1 |
"Duncan" <1i5t5.duncan@×××.net> posted efqqvt$sqh$6@×××××××××.org, |
2 |
excerpted below, on Mon, 02 Oct 2006 10:51:41 +0000: |
3 |
|
4 |
> I'll probably have a bit more to say about it after checking the |
5 |
> tinyurls, which at this point I don't even know where they go yet, let |
6 |
> alone their content. |
7 |
> |
8 |
> <quote from Ryan Hill> |
9 |
> |
10 |
> Duncan wrote: |
11 |
> |
12 |
>> Could you point me at some info on this one (-ftree-vectorize)? It |
13 |
>> came up on the amd64 list a week or so ago[] |
14 |
> |
15 |
> http://tinyurl.com/l75we |
16 |
|
17 |
OK, this one's the gcc bugtracker list of -ftree-vectorize bugs. Lots of |
18 |
reports, many of them reduced to skeleton programs with just the |
19 |
triggering code. Of note, /most/ of the issues involve x86/ppc and/or size |
20 |
increasing loop optimizations that aren't of interest to those whose |
21 |
strategy targets -Os in any case. For example, one of the triggers seems |
22 |
to be -funswitch-loops, enabled by default with -O3, another is |
23 |
-funroll-loops, neither of which (again) are of interest to those |
24 |
targeting size for cache efficiency among other reasons. |
25 |
|
26 |
> See http://tinyurl.com/rt3aa for some real-world examples. |
27 |
|
28 |
This one's the Gentoo bugtracker -ftree-vectorize related bug list. |
29 |
Again, most are x86/ppc and/or involve flags folks interested in -Os |
30 |
won't be worrying about. |
31 |
|
32 |
Looks like on amd64 it may be less trouble than -combine is, and Flameeyes |
33 |
said he has been using it for awhile, so it looks like it'll be standard |
34 |
in my cflags now, to be removed only where it seems to cause problems, |
35 |
which shouldn't be that often, hopefully. |
36 |
|
37 |
-- |
38 |
Duncan - List replies preferred. No HTML msgs. |
39 |
"Every nonfree program has a lord, a master -- |
40 |
and if you use the program, he is your master." Richard Stallman |
41 |
|
42 |
-- |
43 |
gentoo-amd64@g.o mailing list |