[svlug] motherboard recommendations

Ivan Sergio Borgonovo mail at webthatworks.it
Wed Mar 22 00:27:27 PST 2006


On Tue, 21 Mar 2006 22:27:25 -0800 (PST)
Jeff Frost <jeff at frostconsultingllc.com> wrote:

> Yes, I've done this in the past, but this is for a database playground test 
> system and I'd like to hang onto it for more than a year, not just a browse 
> the internet and run pine workstation which is what the original system it's 
> replacing was spec'd for.

I know this doesn't absolutely compare to a real world test but... my athlon x2 3800 "perform" ~2000 per CPU bogomips. My athlon xp 2500 3800 bogomips.

I bought an 64bit x2 cpu cos I was curious and cos I have good memories of a 2xPentium 133 and 2xPentiun Pro 200... since the feeling was they were really multitasking (expecially during desktop application freez).
Comparing my old xp to my new x2 I can't apreciate the same difference I was apreciating between my PII 233 and my 2xP133 (even and expecially for GUI).

I had to compile a small utility for motorola mobiles and the x2 looked a bit faster than the xp.

Maybe for "real works" expecially db x2 and 64bit make the difference. But at this point I'm a bit disappointed.
amd64 doesn't seem enough mature when it comes to desktop apps.
BTW does anyone know where I can find flash for dapper (amd64)? It doesn't seem to be in multiverse.

> What sort of numbers do you get from lmdd of an 8gig file (you are running 

If anyone can tell me in which package I could find lmdd in dapper... or better how to find in which package lmdd is...

> x86_64 right?).  I ask because some nforce boards are known for not so good 
> SATA performance, thus my surveying for real world results.

Where did you read it?
The few benchmarks around I saw say nforce4 performs not differently than other onboard ctl.
As Rick pointed out... your main concern should be... there are no open specification.
I just read about filesystem corruption on nvidia forum when nforce4 is associated with maxtor and samsung(?) hds... but it is hard to find the culprit... expecially without open specification.
Anyway it shouldn't be a big problem... most 939 boards comes with a sil ctl too. External sata ctl will increase the cost of the box by a considerable amount if you want apreciable differences in performance over a sil or an nforce4.

An asus a8n-sli comes with 8 sata ports... + 4 ata...

In soft raid 0 I got:
/dev/md0:
 Timing cached reads:   3084 MB in  2.00 seconds = 1541.46 MB/sec
 Timing buffered disk reads:  358 MB in  3.00 seconds = 119.23 MB/sec

So still far from the theoretical 3Gb/s of the nforce4 ctl.
I'm using 2x WDC WD1200JS-00MHB0.


oooh BTW... I've been a gamer when I was 10... it lasted 6 to 8 years. In the same period I was really interested in assembler, reverse enginering and security. Now I'm curious about games, I gave up with reverse enginering and security.
Sometimes I dream to have a console but I feel they are too expensive compared to the time I'll use one and I feel it won't be too cool to have a cell if the only way you know to program it is not too different from programming an x86.
So you don't have to miss a part of your brain to be a gamer...





More information about the svlug mailing list