AMD Frame Pacing Explored: Catalyst 13.8 Brings Consistency to Crossfire
by Ryan Smith on August 1, 2013 2:00 PM ESTTotal War: Shogun 2
Our first detailed benchmark is Shogun 2, which is a continuing favorite to our benchmark suite. Total War: Shogun 2 is the latest installment of the long-running Total War series of turn based strategy games, and alongside Civilization V is notable for just how many units it can put on a screen at once. Even 2 years after its release it’s still a very punishing game at its highest settings due to the amount of shading and memory those units require.
For the sake of completeness we’re posting our frame rate charts for each of our individual games, but in general there’s nothing here we haven’t seen before in the 7990 review, in other reviews, or in Bench. The 7990 and GTX 690 still swap places fairly regularly.
Looking at our expanded delta percentages for Shogun, we can see how the 7990 and other Crossfire solutions stack up to the GTX 690 and other SLI solutions. For all AFR configurations the results match what we saw in our summary, with NVIDIA’s solutions offering lower deltas than AMD’s even with the new drivers.
This is actually AMD’s weakest game, with both the 7970GECF and 7990 exceeding 20% variability on this game. However it’s also the only non-action game in this collection, so it’s the game least affected by higher levels of variation and consequently the game AMD can afford to do the worst at. Nevertheless the improvement over Catalyst 13.6 without frame pacing is nothing short of amazing.
Meanwhile we’ll hit upon this a few times, but as a reminder AMD’s frame pacing improvements apply to older cards too, so the 6990 has its frame pacing problems resolved like the rest of AMD’s multi-GPU cards. It actually does better than the rest, we believe due to the fact that the lower framerate and higher frame times give AMD’s drivers more time to analyze and schedule frames.
Looking at the FCAT graphs, we can see that the higher variability of the 7990’s frame times is represented well. Though NVIDIA’s frame time spikes are more extreme than AMD’s.
Finally we have our 95th percentile frame times. Despite the fact that AMD’s framerates are down slightly versus Catalyst 13.6, their 95% percentile times are way up. Simply by instituting frame pacing they’ve dropped from 36.2ms to 21.5ms per frame.
Facebook
Twitter
RSS
102 Comments
View All Comments
chizow - Wednesday, August 7, 2013 - link
There was discussions of microstutter on various forums associated with multi-GPU, but PCGH was the first site to publish it's findings in detail with both video evidence and hard data. From what I remember, they were the first to develop the methodology of using FRAPs frametimes and graphing the subsequent results to illustrate microstutter.BrightCandle - Friday, August 2, 2013 - link
One of the most shocking revelations to me is that AMDs quality assurance did not include checking the output of their cards frame by frame. I had always assumed that both NVidia and AMD had HDMI/DVI/VGA recorders that allowed them to capture the output of their cards so they could check them pixel by pixel, frame by frame and presumably check they were correct automatically.Such a technology would clearly have shown the problem immediately. I am stunned that these companies don't do that. Even FCAT is a blatantly blunt tool as it doesn't say anything about the contents of the frames. We still don't have any way to measure end to end latency for comparison either. All in all there is much to left to do and I am not confident that either company is testing these products well, its just I couldn't believe that AMD wasn't testing theirs for consistency (it was obvious when you played it something was wrong) at all.
krutou - Friday, August 2, 2013 - link
AMD is in the business of being the best performance per price entry in every market segment. Technology and quality come second.How often does AMD introduce and/or develop technologies for their graphics cards? The only two that come to mind are Eyefinity and TressFX (100 times more overhyped than PhysX).
Death666Angel - Saturday, August 3, 2013 - link
I think ATI had tessellation in their old DX8 chips. nVidia bought PhysX, so that shouldn't count. But I don't really see how having exclusive technology usable by a single GPU vendor is anything good. We need standardization and everybody having access to the same technologies (albeit with different performance deltas). Look at the gimmicky state of PhysX and imagine what it could be if nVidia would allow it to be fully utilized by CPUs and AMD GPUs?krutou - Saturday, August 3, 2013 - link
Because OpenCl and TressFX are doing so well right?bigboxes - Sunday, August 4, 2013 - link
March on, fanboi.JamesWoods - Sunday, August 4, 2013 - link
If you think that is all AMD/ATI has ever done for graphics then you sir, are ignorant. I was going to use a more degrading word there and thought better of it.Will Robinson - Friday, August 2, 2013 - link
LOL...what a load of tosh."NVDA had to take them by the hand"?
You and Wreckage ought to post in green text.
chizow - Friday, August 2, 2013 - link
Agree with pretty much of all of this, although I would direct a lot of the blame on AMD's most loyal, enthusiastic supporters as well. Every time microstutter was mentioned and identified as being worst with AMD solutions, AMD's biggest fans would get hyperdefensive about it. If those most likely to have a problem were too busy denying any problem existed, it really should be no surprise it was never fixed.And this is the result. Years of denial and broken CF, finally fixed as a result of the scrutiny from the press and laughter of Nvidia fans which brought this to a head and forced AMD to take a closer look and formulate a solution.
EJS1980 - Friday, August 2, 2013 - link
"Truth favors not one side."