MultiGPU Update: Does 3-way Make Sense?
by Derek Wilson on February 25, 2009 2:45 PM EST- Posted in
- GPUs
Who Scales
As we mentioned, we'll be looking at scaling from both 1 to 3 and 2 to 3 GPUs. This gives us a few more metrics than last time to look at both overall and on a per game basis. From the overall standpoint, we'll first look at scaling from 1 to 3 GPUs. We'll again look at general success as >33.3% scaling and complete failure will be <5% scaling. This will give us information on how many titles seem to be only CPU limited and how many are of zero or negative value as compared to a single card.
Before we get to the numbers, it is important to note that all of this data is out of 21 tests for AMD cards (like the previous article) but out of 20 for NVIDIA hardware. We had an issue with FRAPS running at 2560x1600 with 3-way NVIDIA solutions. We do want to be clear that the game ran fine, and this seems to be a high res high memory usage issue in Race Driver GRID when FRAPS is combined with 3-way and higher SLI. Let's make it clear that this isn't an issue with the game or the hardware per se, but an issue in combination with FRAPS. 3-way SLI runs really well on Race Driver GRID: we just can't tell you how well unless and until this issue is resolved. We leave the game in this article because there is AMD data at 2560x1600 and because there's still usable data at 1680x1050.
First up in our look at who scales is general success (>33.3% scaling) in moving from 1 to 3 GPUs:
| NVIDIA GeForce GTX 285 | 16 |
| NVIDIA GeForce GTX 280 | 18 |
| NVIDIA GeForce GTX 260 | 19 |
| NVIDIA GeForce 9800 GTX+ | 19 |
| ATI Radeon HD 4870 1GB | 18 |
| ATI Radeon HD 4870 512MB | 16 |
| ATI Radeon HD 4850 | 18 |
The cards that sees the least success in moving from single GPU 3-way are the 4870 512MB and the GTX 285. With the 4870 512MB, this is a combination of failures and CPU/system limited situations while the GTX 285 is purely CPU/system limited here. 16 out of 21 tests isn't hugely different than the 18 or 19 out of 21 (20 for NVIDIA cards), but we do see less "success" in general as compared to our two card situation. And keep in mind that this is 33.3% out of a possible performance improvement of 200%. We are being less strict and seeing less success.
Now lets look at complete failure of scaling from 1 to 3 GPUs. This is based on scaling of <5% and ends up catching the cases of negative scaling. While we did this on a per game basis for 2-way scaling, this time we are looking at the results out of the total number of tests (out of 21 tests for AMD cards and out of 20 tests for NVIDIA)
| NVIDIA GeForce GTX 285 | 0 |
| NVIDIA GeForce GTX 280 | 0 |
| NVIDIA GeForce GTX 260 | 0 |
| NVIDIA GeForce 9800 GTX+ | 1 |
| ATI Radeon HD 4870 1GB | 0 |
| ATI Radeon HD 4870 512MB | 2 |
| ATI Radeon HD 4850 | 3 |
Again, this is generally failure (negative scaling) at 2560x1600 and is generally an issue we can attribute to 512MB of RAM not being enough at high resolution. There are some differences here as compared to 2-way scaling, but generally this isn't that many cases of abject failure to contend with. We'd still love to see AMD and NVIDIA implement something that caught multiGPU failure and reverted to running on a single card in those cases rather than producing a negative experience. But since we can manually disable both SLI and CrossFire, this isn't a deal breaker (it's just an annoyance).
When we look at scaling up from 2 to 3 cards, things get a bit more dim. Since the maximum scaling percent is 50%, we decided to lower our bar for calling a configuration "successful" by reducing our threshold to 10% (which is very generous at only 1/5th of the theoretical maximum). Our results show that much reduced improvement when moving from two to three cards. Here's the data on the number of "success" (>10% improvement) we saw when going from 2 to 3 cards:
| NVIDIA GeForce GTX 285 | 11 |
| NVIDIA GeForce GTX 280 | 12 |
| NVIDIA GeForce GTX 260 | 14 |
| NVIDIA GeForce 9800 GTX+ | 12 |
| ATI Radeon HD 4870 1GB | 13 |
| ATI Radeon HD 4870 512MB | 8 |
| ATI Radeon HD 4850 | 14 |
With the best success rate coming in at 14 out of 20 with the GeForce GTX 260 3-way SLI, and our threshold for success so low, 3-way isn't looking so great out of the gate. Let's take a look at failure to round that out. We'll consider failure to be <2.5% scaling.
| NVIDIA GeForce GTX 285 | 6 |
| NVIDIA GeForce GTX 280 | 5 |
| NVIDIA GeForce GTX 260 | 7 |
| NVIDIA GeForce 9800 GTX+ | 7 |
| ATI Radeon HD 4870 1GB | 5 |
| ATI Radeon HD 4870 512MB | 9 |
| ATI Radeon HD 4850 | 4 |
These numbers show that the majority of the time that cards don't scale up well from 2 to 3 GPUs, they either make no statistical difference or they degrade performance. This is in contrast to scaling from 1 to 2 GPUs. So despite the fact that 3 GPUs can offer good improvement over 1 GPU, it doesn't seem that 3 GPUs consistently offers good improvement over 2 GPUs.
But this is the high level overview. Let's take a look at each game test to get a better idea of what's going on. First we'll recap prices and the test setup and then we'll get to the analysis.
Facebook
Twitter
RSS
46 Comments
View All Comments
MagicPants - Wednesday, February 25, 2009 - link
I've been playing a bit of GTA4 recently, it runs well on my dual 285 system but I've heard there is no SLI support. It might be nice to include a few of these types of games in the mix.Honestly the only game I've played where SLI matters (on 1920x1200) is Crysis.
MagicPants - Wednesday, February 25, 2009 - link
Having the cutoff of 25fps really effected the value of cards. It was interesting to see the values at different resolutions as well.Now I just want to see an interactive graph where I can enter a game and a resolution and it will tell me what video card is the best value. That's not asking too much is it? :)
... or enter a game and resolution and the thing tells me what to put in my system (cpu, memory, motherboard, video card)
plonk420 - Wednesday, February 25, 2009 - link
i'm not even a proponent of SLI/dualGPU until 100% of games work with the technology (and see a worthwhile increase of performance).mastrdrver - Wednesday, February 25, 2009 - link
I think it would have been interesting to see a 2 and 3 way of the 4830 added to all this. Sure it maybe on the lowend of things, but it could have a great value at maybe 1920 and 1680 compared to the more expensive counterparts.stym - Thursday, February 26, 2009 - link
I would like to see that too. I am going to buy a new system next month and I am torn between a single 4870 and two 4830. Same price tag, but what about performance? The problem is, it should have been considered in the previous article. Although I am convinced a two-way 4830 crossfire configuration may provide great performance at a budget price, I doubt a 3-way 4830 makes a lot of sense in a system. You would have to buy a MoBo with three x16 PCI Express slots, and I would not pair that with lower end cards.mastrdrver - Friday, February 27, 2009 - link
It could be a cheap way to go to an i7 platform with power. Spend all the money on the board/memory/cpu and spend ~$300 USD on 3 cards that have a lot of power. If the 4380s scales as well as either the 4850 or 4870, you could have a very powerful but cheap card setup. Not even 300 will buy you a 4870x2. Sure 3 4830s won't beat it, but it will be between a 4870 and the x2. For $300, it sounds like a great deal.Razorbladehaze - Wednesday, February 25, 2009 - link
"Pairing a single card dual GPU AMD card with a single card single GPU option to get 3-way CrossFireX also seems to have a positive impact on microstutter. "I am a little unclear by this statement, I read it as, pairing this combo eliminates the microstutter. But i am concerned that a positive impact could also mean that the FPS in spite of microstutter increases.
This really was the article of most interest to me, as opposed to the 2-way, or 4-way configurations. I find the graphs to be clear and concise with the information they convey.
I find it surprising that there is less discussion on image quality or distortions during benches (yes i know it is difficult to qualitative judge this). I find it hard to believe that these configurations run these game without much flaws, glitches, tearing, flickering in image quality, as my experience has been. I suppose though that if all these issues are resulting from driver optimizations as i suspect, then these commonly benchmarked, newer games get those driver tweaks.
Anyways the only real comments that may be helpful to the actual presentation of material is i agree with the other fellow that the zero point is not contiguous within the graphs. The more accurate the information the better, as opposed to creating a null value, most people understand what is "playable" for their tastes in different genres (at least most people that i believe read these sites). Further I know that my next suggestion is not as mathematically clean as what you have done, but would produce more useful (based upon card prices/selling points) results. Instead of the FPS per $100 spent, change to FPS per $20 or $50 ($50 would be my choice).
Antman56 - Wednesday, February 25, 2009 - link
I wrote an article about it weeks ago. Its a 4850X2 2GB crossfired with a 4850 1GB. Its good.http://forums.anandtech.com/messageview.aspx?catid...">http://forums.anandtech.com/messageview.aspx?catid...
Denithor - Wednesday, February 25, 2009 - link
to the third card option - when the addition of that extra card results in decreased performance? Shouldn't those ones get "0" value ratings?DerekWilson - Wednesday, February 25, 2009 - link
good point ... we'll try and refine it a little more.