news
news archive
SEND NEWS!

articles
FORUMS!
links
contact

PC Database
([an error occurred while processing this directive] entries)

Folding Team

SETI@HOME Team
RC5 Team
Genome Team

OCAU chat

Creative Labs 3DBlaster GeForce2 Ultra 64MB - Page 2
Review by James "Agg" Rolfe

Overclocking
Sadly both my testbeds are suffering instability at high FSB's at the moment. This means I can't really tell you what AGP speed the Creative was happy at - at anything above stock I had problems, but I've been having that with cards I know can handle a high bus speed lately. Even the total rebuild of both for this article didn't fix it.

As for overclocking the card itself, I found it was stable right up to a core speed of 285MHz and a memory speed of.. wait for it.. 500MHz DDR! This compares to 222/400 for the GeForce2 GTS card we reviewed earlier and 220/380 for a Leadtek Winfast GTS I borrowed for this review.

Performance Comparison - Thunderbird 950MHz
I compared the Ultra card to a few others on a couple of different systems. First up, the firebreathing system. This should show clearly the difference between the various cards because they should never be waiting for data from the memory or CPU.

Testbed Info:
Asus A7V KT133 Motherboard
AMD Athlon Thunderbird 950MHz
Corsair PC133 CAS2 SDRAM
Leadman 400W PSU
IBM 75GPX 30GB HDD

Win98SE
VIA 4-in-1 v4.23
Detonators v5.32

I bet you're wondering why I used 5.32 detonators. Well, the short version is - I screwed up. The long version is, I wanted to use the latest versions of both the 4-in-1 drivers and Detonators that were available for download from the vendors sites when I got the card - I'm not comfortable running "leaked" or beta drivers because they may be strangely optimised or provide inconsistent results. However, for some reason I used my previous testbed standard Detonators when building the 2 boxes for this review and, given it takes 2 hours per card per machine to benchmark, by the time I realised, it was too late. I think if anything, the 6.x Detonators would have optimizations for the GeForce2, and hence show perhaps a greater margin between the two GF2 cards and the other two cards, rather than distinguishing between the two GF2 cards themselves - as discussed earlier they're essentially the same except for the clock speeds.

Cards Tested:
Creative Labs 3DBlaster GeForce2 Ultra 64MB @ Default (250MHz Core / 458MHz DDR Memory)
Creative Labs 3DBlaster GeForce2 Ultra 64MB Overclocked to 285MHz Core / 500MHz DDR Memory
Leadtek Winfast GeForce2 GTS 32MB @ Default (200MHz Core / 333 MHz DDR Memory)
Leadtek Winfast GeForce2 GTS 32MB @ Overclocked to 220MHz Core / 380 MHz DDR Memory
MSI 8809 AGPhantom 3D GeForce SDR 32MB @ Default (120MHz Core / 166Mhz Memory)
Diamond Viper V770U TNT2 Ultra 32MB @ Default (150MHz Core / 183MHz Memory)

I've presented the data in graph form, if you'd prefer a spreadsheet so you can see the actual numeric values (they wouldn't fit and be readable on these graphs), you can grab the data file here - it's an Excel 97 file so 2000 should open it fine.

First up is MadOnion.com's 3DMark2000 benchmark:

There's quite a lot of information there. The GeForce SDR and TNT2U unsurprisingly get left for dead pretty much right from the start, but it's only at about 1024x768 that the Ultra starts to really pull ahead of the GTS - at default clocks, only 1.5% difference at 640x480x16 (which is basically zero, allowing experimental error). This stretches to over 20% by 1024x768x32. In the higher resolutions the scores for all cards level out as the system is maxed out and the difference between the GTS and the GF2U stays fairly static, at 1600x1200x16 the Ultra is 17% faster, in 32bit at that resolution it is an impressive 34% faster in this benchmark. 6000 3DMarks at 1600x1200x32 shows an absolutely staggering fill rate.

Next is id-software's Quake3, running a timedemo of Demo1:

The "Fastest" and "High Quality" settings were as default in Quake3, but for the HQ 1024x768 as well as changing the resolution I set the texture detail slider to maximum and geometric detail to high. This one is pretty self-explanatory, as the detail level and resolution goes up the Ultra pulls pretty obviously ahead of everything else including the GTS. In HQ at 1024x768 at default clock the GF2U is 26.5% faster than the GTS.

NEXT PAGE - P3-550E Performance

Major Sponsors:

All content copyright 1999-2002 James Rolfe. All rights reserved. No reproduction allowed without written permission.
Interested in advertising on OCAU? Contact us for info.