On Friday January 31 2003 09:51 am, Myers, Dennis R NWO wrote:

> Nvidia Geforce 2-32 megs Ram         Nvidia Geforce 2-64 megs Ram
> (made by Verto-PNY)                         (made by MSI)
> 4191 drivers                                      Drivers that come with
> v9.0
>
> Now I know that GLXGears is not the best thing in the world to use
> for a comparison, but the results surprised me:
>
> 2600fps                                               2200fps
>
> Why would the system in column 1 outperform (video wise) the
> machine in the 2nd column? It seems like it would be the other way
> around? (could the drivers make that much of a difference?)
>
> Thanks! :-)

   Besides Charles' pertinent comments, no two vendors nvidia cards 
are equal. Could be the PNY card has a higher memory and core (GPU) 
clock than the MSI version, or vice versa. If you don't know these 
parameters, 'nvclock -i' (as root) will spit 'em out 
(nvclock-0.6.1-2mdk rpm). See 'man nvclock' if you want to try'n 
overclock the card. +10% is safe, even without a fan on the card's 
heatsink. Either cards' onboard memory size (ie, 32 vs. 64) makes 
neglible difference. The MHz it cycle's at, and also the core's,... 
_does_.

   Actually I'm surprised your results are apparently as close as they 
are, altho no 'glxgears' results should be construed as a bench, are 
only some what relevant if resolution an color depth (bbp) are 
included, and the output is run fullscreen, and as Charles said, 
background running procceses are taken into account. To expound a 
little further on Charles' other point, no two systems are equal in 
terms of bios (and it's configuration), motherboard capability, and 
on an' on. So in the end, system to system comparisons are rather 
pointless. Benches are only relevant for gauging system adjustments.
'Sides, glxgears is not a bench.

   I've got a Gef2-64mb (Abit), 167mhz memory, 200mhz core (oc's to 
185/225 with ease). Altho I use it with the Xfree driver that has 
some 3d hardware acceleration (minus direct rendering), I have tested 
it occaisonally with the the proprietary guess who knows what binary 
ones from nVidia. Last time with 1491's vs. XFree-4.2.99 (4.3). The 
secret crap is still 3+ times the performance, but with added system 
trade offs. http://www.mandrakeforum.org/article.php?sid=427&lang=en
Plus the Xfree drivers don't need 'mem=nopentium' to keep from fsck'n 
up your linux install ;~>>
-- 
    Tom Brinkman                  Corpus Christi, Texas

Want to buy your Pack or Services from MandrakeSoft? 
Go to http://www.mandrakestore.com

Reply via email to