About Crossfire scaling...

gran172

Reputable
Jan 22, 2016
327
1
4,810
I was seeing some R9 380 CF benchmarks and i saw this:
1080p: http://i0.wp.com/www.eteknix.com/wp-content/uploads/2015/11/bf414.jpg?w=800
1440p: http://i1.wp.com/www.eteknix.com/wp-content/uploads/2015/11/bf424.jpg?w=800

My question is, why is the min on the 1080p gameplay so low compared to the 1440p one?

Also, if somebody had experiences with crossfiring AMD GPU'S i'd love to hear pros and cons, i've seen lots of bad comments about amd's crossfire regarding their bad drivers but most of them are from like 2013, how are crossfires working now?
 
Solution
This is usually due to the crippling nature of game works but ultimately no they are about the same in most cases. However crossfire scaling is better than sli.
The 1080 bench has to be an error.
There is no synergy here.

AMD crossfire is less reliable than Nv. Still true today. Results can vary massively between games, so best find benchmarks for the games you'll play.

Pros: Mid range: Can add an extra card (often at a lower cost than when you bought the original), which makes it cheaper than selling an old card to buy a new one.
Pros: High range: Allows you to play at resolutions not not achievable by a single card.

Cons: 1+1 = 1.8 (at best)
Double power consumption and heat.
No additional ram
Variable results between games
Selling two cards down the road is harder.
 
I couldn't disagree more.. Cross fire is about as reliable as sli meaning when it works it works well and when it doesn't it's a mess even if it works at all. Crossfire today is pretty good assuming the game gets the support it needs but it's the same for nvidia. Some games have support at launch, others don't and some times the feature is broken due to an update. Crossfire scanning is actually better than sli though and there are plenty of sites that report that.
 
Thanks for the response man, what i don't quite understand is AMD's CF being less reliable than Nv SLI, is that because AMD doesn't optimize games properly in their drivers (even for single gpu rigs) or because CF scales worse than SLI?
 
Technical scaling advantages are meaningless if the drivers aren't optimised for a specific game.

Project Cars for example, there were massive complaints from AMD users experiencing terrible performance. Turns out AMD didn't respond to a single email from the developer. Meanwhile Nv assisted heavily.

Until recently I had to use AMD beta drivers for GTA as they didn't fix gfx glitches for months.

AMD are getting better, but this sort of thing happens regularly.
Then, you have to remember they are running two entirely different card architectures atm. The 3xx series which are basically ancient, and the Fury range which are quite advanced. Both of these perform differently.
The 3xx cards cannot be OC'd, while the Nv 9xx series are well known for OC-ing AND lower power consumption.

These issues don't magically disappear by adding a 2nd card.

As said, it's ALL game dependent.
 


Ah! That's a great example... Project cars... another example of the crippling effects of another Game Works title. So I agree it is very dependent on the game.