1 |
You're wrong on the performance front. That is a sparc |
2 |
myth, and only applies to sparc. It is true that executables |
3 |
will generally consume more memory, and therefore have longer |
4 |
startup times, so for small apps that get run very often but |
5 |
rarely do anything at all, it could theorhetically be slower. |
6 |
But the fact is the performance numbers as a whole far outweigh |
7 |
the statistically few apps that would be penalized. Things like |
8 |
JAVA, SQL DBs, Mozilla, KDE/Gnome, POVRay, compilations, are |
9 |
all considerably faster in 64bit mode from my experience. |
10 |
|
11 |
Also, with that same argument you made, why do we have a amd64 |
12 |
port ?? Why don't we have just the stanard x86 port, that compiles |
13 |
a 64bit kernel, and has a 32bit userland. Oh and yes, I am also |
14 |
the lead for amd64 ... |
15 |
|
16 |
Basically trust me on this one, unless you have a machine that can |
17 |
prove it. Like I said, sparc's are different beasts (too many poor |
18 |
design choices IMHO) |
19 |
|
20 |
|
21 |
> Brad, |
22 |
> |
23 |
> Why does ppc64 need a 64bit userland? |
24 |
> |
25 |
> It would seem to me that you would not want a 64 bit userland at all. |
26 |
> From my understanding pure 64bit userlands tend to hurt performance |
27 |
> because twice as much data is going across the bus as needed. |
28 |
> Are -m32 -m64 not valid gcc/ld options for you? |
29 |
> |
30 |
> On Wed, 2004-01-21 at 08:50, Brad House wrote: |
31 |
>> We DEFINATELY need ppc64!!!! |
32 |
>> ppc64 is a 64bit kernel and 64bit userland |
33 |
>> What ppc does is a 32bit kernel and a 32bit userland |
34 |
>> What sparc64 does is a 64bit kernel and a 32bit userland |
35 |
>> What sparc does is a 32bit kernel and a 32bit userland |
36 |
>> See the main difference here, it's all in the userlands man! |
37 |
>> |
38 |
>> If anyone removes ppc64 keywords, I will personally lynch you. |
39 |
>> |
40 |
>> -Brad |
41 |
>> |
42 |
>> > maybe i'm lost with this but is there any reason we need ppc64 ? |
43 |
>> > couldnt something be done with ppc/ppc64 like the sparc/sparc64 merge |
44 |
>> ? |
45 |
>> > -mike |
46 |
>> > |
47 |
>> > |
48 |
>> > -- |
49 |
>> > gentoo-dev@g.o mailing list |
50 |
>> > |
51 |
>> > |
52 |
>> > |
53 |
>> |
54 |
>> |
55 |
>> -- |
56 |
>> gentoo-dev@g.o mailing list |
57 |
> -- |
58 |
> Ned Ludd <solar@g.o> |
59 |
> Gentoo Linux Developer |
60 |
> |
61 |
|
62 |
|
63 |
-- |
64 |
gentoo-dev@g.o mailing list |