• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

800 in sli ?

rippling

rippling

rippling

Soldato
Joined
30 Aug 2006
Posts
2,682
Right i have a dillema , i want to change gfx cards i have herd that sil doesnt scale very well compared to crossfire so buying another 8800gtx prob not the best way to go?

so i need a new mobo anyway , do i get a 3570 or wait for a new mobo that supports 2 3570 x2 ie 4 cores?

or does a second gtx blow everything away?
 
There shouldn't be any difference in the scaling between crossfire and SLI at the most basic level they use the same method for multi-GPU rendering... any difference would be down to driver support and nVidia's SLI support gets better with every driver revision...

DS4 won't support SLI but I guess you already know that.

Quad GPUs in either crossfire or SLI is going to result in noticeable input latency with lower framerates... depending how sensitive you are to it some people would need a constant 60+fps with 4 cards to find the input latency acceptable... which doesn't help much if you need all 4 cards just to get 30fps.
 
keep your gtx for now and just upgrade to a 9 series card when they come out q1 next year ;)
 
Probably won't see much difference between any of the 8800 series cards in SLI... 8800GT might be a bit slower but even the new 8800GTS in SLI should be pretty much as fast as an 8800 Ultra SLI in most games, depending on resolution and settings...

The rumours are that the 9800GTX will be 2.5x faster than the 8800GTX, but so far theres no proof of that.
 
rippling said:
anyone know what sort of performance increase the 9800 will bring?
Click to expand...

they are usually double the performance

and that doesn't mean the same performance as sli GTX, it means twice as good as one card (so a lot better than slid cards)

sli isn't that great overall anyway.
 
Rroff said:
I don't see any comparisons between crossfire and SLI in that thread to backup your claim...
Click to expand...

There is no comparison with SLi there, that shows how well Crossfire scales, which is a lot better than SLi, if you look at SLi reviews, you'll see it doesn't scale anywhere near what Crossfire does, thats almost double performance there, Crossfire is better than SLi.
 
Last edited:
LoadsaMoney said:
There is no comparison with SLi there, that shows how well Crossfire scales, which is a lot better than SLi, if you look at SLi reviews, you'll see it doesn't scale anywhere near what Crossfire does, thats almost double performance there, Crossfire is better than SLi.
Click to expand...

eh? you've just said precisely nothing 4 times in a row...
 
OK this is a 7 series card, not 8 series... but some quick tests using my 7950GX2 in both single and multi-GPU modes, the CPU was an E6600 @ only 3.2gig as I'm still waiting on an RMA on my proper RAM:

Crysis: (1920x1200, ini tweaked for ultra high, 0x FSAA, 16x AF)

TimeDemo Play Started , (Total Frames: 2000, Recorded Time: 111.86s)
!TimeDemo Run 0 Finished.
Play Time: 389.86s, Average FPS: 5.13
Min FPS: 2.63 at frame 1658, Max FPS: 6.29 at frame 63

TimeDemo Play Started , (Total Frames: 2000, Recorded Time: 111.86s)
!TimeDemo Run 0 Finished.
Play Time: 198.19s, Average FPS: 10.09
Min FPS: 4.67 at frame 137, Max FPS: 13.84 at frame 786

Showing a 96.7% gain in SLI mode on the average, as min and max were on different frames between runs a direct comparison isn't easy but again they show approx 90-100% gains.

ETQW: (1920x1200, 4x FSAA, 16x AF, all ultra)

Single Card: 43.4 avg
Multi Card: 80.3 avg

Showing an 85% gain.

HL2 - Lost Coast stress test: (1920x1200, 6x MSAA, 16x AF, all max in menu, reflect all)

Single Card: 65.44
Multi Card: 154.62

This one shows an anomally... I would need to run the benchmarks multiple times at each mode to get an average for comparison as the results change slightly each time but its good enough to show that the gains are close to 100%.

If your not getting good scaling with SLI with the 8 series its probably due to either: CPU bottleneck, other bus bottleneck or you just don't know what you are doing... the cards and drivers are capable of just as good scaling as crossfire if not better...
 
This suggests they scale at different rates. At a g;ance, Crossfire scales slightly better than SLI, but it would take some good eyes to notice the difference.

I really do wish people would stop posting rumours as to future generation performance. There were rumours about the GeForce 5800, the Radeon 1950XT/2900XT/etc. and they all performed quite significantly below expectations... simply because those expectations were based on absolute nonsense.

I will wait until I see how much faster future generation cards are before voicing an opinion on their actual performance.
 
Thats hardly a good test to give a comparison in scaling... except for Oblivion the 8800GTX is blowing the 2900XT out the water and is probably hitting CPU bottlenecks artificially limiting its percentage gains...
 
Ok folks I have a 2x3870 in Crossfire on a X38 today and will receive a 780i SLI board tomorrow. I've got either 2 8800GTXs or 2 8800GTS 512 to put in, not decided which yet so will do some direct compares . I'll be running at 1920x1200 and probably using Vista 64 or 32 and UT3, Crysis and World in Conflict.
 
J1nxy said:
Ok folks I have a 2x3870 in Crossfire on a X38 today and will receive a 780i SLI board tomorrow. I've got either 2 8800GTXs or 2 8800GTS 512 to put in, not decided which yet so will do some direct compares . I'll be running at 1920x1200 and probably using Vista 64 or 32 and UT3, Crysis and World in Conflict.
Click to expand...

how do the 2 3870 run in crossfire with crysis?
 
one 3870 is about the same performance as the 7950GX2, so if you use my example above, 2 of them in crossfire with high res, ultra high settings would get around 20fps average... while that might not sound like much 2x 8800GTX in SLI with same settings would only be getting around 35fps.
 
FEAR:
C/Fire- 73%
SLI - 95%

FarCry:
C/Fire - 70%
SLI - 54%

Oblivion:
C/Fire - 106%
SLI - 51%

Prey:
C/Fire - 54%
SLI - 49%

Supreme Commander:
C/Fire - 83%
SLI - 15%

X3:
C/Fire - 65%
SLI - 29%

No DX10, but that's not the point.

It would appear that the CrossFire implementation of the 2900XT scales much more efficiently that the SLI'd 8800GTX in everything bar Fear. Could it be that the games are CPU-limited with the GTXs in Crossfire? Possibly, but difficult to isolate considering the games are being benched with 8xAA and 16xAF at 1920x1200.

Now, the GTX wins pretty much every benchmark, but again, that's largely irrelevant when discussing scaling. It would appear the 2900XT scales much more effectively when CrossFire'd with another similar card. That doesn't mean it a good-value proposition, but it means that (soft- or hardware) architecture is potentially more efficient.

And I'm not quite sure which benchmarks you're looking at, because the ones in the review I posted clearly have Crossfire'd 2900XTs outperforming the SLI'd 8800GTXs at both 1920x1200 and 2560x1600 with the IQ settings as posted above...?
 
You must log in or register to reply here.
Back
Top Bottom