First 3rd party benchmarks/review V8 ultra duo

Started by qrazi, 03 November 2003, 15:22:44

Previous topic - Next topic

qrazi

A german site, the Tecchannel received a beta board of the Volari duo V8 ultra.
http://www.tecchannel.de/hardware/1271/index.html

some benchmarks numbers for all of you who dont read german:
http://www.tecchannel.de/hardware/1271/images/0014169_PIC.gif
http://www.tecchannel.de/hardware/1271/images/0014162_PIC.gif
http://www.tecchannel.de/hardware/1271/images/0014166_PIC.gif
http://www.tecchannel.de/hardware/1271/images/0014167_PIC.gif
http://www.tecchannel.de/hardware/1271/images/0014168_PIC.gif
http://www.tecchannel.de/hardware/1271/images/0014163_PIC.gif
http://www.tecchannel.de/hardware/1271/images/0014164_PIC.gif
http://www.tecchannel.de/hardware/1271/images/0014165_PIC.gif

the used system is a P4 3,2 GHz mit 512MB pc3200 memory.
the Board and drivers are still beta. the memory (DDr2) on release is supposed to run at 500 mhz (1ghz effectively) instead of the 450mhz memory used on this board.
the drivers are still beta too, and havent got any FSAA or AF options in them.... that does make me a bit weary of the FSAA and AF quality and performance hit.
good thing is that there werent any artifacts during this admittedly small benchmark run...


ECS K7s5a@ 147 MHz, Athlon XP 1600+@1825+, Voodoo5 5500 @166 MHz, 512mb pc3200

NitroX infinity

I don't have much faith in these chips as high-end chips since they need a dual chip card to get close to the performance of today's high-end chips.
Still, you have to start somewhere and I think XGI's next generation will be a lot better.
 

qrazi

well, i didnt think they would outperform ati and nvidia anyway, but the flagship does come close to the 9800pro 5900 ultra cards, especially if you take the beta status in account.
i think also that XGI needs to offer good price/performance, because not only is the current flagship from ati and nvidia faster, but XGI will have to compete with the next ATi and nvidia cards in a few months... :/

ECS K7s5a@ 147 MHz, Athlon XP 1600+@1825+, Voodoo5 5500 @166 MHz, 512mb pc3200

NitroX infinity

Well, it comes close to a Radeon but it outperforms a GeforceFX.
They used the 52.16 drivers for the GeforceFX which are the wrong drivers. The 43.03 are the last correct working drivers. And with those drivers the GeforceFX cards perform at about 55% of their Radeon counterparts while the Volari's perform better than those 55%.
 

qrazi

as long as those drivers produce the right image, which they do as i understand from most sites, and there are other benchmarks used then just a 3dmark one, i see no reason why 52.16 shouldnt be valid drivers... most users will be using them anyway....

ECS K7s5a@ 147 MHz, Athlon XP 1600+@1825+, Voodoo5 5500 @166 MHz, 512mb pc3200

NitroX infinity

I don't believe those 52.16 drivers produce the right image. It would mean that the 43.03 drivers would have been crap. Which means that all GeForce (1/2/3/4) cards have been underperforming up till now. Unless there were some GeForceFX specific issues that were causing the bad performance.
 

qrazi

i am not sure how long you have been following the development of the detonator drivers, but there have always been increases in speed a few months after the release of a certain card. it is what made the geforce2 gts decisevily faster then the radeon ddr and voodoo5.
just as we can see speed improvement in our own 3dfx driver, so should nvidia be able to optimize the coding of its drivers. i actually dont follow the latest new that closely anymore, but what i get from www.tweakers.net (lets say, a dutch hardocp.com) 52.16 drivers are cheat free, or at the very least have the biggest cheats removed.

ECS K7s5a@ 147 MHz, Athlon XP 1600+@1825+, Voodoo5 5500 @166 MHz, 512mb pc3200

NitroX infinity

#7
*cough* t.net full abbo here *cough* [:P]

Still, if the 52.16 drivers are mostly cheat free than they've improved their performance by let's say 40% - 80% compared to the 43.03 drivers. I find it hard to believe that you can obtain such a high performance increase from just the drivers.

The only immaginable situation in which that could happen is when the 43.03 would have been real bad. At least for the GeForceFX cards.
Then I wonder, what went wrong with driver development at nVidia?
 

qrazi

:) didnt know that the performance increase was 80%.... hmmz... that sounds a bit to much indeed...

we will have to wait for more benchmark numbers to say something significant about volaris perfomance though... cant wait to see em...

ECS K7s5a@ 147 MHz, Athlon XP 1600+@1825+, Voodoo5 5500 @166 MHz, 512mb pc3200

Glide

XGI package (card + driver) is very, very young in comparison with Radeon + Catalyst and Geforce + Detonator products so we must wait driver updates to begin getting the real Volari hardware performance (imho).

;)

Bye bye


lecram25

#10
Actually, the 52.16's are full of cheats...in the dx9 aspect of it. Apparently, the 52.16s do not run any form if not no form of PS 2.0...Someone from x3dfx has confirmed this by running Aquamark. If you purchase the full version, there is an option that lets you see a vid as to what the card is rendering. As I said, with the 52.16s, little to no PS 2.0 was going on. So the nVdia test is moot, i.e. the V8 Duo is faster than the 59xx. In all dx9 intensive tests, except for ogl based games of course, as we all know nVidia fairs really well in ogl...

EDIT: but the Volari does look good...REAL GOOD! [:p]

| AMD Athlon XP (Barton) 2800+ @ 3200+ (11x200) | Cooler Master Jet 7 | ABIT NF7-S Rev 2.0 (nForce 2) | Corsair TWINX1024-3200C2 | Sapphire ATi Radeon 9500np>9700np | Sound Blaster Live! X-Gamer 5.1 | 3dfx VoodooTV 200 PCI | Western Digital 40GB & 120GB HDD | Pioneer 106s DVD | LG 52x24x52 CDRW | Cambridge Soundworks Desktop Theater 5.1 DTT3500 Digital Speakers |

NitroX infinity

Whahaha, nVidia screwed up big time. [:P]

ASUS P4S533 | Intel Pentium 4 2,261GHz | 512MB Samsung PC2700 DDR-SDRAM | Voodoo 5 5500 AGP
 

qrazi

hmmmz, i just read Firingsquads review of the forceware 52.16 driverset, and i never saw 80% performance increase. in fact there were a lot of decreases in perfomance as well.

the major increases were in the benchmarks that utilize Pixel Shader 2.0 (directx9 games), which can infact be explained by the new compiler and other optimalizations....

they ran Shadermark to see what the nv35 actually could do, and while it still cant do all tests, it did a few more, and performed better....

read their conclusion http://www.firingsquad.com/hardware/forceware_52.16/page16.asp

ECS K7s5a@ 147 MHz, Athlon XP 1600+@1825+, Voodoo5 5500 @166 MHz, 512mb pc3200