G80 vs G92 with 4Ghz Yorkfield in 3dmark06 under vista x64

[safan80]

the test setup is as follows

yorkfield at 4Ghz (10x400)
G.skill (2 x 2GB) DDR2 1000 (PC2 8000) F2-8000CL5D-4GBPQ at 1066
Asus Maximus Formula SE bios 0903

3dmark06 was run at stock resolution (1280x1024) and 2560x1600 with and without AA and AF
vista x64 ultimate w/sp1 final
drivers were 169.32
G92 (8800GTS) on AIR 800/2000/1030
G80 (8800GTX) on water *661/1549/1000

*was only 3dmark06 stable

stock 3dmark06 (1280x1024)

G80


G92



1280x1024 4xAA 16AF

G80


G92



2560x1600

G80


G92


2560x1600 4xAA 16AF

G80


G92

[safan80]

using drivers 169.39

1920x1200

G80


G92



1920x1200 4xAA 16AF

G80


G92

[Psycho_eddie]

Nice review, although cant help but feel test bit unbalanced?

I.e

G92 800/2000/1030

G80 661/1549/1000


Its not really comparison between difference in performence between two cores as clocks way different and is reason for large differences in end scores

Should really have got yourself an

8800Ultra @ 700/1730/2000

8800GTS @ 700/1730/2000

That would be a compairson, sorry if you feel like im sniping just giving contructive critizism here as i would be interested in finding on a 'Comparison'

[dinos22]

no i think you miss the point

he is testing his cards with max or close to max clocks he'd run them in his system

i think what he's done is great. same deal with CPUs i think......a CPU is only as good as it overclocks.

great insight and it definitely paints an interesting story
now where are those single core tripple sli capable GFX cards Mr. Nvidia

[xwards]

great test GTS is the best

[safan80]

Nice review, although cant help but feel test bit unbalanced?

Its not really comparison between difference in performence between two cores as clocks way different and is reason for large differences in end scores

Should really have got yourself an

8800Ultra @ 700/1730/2000

8800GTS @ 700/1730/2000

That would be a compairson, sorry if you feel like im sniping just giving contructive critizism here as i would be interested in finding on a 'Comparison'

it would be nice to have an ultra but I don't have one.

great test GTS is the best

at 3dmark... I've put my GTX back into the system because at 2560x1600 with 8x AF in Quake Wars the GTX rules the GTS. you might want to check out the last set of benches with 2560x1600 4xAA and 16AF.

i think what he's done is great. same deal with CPUs i think......a CPU is only as good as it overclocks.

thanks

[Leeghoofd]

Good clocks Safan... I prefer my 8800GTS 512 over my GTX in 3dmark lately...

[Martijn]

Can you do the same runs on 1920x1200 and 1650x1080? I think quite a few people (including me) are running 22" and 24" LCDs

[safan80]

Can you do the same runs on 1920x1200 and 1650x1080? I think quite a few people (including me) are running 22" and 24" LCDs

1650x1080 will be like 1280x1024... I will try to do 1920x1200 later in the week, but it will be with different drivers.

[scottc19]

Great job, very interesting

[safan80]

updated see second post for 1920x1200

[WeStSiDePLaYa]

Looks like the extra bandwidth really helps the G80 with AA/AF.

Even clocked way lower it beats the GTS with aa/af.

[railmeat]

Nice review, although cant help but feel test bit unbalanced?

I.e

G92 800/2000/1030

G80 661/1549/1000


Its not really comparison between difference in performence between two cores as clocks way different and is reason for large differences in end scores

Should really have got yourself an

8800Ultra @ 700/1730/2000

8800GTS @ 700/1730/2000

That would be a compairson, sorry if you feel like im sniping just giving contructive critizism here as i would be interested in finding on a 'Comparison'

[QUOTE=Psycho_eddie;2779328]Nice review, although cant help but feel test bit unbalanced?

I.e

G92 800/2000/1030

G80 661/1549/1000






agree %100,make the clocks dead even then retest...


gtx will beat it out higher and higher as the res goes up.my gtx crushes my buddys 8800gt 512 at 2048x1536 in my rig using INDECTICAL clocks,we been there already.lower to med res there bout dead even.the gtx is more of a hi res firepower gamers card.were the gt is def the best deal on the planet for any res under 1600x1200.both amazing cards but the gtx still wins 1 year later strong.i feel it was money well spent.cannot wait for the new 9800gtx.....

[WeStSiDePLaYa]

[QUOTE=railmeat;2783883]

Nice review, although cant help but feel test bit unbalanced?

I.e

G92 800/2000/1030

G80 661/1549/1000






agree %100,make the clocks dead even then retest...


gtx will beat it out higher and higher as the res goes up.my gtx crushes my buddys 8800gt 512 at 2048x1536 in my rig using INDECTICAL clocks,we been there already.lower to med res there bout dead even.the gtx is more of a hi res firepower gamers card.were the gt is def the best deal on the planet for any res under 1600x1200.both amazing cards but the gtx still wins 1 year later strong.i feel it was money well spent.cannot wait for the new 9800gtx.....

Running them at the same clocks is dumb.

When we run our new cards, do we limit them to the clocks of our old cards?

We run them at their max, which is what he is showing.

[DStealth]

great test GTS is the best

In 3dmark sure.
Ive got stock all GTS with old Q6600 + water ~ 19k in 2006

My year and 3 months GTX is still in my slot you know why i'll give you some hints :

GTX@720/2240 - 24/7 def.volts
1680*1050 NoAA HQ
SM3 - 6332 HDR1-60+fps HDR2-66.6fps


GTX@720/2240 - 24/7 def.volts
1680*1050 SSAA4 AF16 HQ
SM3 - 4460 HDR1-49+fps HDR2-40+fps


You don't think there are still sm2 using games out there, do you ?
Greater score to GTS comes from SM2

If you compare the consistency of FPS specially ingame, you'll change your mind.
Ahh and here something to try to show with GTS 24/7 settings:
1680*1050 SSAA2 AF16 HQ

Don't forget it's almost year and a half old card.
I have them both, so got an opinion

[WRC]

the old school GTX is still a beast

[jcool]

Great comparison. For those of you that are interested, I have done a similar thing recently, comparing my G80 GTS vs. a G92 GT with 3DMark and Crysis, both 1280 and fullHD with and without AA.
Check it out here - even though most of you won't be able to read German, you can still compare scores. The G80 does kick some ass especially with Crysis. The main downside of G92 is its limited 512MB Vram, that's why the 1GB version of the GTS will most likely be the way to go for future games (missing bandwidth is not that tragic)

[safan80]

sorry I don't have xp 32 installed anymore. I might reinstall it when sp3 is released but that's doubtful.

[mascaras]

great test GTS is the best

In 3dmark sure.

yep , in 3dmark its obvious because of GTS G92 Higher clocks , but in games with High resolutions and filters (AA & AF) i think GTX G80 still better


regards