Sli/Crossfire? hA!

[bartvandenberg]

i jsut wanted to post something quick here, for all those out there who dont like to research stuff.

i think there is a bit of a misconsception about the performance gains of sli/crossfire.

so.. im posting these links of benchmark results so people can see for themselves if its really worth the extra double amount on gpu's.

Far Cry http://www.pcstats.com/articleview.cfm?articleid=1930&page=7

Doom3 http://www.pcstats.com/articleview.cfm?articleid=1930&page=9

Quake 4 http://www.pcstats.com/articleview.cfm?articleid=1930&page=10

Fear http://www.pcstats.com/articleview.cfm?articleid=1930&page=11

i really hope this opened some peoples eyes to see you are much better off buyin one good card, then 2 crappy ones.

and hey .. it im completely out to lunch. please.. let me know...

[jrabbitb]

yeah, the only reason to do sli that i can think of is to get full power to nice monitors, two or more monitors.

[bartvandenberg]

but.. in that case.. you dont combine them with sli. you run them seperately.

[tohito]

Odd, I come to an entirely different conclusion from looking at these benchmarks. Most of these cards are realizing a nearly doubled frame rate when connected via SLI. For example on Splinter Cell, Chaos Theory 1600x1200 SM3, a single Asus EN7800GT had a frame rate of 34.34 FPS. Two of the same cards in SLI mode achieved 66.86 FPS! That looks like a substantial gain to me. Now, perhaps for the price of two 7800's you could buy a card that could get 66.86 FPS on its own, but if so pcstats.com didn't test it. The highest frame rate listed in their benchmark table belongs to dual 7800 GTXs. Of course, we certainly don't see such a dramatic increase in frame rate on most benchmarks. For example, on 1024x768 SM1 that same card sees only about a 50% increase in frame rate.
But here's another reason to consider SLI: suppose a person has about $200 to spend on a video card right now. They can afford perhaps a geforce 6600-based card, but not a 7900. However, in a few more months when they have some more money, rather than buying a whole new card they could buy another 6600 and link the two together. SLI: It's scalable!

[Cevinzol]

The games themselves seem to be more the deciding factor. some make better use of the second card.

So the real question is which game do you plan on playing?

But here's another reason to consider SLI: suppose a person has about $200 to spend on a video card right now. They can afford perhaps a geforce 6600-based card, but not a 7900. However, in a few more months when they have some more money, rather than buying a whole new card they could buy another 6600 and link the two together. SLI: It's scalable!
"In a few more months".. Do you know how long that is in computer years much less dog years?

I have learned over the past 15 years building computers you don't plan on upgrading. in a few months tha $200 dollars will get you a card that is better than the 7900 you were originaly aiming for. So what do you do? buy a second 6600?. I wouldn't, quad GPU's are already out (http://www.tomshardware.com/2006/05/02/nvidia_goes_for_four/). By the time you have your next $200 they may have changed the interface to SLI_2.0 or something.

personaly I think spending twice as much for 2 cards to get a 50% increase in speed is a poor investment. its costs more than just money.
It takes at least another SLI socket.
If plan on overclocking your card it can take up another slot each (4 total).
You have to deal with all that extra heat.
If your M/B doesn't have the room, you'll need a new board (more money).

crossfire and SLI to me is a lot like that line from spinal tap
"...but our amps go to 11."

[bartvandenberg]

The same goes for every game. only at a resolution of 1600 x 1200, do you see a real benefit of sli, and even then do you only see 20 fps more. you do not ever see a 50% increase in any benchmark ive ever seen. in fact, if you looked a little more, you would see that most games see roughly the same or even less framerates when set at a lower resolutions.

realisticly, there is no point to having a benchmark for 1600 x 1200, because 95% of people dont have a monitor that will support that resolution anyway. so.. we should be looking at lower res results, and in doing that, you notice that im right on with what im saying.

take for instance, f.e.a.r benchmarks. you see the same/less framerates with a single card as with sli.

this goes for every game, besides splinter cell. so.. it seems to be the only real exception. i know i know, i can see that its not a solid arguement to say it has no benefit, but why would you want a setup that only benefits you half the time?????