Skip to main content

Unit 3.4.7 Using in-lined assembly code

Even more control over where what happens can be attained by using in-lined assembly code. This will allow prefetching and how registers are used to be made more explicit. (The compiler often changes the order of the operations even when intrinsics are used in C.) Details go beyond this course. You may want to look at the BLIS micro-kernel for the Haswell achitecture, implemented with in-lined assembly code.