how much increase in 3DMark05 after going with SLI?

[CHaynes112]

How much of an increase does SLI give you in 3DMark05? Like 50%, 70%, etc... For example, a single card at 8500... what (approx) would the result be if I went SLI (with the exact same card/clocks)?

-Thanks

[saaya]

depends on what card your un in sli/crossfire and how strong your cpu is
with a x1600pro and x1600xt its around 70% and cpu speed doesnt matter at all
with a x850xt its around 60% cpu speed matters
with a 7800gtx its around 50% cpu speed matters a lot
with a 6800U its aroud 80% cpu speed matters

[CHaynes112]

So I'm assuming a GT will be about 50% also (since a GTX is)?

[Elfear]

I got around a 55% increase going with SLI. ~9000 to ~14,000

[CHaynes112]

very nice to hear :D

-thanks

[mr_knowitall15]

I got around a 55% increase going with SLI. ~9000 to ~14,000
actually, thats 64.2857% according to the ol calculator :fact:

[afireinside]

8500 to 14100 with GT sli at lower clocks than single

[saaya]

gts scale better in sli than the gtx because theres more cpu headroom.
if we all had a 4ghz a64 x2 i think gts and gtx cards would scale the same :D

[Elfear]

actually, thats 64.2857% according to the ol calculator :fact:

Wweeelllll Mr. Know-it-all (J/K) if my math is correct (14,000/9000) -1 = .55 or 55%, or done another way, 14,000 - 9,000 = 5,000 --> 5,000/9000 = .55 or 55%. Someone correct me if I'm wrong here.

Sorry to get off topic.

[CHaynes112]

It seems like 55% to me :D

[IYP]

err....9000/14000 = .64285.... = 64.285...%

[Elfear]

err....9000/14000 = .64285.... = 64.285...%

That's true but I don't believe that's how you calculate the % increase from a single card. Someone correct me if I'm wrong but I thought the way to calculate % increase was the way I showed in my post above. The score increased 5000 points which would be 55% of the original score.

[craig588]

Yes, there's a 55% increase and a 64% difference.

Just a random example of percents messing you up: Your score goes up 10% then down 10%, you're left with 99% of your original score.

[CHaynes112]

The answer is 55%... anything else is wrong :) It looks like a couple of people above were dividing the numbers wrong and/or using the wrong formula...

[saaya]

err....9000/14000 = .64285.... = 64.285...%

9k is 64% of 14k so?
you calculated that a single card gets 64% of the performence of an sli config :slap: :D

[Elfear]

Pheww. I thought my budding career as an accountant was about to die an early death.:explode2:

[CHaynes112]

Pheww. I thought my budding career as an accountant was about to die an early death.:explode2:

LOL

[cadaveca]

went from 9430 in single GTX to 13848 in SLI.

[CHaynes112]

so it seems that 45% to 55% is a rough average... not too bad :)

[saaya]

thats not too bad indeed :)

[cadaveca]

but still not enough to justify the cost of another card. maybe if adding the card added exra rendering features, but when one card is more than enough to play almost every game on high settings...i had to say bye-bye to SLi.

[Elfear]

but still not enough to justify the cost of another card. maybe if adding the card added exra rendering features, but when one card is more than enough to play almost every game on high settings...i had to say bye-bye to SLi.

I think it kind of depends on what monitor you're using and what games you play. 1600x1200 and above and SLI is a huge help. If you game at 1280x1024 than there are only two or three games out right now that a single card would struggle at in which case SLI might not be a great option.

[cadaveca]

well, with most LCD's only doing a max res of 1280x1024, that kinda kills the monitor idea...i use a CRT @ 1600x1200...only FEAR caused any issues, but going from singlecard to sli only made it go from 1024x768 w/4xAA to 1280x960 w/4xAA. that's not that big a jump for adding even 40% of the processing power.

Don't get me wrong, there ARE situations were SLI is nessecary, but those situations, currently, are few and far between. Maybe i nthe future that situation will change, but i doubt it will happen soon.

[CHaynes112]

well, with most LCD's only doing a max res of 1280x1024, that kinda kills the monitor idea...i use a CRT @ 1600x1200...only FEAR caused any issues, but going from singlecard to sli only made it go from 1024x768 w/4xAA to 1280x960 w/4xAA. that's not that big a jump for adding even 40% of the processing power.

Don't get me wrong, there ARE situations were SLI is nessecary, but those situations, currently, are few and far between. Maybe i nthe future that situation will change, but i doubt it will happen soon.

huh? can't 7800GT SLI run FEAR at 1600x1200 @ max AA...?

[afireinside]

I got 45 fps average and a playable (For single player) frame rate with my GT SLI... FEAR at 4aa/16af MAX everything but soft shadows and 1680x1050. 2.8ghz dual core and 475/1180 card clocks of course :)

[cadaveca]

huh? can't 7800GT SLI run FEAR at 1600x1200 @ max AA...?
not without vsync issues in various areas that made it really hard to play, at least for me. I figured i'd get flawless graphics...i did not.

[afireinside]

people actually use vsync?

[cadaveca]

actually, i almost wanted to because FEAR was so bad, but it dropped max frames VERY low. But i did try.

[CHaynes112]

that's really upsetting :(

[Elfear]

FEAR runs very well for me at 1920x1200 w/ 2AA/8AF everything at max and soft shadows off. I get the following for the stress test:

Min: 48
Avg: 65
Max: 202

4AA causes too big of a hit to be playable. At 1600x1200 with same settings as above except 4AA I get:

Min: 26
Avg: 49
Max: 183

0% below 25fps
3% between 25 and 40fps
97% above 40fps

I think a single card would get pummeled at those settings.

[Rayman]

i have xfx6800gs and went from 5800 stock single card 2 9500 stock sli