3-Way CrossFire, Eyefinity and 4K Issues

Based on the previous 8 pages of this article it would appear that AMD has gotten quite a bit right with the new 13.8 driver with frame pacing integration.  Testing at 1920×1080 and 2560×1440 has shown has that frame time variance is way down and the frame rates are much more consistent without the runt frames that we saw with CrossFire configurations for the last half-year. 

But not all is right in the kingdom quite yet.  There are a couple of other hurdles AMD needs to address to perfect the solution.

First, I already mentioned the DX9 limitation and that the current frame pacing implementation won't work with it.  AMD claims that is next on this list and it should be addressed soon.

Second, it would appear that even though AMD claims that 3 GPU configurations should be supported with the frame pacing implementation in today's driver.  But our results showed some complications with that.  The below graphs compare a 2-Way CrossFire Radeon HD 7970 configuration to a 3-Way CrossFire HD 7970 setup.

sleepingdogs-2560x1440-frapsfps-2.png

sleepingdogs-2560x1440-ofps-2.png

sleepingdogs-2560x1440-plot-1-0.png

sleepingdogs-2560x1440-per-2.png

sleepingdogs-2560x1440-stut-1.png

Notice that there is once again a difference between the FRAPS and observed frame rates with the 3-Way configuration that indicates dropped or partially missing frames.  The frame times plot clearly shows some problems – the frame time graph of the 3-Way setup (orange) is less consistent than the 2-Way config and shows some dropped frames and runt frames along the way too.  While still better than the result we saw pre-frame pacing driver, there is a lot of work AMD can do to improve the experience for users tripling their investment.

Finally, the biggest issue that still faces AMD is the lack of improved support for frame pacing with Eyefinity configurations.  Testing with three panel Eyefinity at 5760×1080 results in no improvements to runts or dropped frames.  While I realize that the number of people that are using Eyefinity is pretty small relative to the whole market, these are still the customers investing the most in the PC gaming medium and hopefully AMD will find a solution to the CrossFire + Eyefinity dilemma.

When I asked AMD for more details on WHY Eyefinity wasn't fixed with this release and why it technically was presenting more of a problem, they didn't want to get into it.  My theory still revolves around the compositing engine that AMD is using for CrossFire and the amount of bandwidth it can handle.  Moving a fame of 2560×1600 pixels 60 times a second is taxing but 5760×1080 uses about 50% more pixels is where things seem to break down for AMD.  I will be curious if its even possible for AMD to fix the frame pacing issues with CrossFire and Eyefinity this generation.

Even worse though is that with all the excitement over 4K displays and in particular the ASUS PQ321Q 4K 60 Hz monitor I recently reviewed, the Eyefinity limitation will pop up again in this niche case too. 

4k.jpg

Remember that that PQ321Q 4K display is a tiled monitor, meaning that is supports DisplayPort 1.2 multi-stream transport technology and thus it shows up at two different monitors to the graphics system.  NVIDIA uses a type of Surround technology and AMD uses Eyefinity to enable it – thus the same issue with frame times and frame pacing exists.

« PreviousNext »