[LLVMdev] Any LLVM passes for integers type compaction?

[Dmitry Mikushin]

Dear Henrique,

Thanks for reply, I think I saw this project before. We have an LLVM-based
GPU toolchain, where variable size is usually one of the major performance
factors.

- D.

2013/5/7 Henrique Santos <hnsantos at gmx.com>

> A few colleagues of mine implemented an inter-procedural range analysis
> algorithm for LLVM 2.7 a few years back.
> Their paper (see
> http://homepages.dcc.ufmg.br/~fernando/publications/papers/SBLP12_victor.pdf)
> states a 13-14% reduction in variable size
> for the SPEC CPU2006 benchmarks and 36% for the Stanford benchmarks.
> Check out http://code.google.com/p/range-analysis/ if you're interested
> and feel free to email me if you have any questions.
>
> H.
>
>
> ----- Original Message -----
>
> From: Dmitry Mikushin
>
> Sent: 05/07/13 02:48 AM
>
> To: LLVM Developers Mailing List
>
> Subject: [LLVMdev] Any LLVM passes for integers type compaction?
>
>
> Dear all,
>
> I'm wondering if anybody is working on optimization passes that could use
> smaller integer types for variables that are known to have values within
> smaller type range?
>
> Thanks,
> - D.
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20130507/4ce06c63/attachment.html>