2-way SLI 9800GX2 vs. 3-way SLI 9800GTX

[hqnffruh]

any sites done benchmarks with the graph things? are the drivers supporting 9800GX2 SLI out yet? tell us what you know. which setup is the better overall performer?

 

[ChrisRay]

http://www.rage3d.com/previews/video/nv9800gtx/

((my review))

and techreport have done some direct comparisons too.

http://www.techreport.com/articles.x/14524

Drivers have been available since a week after the 9800GX2 launch

 

[Fenriss]

interesting review Chris, but why no 8800GTX Ultra?
i really think a threeway ultra will trump any setup including the 9800's and it would be interesting to know by how much. the TechReview sorta points it out but it's hard to quantify since they used diff mobo's to test the Ultras and the 9800's. i know the 780i is basically a 680i but it's sort of an improvement of it and is suppose to be faster. so it's kind of handicaps the Ultra.
also, shouldn't the setup be in a baseline config? that is no OCing @ all. since the cards have diff memory bandwidth and memory size which has an affect on OCing.

 

[angry hampster]

Originally posted by: Fenriss
interesting review Chris, but why no 8800GTX Ultra?
i really think a threeway ultra will trump any setup including the 9800's and it would be interesting to know by how much. the TechReview sorta points it out but it's hard to quantify since they used diff mobo's to test the Ultras and the 9800's. i know the 780i is basically a 680i but it's sort of an improvement of it and is suppose to be faster. so it's kind of handicaps the Ultra.
also, shouldn't the setup be in a baseline config? that is no OCing @ all. since the cards have diff memory bandwidth and memory size which has an affect on OCing.

You're probably right -- but I think the verdict would be heavily dependant on resolution. At 1680 or smaller, a 9800 card setup will probably come out on top, but at larger resolutions, the huge memory bus of the Ultra really helps.

 

[Fenriss]

good point, hampster.
still, since higher rez is the trend in gaming, i would put more weight on the card that faster in high rez.

 

[sgrinavi]

Originally posted by: Fenriss
interesting review Chris, but why no 8800GTX Ultra?
i really think a threeway ultra will trump any setup including the 9800's .

A tri-9900-Ultra rig is a $2000 solution, it had better trump a tri-9800's at less than half the cost.


I think nvida is doing a great job considering the performance they are delivering at some amazing street prices. I think they are spot-on with their new product lines albeit the goofy naming/numbering conventions.

Slap a $500 gx2 on a $100 p35 board with a $75 CPU, add $300 for some ram, psu, drives etc and blow away a rig that you would have sunk $2000 into last summer.

 

[phexac]

This is such graphical overkill

I cannot ever see myself spending that much on graphics in a rig, especially seeing that an good upper-mid-range card such as 8800GTS 512MB can run every game out there, besides Crysis, maxed out at 1920x1200.

 

[hqnffruh]

Originally posted by: phexac
This is such graphical overkill

I cannot ever see myself spending that much on graphics in a rig, especially seeing that an good upper-mid-range card such as 8800GTS 512MB can run every game out there, besides Crysis, maxed out at 1920x1200.

Doubt it, I can't run everything maxed out with two 8800GTs at 1920x1200. When I say maxed out i mean 60fps constant. Sure you might get 30-60 FPS but it's so unstable. I only find this scenario with unreal engine 3 games that just came out like frontlines, rainbow six vegas 2, unreal tounrment 3, etc. Now if you're talking DX9 that's different era and isn't even really an viable option at this point.

 

[sgrinavi]

Originally posted by: phexac
This is such graphical overkill

I cannot ever see myself spending that much on graphics in a rig, especially seeing that an good upper-mid-range card such as 8800GTS 512MB can run every game out there, besides Crysis, maxed out at 1920x1200.

I have/had an 8800GTX, an 8800GT, 8800GTS /g92, 3870 crossfire and, now, the 9800 GX2. None of them will run everything maxed out at 1900x1200 at (what I consider) reasonable frame rates (50-60)...

 

[Jax Omen]

Originally posted by: phexac
This is such graphical overkill

I cannot ever see myself spending that much on graphics in a rig, especially seeing that an good upper-mid-range card such as 8800GTS 512MB can run every game out there, besides Crysis, maxed out at 1920x1200.

Lost Planet is a slideshow maxed out on DX10 on my rig, and I haven't even tried it with AA yet.

Most high-end games run between 30-60fps when maxed out with no AA, in my experience (fast enough for me, but certainly not for everyone, and if I could afford it I'd go SLI )

 

[Naeeldar]

Originally posted by: sgrinavi

Originally posted by: phexac
This is such graphical overkill

I cannot ever see myself spending that much on graphics in a rig, especially seeing that an good upper-mid-range card such as 8800GTS 512MB can run every game out there, besides Crysis, maxed out at 1920x1200.

I have/had an 8800GTX, an 8800GT, 8800GTS /g92, 3870 crossfire and, now, the 9800 GX2. None of them will run everything maxed out at 1900x1200 at (what I consider) reasonable frame rates (50-60)...

Human eye is only capable of seeing 36 fps I thought? So why is 50-60 reasonable?

 

[hqnffruh]

in games like UT3 less FPS = more lag because it's a really fast game

 

[sgrinavi]

Originally posted by: Naeeldar


Human eye is only capable of seeing 36 fps I thought? So why is 50-60 reasonable?

Says who? and it probably depends on the media and venu.

One other thing to consider; if I am looking for an average of 50-60 then I am going to have lags in the 30's, no doubt.

 

[CP5670]

Time to post this yet again:

http://www.100fps.com/how_many...mes_can_humans_see.htm

 

[hqnffruh]

well I can sure as heck tell the diff between 60fps and 40. it's slower to my eyes. sure go ahead and post your scientifical study it's probably right, but any gamer can tell the engine is definately running slower at 40 rather than 60 to the naked eye even.

 

[CP5670]

If you actually read the link, it fully backs up what you're saying.

In general, I have no trouble seeing differences up to at least 100fps in almost any game, as long as the refresh rate is high enough.

 

[v8envy]

Depending on the title my ability to correctly determine faster vs slower frame rate can be as little as 10% frame rate difference, also up to about 90fps-ish. One of the reasons I elected to have a 3 year old 1600x1200x120hz 21" trinitron repaired for more than a 1900x1200x60hz 24" LCD would have cost me.

Just do a high speed 180 degree turn. The difference between 36 fps and e.g. 60 (never mind 90!) is night and day.

 

[FiLeZz]

I have 3 8800 gtx card in tri SLI



Currently no 9800gtx score has beaten my 3dmark scores.. at either hi res or the default test.. here are my links to my scores to help.

P13438 Score
http://service.futuremark.com/...Id=15853&resultType=19

X6363 Score
http://service.futuremark.com/...tId=7627&resultType=19


However I have been beaten by the 9800gx2's in quad SLI