[prev in list] [next in list] [prev in thread] [next in thread]
List: llvm-commits
Subject: Re: [PATCH] D11662: Filter cold blocks off the loop chain when profile data is available.
From: Xinliang David Li <davidxl () google ! com>
Date: 2015-07-31 21:44:22
Message-ID: CAAkRFZ+u7Zyyb7tu0AVffPD0s=k1D+=5P6nGmNitEeMXpL8Qfg () mail ! gmail ! com
[Download RAW message or body]
I suggest making it an option. The other usage is slightly different
-- we can leave it out for now (if needed can be changed later).
David
On Fri, Jul 31, 2015 at 2:41 PM, Cong Hou <congh@google.com> wrote:
> congh added inline comments.
>
> ================
> Comment at: lib/CodeGen/MachineBlockPlacement.cpp:826
> @@ +825,3 @@
> + // We use the threshold 20% here to filter cold blocks.
> + const unsigned LoopToColdBBRatio = 5;
> + for (MachineBasicBlock *LoopBB : L.getBlocks()) {
> ----------------
> davidxl wrote:
> > Try to avoid hard-coding parameter like this.
> Should I make it a global constant variable or an option? The other 20% probability \
> used in selectBestSuccessor() is also hard-coded and we could use the same variable \
> to replace them.
>
> http://reviews.llvm.org/D11662
>
>
>
_______________________________________________
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
[prev in list] [next in list] [prev in thread] [next in thread]
Configure |
About |
News |
Add a list |
Sponsored by KoreLogic