MultiGPU Update: Two-GPU Options in Depth
by Derek Wilson on February 23, 2009 7:30 AM EST- Posted in
- GPUs
Who Scales: How Much?
To calculate this scaling data, we simply looked at percent performance improvement of two cards over one. With perfect scaling we would see 100%, while no improvement is 0% and a negative performance improvement means that the multiGPU solution actually produced worse numbers than the single card. There's a lot of data here, so we'll break it down a bit before we present it all.
It is possible to see more than 100% scaling in some tests for different reasons. Fluctuations in benchmark performance can contribute to just over 100% situations, and some times optimizations to enable better multiGPU performance can cut some work out enabling higher performance than would otherwise have been possible. In one of the cases we test today we have a situation where single GPU performance is limited at some framerate while multiple GPUs aren't hindered by the same limit. This artificially inflates the scaling percent.
When looking at games that scale overall, we end up seeing both Radeon HD 4870 configurations (512MB and 1GB) performing worse than we expected. Granted, the 4870 1GB looks better if we only take 2560x1600 into account, but even then the Radeon HD 4850, GeForce GTX 260 and GTX 280 beat out the 4870 1GB in terms of average performance improvement (when performance improves). When we add in CPU limited cases, the 4870 cards look even worse. Consistently, most of the ways we attempted to analyze the magnitude of performance improvement (averages, geometric means, per game, across games where call cards scaled, etc.), the Radeon HD 4850 and GeForce GTX 260 (and sometimes the GTX 280) did pretty well, while the Radeon HD 4870 cards came in pretty low on the list with the 1GB often looking worse because it hit harder CPU limits at lower resolutions.
Hitting CPU or system limits does speak more to value than desirability from a performance standpoint, but it's still important to look at all the cases. Configurations with lower baseline single GPU performance will have more headroom to scale, but these might not always scale enough to be playable even if they scale well. So it's important to take both value and absolute performance data into account when looking at scaling.
We've put all this data on our benchmark pages with the performance data to make it easier to see in context. There just isn't one good way to aggregate the data or we would talk about it here. Depending on the type of analysis we try to do, we could present it in ways that favor AMD and NVIDIA and since there really isn't a "correct" way to do it we've decided to just present the data per game and leave it at that.
Facebook
Twitter
RSS
95 Comments
View All Comments
MamiyaOtaru - Tuesday, February 24, 2009 - link
So we have to be perfect in every way to point out errors? NBA players shouldn't listen to their coaches because their coaches can't play as well as they do? Game reviewers shouldn't trash a game because they couldn't make a better one?ggathagan - Tuesday, February 24, 2009 - link
When it comes to grammatical errors as insignificant as the ones pointed out, yes.If you're going to be that critical, then you best check your own grammar.
cptnjarhead - Wednesday, February 25, 2009 - link
Grammar shmammar, you guys need to move out of your mom’s basement and get laid. :)bigboxes - Wednesday, February 25, 2009 - link
+1stym - Monday, February 23, 2009 - link
I am curious to see how a pair of radeon 4830 would perform in this lineup. A single one is quite weak at those resolutions, but I am willing to bet a pair of those would hold its own against a single GTX280.Oh, and it would be much cheaper, too ($180 including the bridge).
Could you possibly include that setup next?
DerekWilson - Monday, February 23, 2009 - link
You are right that a single 4830 won't be enough perform on par with these guys ... but I don't think two of them would really be worth it against the GTX 280 except maybe at lower resolutions. The 1GB 4830 will run you at least $145, so you're looking at $290 for two of them and the 4850 X2 2GB is the same price. The 512MB 4830 will be limited by memory usage at higher resolutions just like the 4850 512MB.We might look at the 4830 in CrossFire internally and see if it warrants an update, but so far it isn't in the roadmap for the rest of the series.
stym - Monday, February 23, 2009 - link
I was thinking 512MB 4830s, which are in the $90~$110 price range. That price range is the only reason I mention them, because it puts the price tag of a pair of those in the exact same range as a Radeon 4830 512MB or even a GTX260.You said that a 4850 1GB doesn't make sense, and that's even more obvious for a 4830.
pmonti80 - Monday, February 23, 2009 - link
I find too that this would be an interesting match at the $200+ pricetag.wilkinb - Monday, February 23, 2009 - link
why not just drop AoC, it was bad when it came out, has always had issues and odd results and no one i know played for more then 2 months...If you want to have a mmo, why not use one that people play? and maybe even more mature in development...
I know you will say it adds value, but you dont know its it bad code or showing a different view.
ajoyner - Tuesday, February 24, 2009 - link
Most of the issues with the game are gone. There are currently no other MMO's out there that have the graphics or combat system to tax a gpu like this game. Your comment on testing a game that people play is very subjective. There are many MMO's out there that I would not touch....WOW, cough, cough.....but that doesn't mean other people don't enjoy them. I think having this game as one that is regularly benchmarked adds a great deal of value to the article.