this test is all cpu after a point but where does that point start for a 4500, 5500 and a 6000, in different places for sure, it seems a 6k card with  2600mhz or higher cpu needs a fsb up closer to 200 or higher[like tm30], i like the barton for this as the early chips were unlocked and they had 512 cache, cpu maufacuters found that as the speed incressed the performance wasnt keeping pace, it was all tied to the fsb, so if i was gonna attempt this and by looking at others results i'd try a little fsb to get over hump, after all this is a test of not a video card or a cpu but of a system, so its the point of having the fastest 3dfx system, be it a 4500 or a 6000. good luck to you all  

Wooot, another 2 pages  in topic

Well, I didn't think about high FSB problems with KT333... My last experience with Socket A was almost three years ago (before I switched to P4+875P platform), and for VIA chipsets it's almost four years (with EPoX 8KHA+ as my last mobo on these chipsets), so I'll need to recover some knowledge from backup ))

Regarding the statement "And about the higher scores with 3DMark2001 and faster CPU's --> Some CPU speed is being calculated 'into' the final score. When using a faster CPU you'll always get a higher score. Even if the graphics card won't go any higher anymore"

That's not exactly true. 3DMark2001 don't _calculate_ CPU power into final score. Score is counted as following:

(CL+DL+LL)*10+(CH+DH+LH+N)*20=SCORE

C = Car Chase
D = Dragothic
L = Lobby
N = Nature

L = Low detail
H = High detail

But some tests gladfully react on CPU power. That's most true to Car Chase Hi, which is almost CPU test, but also for Car Lo... Also, both Lobbys are xtremely sensitive to memory bandwidth, I even use Lobby Hi as a test for memory sticks Dragos looks like to be complex tests and Nature is strictly GFX-sensitive.

That's make 3DMark 2001 (not 03 or 05!) an excellent SYSTEM test and the best proving ground for benchmarkers - to get a decent score you should create a system without any weak points. CPU power, FSB speed, memory timings and of course graphics power - everything make the difference. Of course, with GPU of the past (i.e. GF2 or VSA-100) there isn't so big difference between aircooled or phase-change cooled modern CPU just because videocard will be a bottleneck, but still it will help to get one or two hundreds more points...

Aaaahhh.... Good ol' days with overnight 3DMark01 benching sessions looks like to return for me soon )) And for the first time with 3dfx hardware!
Usually I get myself to Futuremark absolute TOP-10 with each new hi-end card since Radeon 9800 Pro but last time I seriously benched this summer (FX-57 + 7800GTX SLI)... In addition, I still own world records for FX5900XT (27800 points), PCX5750 (20002), PCX5900 (22200) and few other older cards So to add 3dfx WR will be an honor to me

It thrills me just to know that you're accepting the challenge and will be stepping into the ring soon

Quote:


of course...but what will you do when your football team is going to win the superbowl? ok ok...its not the same but its nice to see that a german tweaker is on top...
if alt-f13 get the highest score it is also cool to see what is possible to get with 3dfx card...

well If he can make world records with experimental equipment he surely knows how to handle a 6K in a situation like this, maybe he informed himself alot better before buying which I did do also D'oh 

He is Russia's and Unkraine's most well known Hardware expert lol

maybe it would be best to end topics like this I'm sense them mostly in to no good anyways, the whole 3D Mark thing is a pointless discussion.

Quote:


what do you mean? probably that I do not inform myself when i buy things... or i inform myself not as good as ALT-F13? C'mon! this is really stupid. I am an overclocker since i got my P133...

I can read out that you DISLIKE THAT I AM THE ACTUAL W/R HOLDER. now you can say what you want but it sounds like this. i think other people might think that too... so F**k off!