| 1 |
On Tue, 22 Apr 2014 08:45:31 +0000 (UTC) |
| 2 |
Martin Vaeth <martin@×××××.de> wrote: |
| 3 |
|
| 4 |
> Ryan Hill <rhill@g.o> wrote: |
| 5 |
> > |
| 6 |
> > One thing I forgot to mention - LTO can also have detrimental effect on |
| 7 |
> > certain architectures. On some (eg. ppc), performance can actually |
| 8 |
> > be degraded due to increased register pressure. |
| 9 |
> |
| 10 |
> If this really is the case it is not the problem of LTO but |
| 11 |
> of the optimizer: If the optimizer really produces *worse* |
| 12 |
> code when he *can* see the full program instead of only parts of it, |
| 13 |
> something is severely broken in the optimizer. Only decreasing the |
| 14 |
> possibilities of the optimizer by removing LTO would be the wrong way |
| 15 |
> to "solve" this problem. |
| 16 |
|
| 17 |
Yes, this is a problem caused by aggressive inlining, and is being worked on |
| 18 |
upstream[1]. I meant that currently released versions exhibit this behaviour. |
| 19 |
|
| 20 |
|
| 21 |
[1] see for example http://gcc.gnu.org/ml/gcc-patches/2014-04/msg01098.html |
| 22 |
|
| 23 |
-- |
| 24 |
Ryan Hill psn: dirtyepic_sk |
| 25 |
gcc-porting/toolchain/wxwidgets @ gentoo.org |
| 26 |
|
| 27 |
47C3 6D62 4864 0E49 8E9E 7F92 ED38 BD49 957A 8463 |
Archives