[PATCH] D147835: [X86][mem-fold] Speed up test by not re-generating the .inc file

[Kan Shengchen via Phabricator via llvm-commits]

skan added a comment.

In D147835#4257774 <https://reviews.llvm.org/D147835#4257774>, @thakis wrote:

> In the GN build, the .inc file is at a different path.
>
> As far as I can tell, nothing in this discussion mentions how time this saves – is it a lot?
>
> I suppose we can add a dedicated `%var` for the path to the .inc line?
>
> (Another point: It's generally possible to build LLVM on one machine and copy just built binaries and the test/ folder and then run tests on another machine. This change makes that harder since you now also need to copy over that .inc file.)

I'm surprised that the path does not work for GN build. For curiosity, could I know where is it for GN build?  I didn't realize anyone have the requirement of building on one machine and testing on another...

On my system, running `llvm-tblgen -gen-x86-fold-tables -asmwriternum=1 llvm/lib/Target/X86/X86.td -I llvm/lib/Target/X86 -I llvm/include -o tmp` takes 14 seconds for DEBUG build, and after this change it takes 0 second. So it saves 14 seconds.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D147835/new/

https://reviews.llvm.org/D147835