| 1 |
On 9/26/05, Matt Randolph <mattr@×××××.com> wrote: |
| 2 |
> Mark Knecht wrote: |
| 3 |
> |
| 4 |
> > So, to check how this hardware platform works in a 32-bit mode I |
| 5 |
> >guess I need to do a complete second install on a separate part of the |
| 6 |
> >hard drive as an Athlon. That's a big job that I'm not anxious to do. |
| 7 |
> > |
| 8 |
> >Cheers, |
| 9 |
> >Mark |
| 10 |
> > |
| 11 |
> > |
| 12 |
> > |
| 13 |
> |
| 14 |
> If you're just trying to compare the performance of your software |
| 15 |
> between 32-bit and 64-bit kernels, you might try a distro that installs |
| 16 |
> more quickly. If I recall, you can install a complete 32-bit Debian |
| 17 |
> system in about half an hour using a Knoppix CD. Google `knoppix |
| 18 |
> "install to hard disk"` for the howto. I believe you can also get a |
| 19 |
> 64-bit Debian system just as quickly for comparison by following the |
| 20 |
> same procedure but using a Kanotix-64 CD instead. (You might just |
| 21 |
> compare against your existing 64-bit Gentoo system instead, but I don't |
| 22 |
> know how fair a fight would be between a hand-tuned Gentoo system and |
| 23 |
> one running a binary distro.) |
| 24 |
> |
| 25 |
> The only hitch I can think of is that some of your applications might |
| 26 |
> not be available as Debian binaries in both 32 and 64-bit versions. |
| 27 |
> Also, naturally, you won't have the ability to test your software with |
| 28 |
> the optimizations afforded by the USE flags you intend to use back here |
| 29 |
> in Gentoo-land. If you would be building your software from source |
| 30 |
> anyway, then this would not be an issue. |
| 31 |
> |
| 32 |
> That being said, I think this experiment would actually be pretty |
| 33 |
> academic for most users. Unless your software is, say, a very CPU |
| 34 |
> intensive scientific or multimedia application that was expertly ported |
| 35 |
> to 64-bit, it is my understanding that you're probably not going to see |
| 36 |
> a performance difference worth writing home about. I would encourage |
| 37 |
> anyone to comment if they know better as I would be delighted to learn |
| 38 |
> it was otherwise. |
| 39 |
> |
| 40 |
> - Matt |
| 41 |
> |
| 42 |
> -- |
| 43 |
> gentoo-amd64@g.o mailing list |
| 44 |
> |
| 45 |
> |
| 46 |
Hi Matt, |
| 47 |
As I stated in the first email: |
| 48 |
|
| 49 |
"Any disk activity seems to be the main cause of xruns on this machine." |
| 50 |
|
| 51 |
If I get xruns (a failure mode of Jack-Audio-Connection-Kit running |
| 52 |
on this hardware) then the machine is useless doing the sort of Linux |
| 53 |
audio work I want to do so this is the problem I'm trying to solve. |
| 54 |
That's my only goal. |
| 55 |
|
| 56 |
In my case I want to run Gentoo so I'm suspect of comparing any |
| 57 |
compiled distro with the results I'm getting on Gentoo's 64-bit |
| 58 |
kernel. I don't know what patches they've applied, etc., so I wouldn't |
| 59 |
know how to compare that. |
| 60 |
|
| 61 |
While it is a bunch of work, it might just be best to do a standard |
| 62 |
stage 3 Gentoo install on another section of the drive. This way I |
| 63 |
have a better comparison, and if it works as well as my other machines |
| 64 |
then at least I have a working platform until I can figure out what's |
| 65 |
causing the problem in the 64-bit environment. |
| 66 |
|
| 67 |
Thanks for your inputs. |
| 68 |
|
| 69 |
Cheers, |
| 70 |
Mark |
| 71 |
|
| 72 |
-- |
| 73 |
gentoo-amd64@g.o mailing list |
Archives