1 |
Alexis Ballier <aballier@g.o> wrote: |
2 |
>> >> |
3 |
>> >> More precisely: When changing the names anyway, |
4 |
>> >> IMHO it would be a very good idea to follow the convention of the |
5 |
>> >> "flag" names in /proc/cpuinfo and add all flags supported |
6 |
>> >> there as possible USE-flags, no matter, whether they are currently |
7 |
>> >> used in some package or not. |
8 |
>> > |
9 |
>> > seriously ? |
10 |
>> > |
11 |
>> > $ grep flags /proc/cpuinfo | head -n 1 | wc -w |
12 |
>> > 94 |
13 |
|
14 |
Well, currently, we have at least 10-20 already: |
15 |
At a very quick check (probably missing some), I found: |
16 |
|
17 |
mmx sse sse2 aes-ni ssse3 sse3 sse4 sse4a sse4.1 sse4.2 |
18 |
3des 3dnow 3dnowext |
19 |
|
20 |
and probably there are more to be added anyway |
21 |
in future packages. |
22 |
|
23 |
I am not even sure whether some of these sse4* are dupes |
24 |
or not since they do not follow /proc/cpuinfo convention. |
25 |
Some others do not follow this convention at all, |
26 |
e.g. "sse3" is called "pni" in /proc/cpuinfo |
27 |
or "aes-ni" is called "aes" in /proc/cpuinfo |
28 |
|
29 |
This is mainly what I meant: Currently, users have to look |
30 |
up how sse3 is called in /proc/cpuinfo to understand whether |
31 |
their processor has support for it. This could be automated. |
32 |
|
33 |
If not all of the 94 flags are included, it is perhaps not a drama; |
34 |
my main suggestion is to follow /proc/cpuinfo convention in the |
35 |
existing ones. |
36 |
|
37 |
> but rather about writing the proper .desc file |
38 |
|
39 |
Well, currently the descriptions are: |
40 |
|
41 |
3dnow: Use the 3DNow! instruction set |
42 |
mmx: Use the mmx instruction set |
43 |
sse: Use the SSE2 instruction set |
44 |
sse4_1: Enable sse4.1 acceleration |
45 |
... |
46 |
|
47 |
So they are practically useless unless you know anyway what the |
48 |
corresponding processor flags mean. |
49 |
I suggest to start with a "stub" .desc list, e.g.: |
50 |
|
51 |
yyy: Use the yyy CPU feature, cf. /proc/cpuinfo |
52 |
|
53 |
This is enough for the user to check on which of his systems he |
54 |
might want it. A more verbose description can be added for some |
55 |
flags some day as a bonus; such a list can grow over time, and some |
56 |
experts might want to send some patches to fix/extend the descriptions. |
57 |
But why require for the start more than we have currently? |
58 |
If there is no list to start with, then you can also not expect |
59 |
that somebody will suggest patches (I mean: to a non-listed feature...) |