffer[a_type, 2, MutAnyOrigin, shape], b_device: NDBuffer[b_type, 2, MutAnyOrigin, shape], a_scales_device: NDBuffer[scales_dtype, 5, MutAnyOrigin, shape], b_scales_device: NDBuffer[scales_dtype, 5, MutAnyOrigin, shape], tensor_sf: Float32, ctx: DeviceContext)