Nvidia NV 31 and NV34 benchmarks

Zardon · Feb 28, 2003, 08:21 AM

Inquirer

IT FINALLY MAKES SENSE why Nvidia does not like 3dmark 2003 and why the company has even made public statements against it. We managed to get scores from the latest Futuremark benchmark tool after running it on 2 Nvidia cards that are yet to be introduced. They were the performance mainstream card NV31, likely to be called Geforce FX 5600, and the NV34 mainstream card with the possible name of Geforce FX 5200.
The machine these two cards were plugged into was an Athlon 2700+ with 512 MB of RAM. Very similar to the machine we use. The only card we have run 3dmark03 on before now was the Radeon 9700 PRO and that baby, as far as we can remember, scored around 4600 marks.

The faster of the two new Nvidia cards, the NV31 scored 2389. We cannot make a comparison to the Radeon 9500 and 9500 PRO as we are away from our LAB at the moment. If someone has a Radeon 9500 PRO and similar configuration we would like to update the article with its result as well.

You can note that card is about half the speed of the Radeon 9700 Pro.
Its mainstream brother, the NV34 was unable to score more than a thousand. To be precise it got 879 marks making it only 37% of the speed of the NV31.
These results may not reflect real in-game performance but sure can give you some conclusions about cards that are not so far away from being publicly launched.
It sure gives you a clue why Nvidia could be so harsh about Futuremark. It will be real interesting to compare these scores with RV280 and RV350, the soon to be introduced ATI parts.

The_Neon_Cowboy · Feb 28, 2003, 11:15 AM

who cares about ... 3dmark 2003 I wanna see 3dmark 2001SE benchmarks!!
3dmark 2003 marks are too unrelyable... to many variables basesd on the users hardware /brands/tests change the way they act based on hardware so not evey card ges the same test( even though they are running the same exact test)/ not the performance of that hardware...

starboy · Feb 28, 2003, 02:35 PM

Agreed. 3dmark03 cannot be relyable. it also seems like not updated comparing 3dmark2001se

digitalwanderer · Feb 28, 2003, 02:42 PM

You're both wrong, nVidia is going to be SERIOUSLY screwed when these cards go public and we get to see what they really can do.

These are gonna make the FX look like a well thought out and high quality part.

BWX · Feb 28, 2003, 04:46 PM

879!!??? WTF? MY ti4200 gets twice that on this old slow ass machine (comparatively speaking)-

Looks like they are pulling the old GF4 MX trick, let me guess- the NV34 is actually a gf2 mx w/ 8 times AGP and non-DDR sdram. LOL- I wouldn't doubt it.

What the hell is Nvidia doing?? I have lost all faith. As soon as I can I am getting an ATI 9700 pro.

***But the Nforce 2 looks like a good chipset.
Hmm, Maybe Nvidia is spreading itself too thin?

mbeeston · Feb 28, 2003, 07:01 PM

why you all say 3dmark is not reliable?.. it's only been out like a monthe.. it'll prove itself

btw... can you say "SLOW!"?

The_Neon_Cowboy · Feb 28, 2003, 07:37 PM

Quote:

Originally posted by mbeeston
why you all say 3dmark is not reliable?.. it's only been out like a monthe.. it'll prove itself btw... can you say "SLOW!"?

lest see to sart useing diffent shaders etc.. on diffrent cardss... that make it make it scores uncompareable to each other... unless you compairing to systems exactly alike with the same card only....it's useless

Desert_Siege · Mar 1, 2003, 02:21 AM

Quote:

Originally posted by starboy
Agreed. 3dmark03 cannot be relyable. it also seems like not updated comparing 3dmark2001se

This benchmark is SO much more dependent on raw video card power than system power. In 3DMark 2001SE, I can get almost a 900 point increase by simply raising my FSB, thereby increasing the AGP bus marginally with only a 25Mhz difference in processor speed. Do the same thing in 3D Mark 03 and you will do good to get a 1-200 point increase, which I think is a much better measurement of VIDEO CARD power instead of overall system power. When you acutally overclock the card AND the system, the scores are more drastically effected.

3DMark 2001SE:

11.5x166=1909/ card default core and mem: 13169
10 x 190 =1907?/ card default core and mem: 13912
10x190 /card at 390/360 core/mem: 14721

3DMark 2003
11.5x166/card default core and mem: 4498
10x190 /card default core and mem: 4649
10x190 /card at 390/360 core/mem: 5299

As you can see, 3DM 2k3 is MUCH more dependent on Video card performance than sytem performance. Its only when the card itself is overclocked that the scores really start to increase noticeably above the margin of error. As far as why they didn't optimize the bench to run differently on different hardware I think the answer to that is pretty obvious. If it acted differently on every video card it would be useless wouldn't it? I mean you might get suckered in to buying a card based off an "optimized" benchmark only to find that it sucks ass in all your games because they aren't optimized for your card and you would be pretty upset wouldn't you? Not that anybody here would buy a card based on its 3D Mark score or anything..heh.

I think as long as the card uses the same routines no matter what video card is installed and those routines are the same routines that alot of the upcoming games are going to use, then its a pretty good indicator of expected performance on alot of the upcoming games isn't it? And shouldn't it be that way? Not everyone optimizes there games to run different paths on 10 different cards right out of the box do they? Don't know, guess I think of it in car terms...lol. I mean, if your video card is your engine and you hook your engine to a dyno and it makes X amount of power and you hook a different engine to the same dyno and it doesn't make as much power, then would you switch the dyno or work on the engine? Stupid analogy maybe, but If you switched the dyno, the results would be pretty useless wouldn't they? I think Nvidia better get under the hood and work on the engine cause its sputtering pretty badly. Thats just my opinion and I could be wrong, but who really cares anyway?

DriveEuro · Mar 1, 2003, 03:49 AM

BWHAHAHAHHAHAHAHA, my Ti4200 is better, shit any Ti4200 is better. But even still. These cards are turning out to be a joke.

Feb 28, 2003, 08:21 AM	#1
Zardon DriverHeaven Founder Join Date: May 2002 Posts: 32,480 Rep Power: 179	Nvidia NV 31 and NV34 benchmarks Inquirer IT FINALLY MAKES SENSE why Nvidia does not like 3dmark 2003 and why the company has even made public statements against it. We managed to get scores from the latest Futuremark benchmark tool after running it on 2 Nvidia cards that are yet to be introduced. They were the performance mainstream card NV31, likely to be called Geforce FX 5600, and the NV34 mainstream card with the possible name of Geforce FX 5200. The machine these two cards were plugged into was an Athlon 2700+ with 512 MB of RAM. Very similar to the machine we use. The only card we have run 3dmark03 on before now was the Radeon 9700 PRO and that baby, as far as we can remember, scored around 4600 marks. The faster of the two new Nvidia cards, the NV31 scored 2389. We cannot make a comparison to the Radeon 9500 and 9500 PRO as we are away from our LAB at the moment. If someone has a Radeon 9500 PRO and similar configuration we would like to update the article with its result as well. You can note that card is about half the speed of the Radeon 9700 Pro. Its mainstream brother, the NV34 was unable to score more than a thousand. To be precise it got 879 marks making it only 37% of the speed of the NV31. These results may not reflect real in-game performance but sure can give you some conclusions about cards that are not so far away from being publicly launched. It sure gives you a clue why Nvidia could be so harsh about Futuremark. It will be real interesting to compare these scores with RV280 and RV350, the soon to be introduced ATI parts.

Feb 28, 2003, 11:15 AM	#2
The_Neon_Cowboy HardwareHeaven Extreme Member Join Date: Dec 2002 Location: U.S.A. Posts: 16,009 Rep Power: 92 System Specs	who cares about ... 3dmark 2003 I wanna see 3dmark 2001SE benchmarks!! 3dmark 2003 marks are too unrelyable... to many variables basesd on the users hardware /brands/tests change the way they act based on hardware so not evey card ges the same test( even though they are running the same exact test)/ not the performance of that hardware... __________________

Feb 28, 2003, 02:35 PM	#3
starboy HardwareHeaven Lover Join Date: Aug 2002 Location: Brunei Posts: 127 Rep Power: 0	.. Agreed. 3dmark03 cannot be relyable. it also seems like not updated comparing 3dmark2001se __________________ One Shall Stand! One Shall Fall!

Feb 28, 2003, 02:42 PM	#4
digitalwanderer Colour Commentator Join Date: May 2002 Location: Highland, IN USA Posts: 5,619 Rep Power: 0	You're both wrong, nVidia is going to be SERIOUSLY screwed when these cards go public and we get to see what they really can do. These are gonna make the FX look like a well thought out and high quality part. __________________ WTF is up with the sigs?

Feb 28, 2003, 04:46 PM	#5
BWX unplugged Join Date: Nov 2002 Location: USNY Posts: 19,669 Rep Power: 110 System Specs	879!!??? WTF? MY ti4200 gets twice that on this old slow ass machine (comparatively speaking)- Looks like they are pulling the old GF4 MX trick, let me guess- the NV34 is actually a gf2 mx w/ 8 times AGP and non-DDR sdram. LOL- I wouldn't doubt it. What the hell is Nvidia doing?? I have lost all faith. As soon as I can I am getting an ATI 9700 pro. ***But the Nforce 2 looks like a good chipset. Hmm, Maybe Nvidia is spreading itself too thin? __________________

Feb 28, 2003, 07:01 PM	#6
mbeeston HardwareHeaven Senior Member Join Date: Oct 2002 Location: alberta, canada (thats north of the usa people) Posts: 1,037 Rep Power: 68 System Specs	why you all say 3dmark is not reliable?.. it's only been out like a monthe.. it'll prove itself btw... can you say "SLOW!"? __________________ R.I.P Roadee - you will be missed T_T peace,love, and war... 3 things people look for... "if your going off the deep end, show off, do a cannon ball!" - me

Mar 1, 2003, 03:49 AM	#9
DriveEuro HardwareHeaven Extreme Member Join Date: Jul 2002 Location: Gurnee Illinois Posts: 4,677 Rep Power: 0	BWHAHAHAHHAHAHAHA, my Ti4200 is better, shit any Ti4200 is better. But even still. These cards are turning out to be a joke. __________________ [color=orange]ASUS P4P800 Deluxe [color=black]-[/color] 2.4C --> 3300mhz [color=black]-[/color] Mushkin PC3200 at 220 5-2-2-2 (cpu limited) [color=black]-[/color] BBAti 9800Pro at 430/375 [/color] [color=black]3[/color][color=gray]D[/color]M[color=black]a[/color][color=gray]r[/color]k[color=yellow]01[/color] [color=black]3[/color][color=gray]D[/color]M[color=black]a[/color][color=gray]r[/color]k[color=yellow]03[/color] 16001200 12801024