AMD Frame Pacing Explored: Catalyst 13.8 Brings Consistency to Crossfire
by Ryan Smith on August 1, 2013 2:00 PM ESTIn Summary: The Frame Pacing Problem
Before we dive into the technical details of AMD’s frame pacing mechanism and our results, we’re going to spend a moment recapping the basis of the frame pacing problem. So if you haven’t been keeping up with this issue, please read on, otherwise feel free to jump a page.
In brief, in multi-GPU setups, be it single-card products like the GTX 690 or multiple cards such as a pair of 7970s, the primary mode of splitting up work is a process called Alternate Frame Rendering (AFR). In AFR, rather than have multiple GPUs working on a single frame, each GPU gets its own frame. This method has over time proven to be the most reliable method, as attempting to split up a single frame over multiple GPUs (with their relatively awful interconnect) has proven to be unreliable and difficult to get working. AFR in contrast is by no means perfect and has to deal with inter-frame dependency issues – where the next frame relies in part on the previous frame – but this is still easier to implement and more consistent than previous efforts at splitting frames.
However due to the mechanisms of AFR, left unattended it can significantly impact the intervals between frames and consequently whether stuttering is perceived. To do AFR well it’s necessary to pace the output of each GPU such that each GPU is delivering a rendered frame at as even a rate as possible; not too soon after the previous frame, and not too late such that the following frame comes up quickly. In a 2 GPU setup, which is going to be the most common, this means the second GPU needs to produce a finished frame when the first GPU is roughly half-way done with its current frame. Should this fail to happen then we have poorly paced frames that will result in perceived micro-stuttering.
Micro-stuttering has been a longstanding issue on multi-GPU setups. Both NVIDIA and AMD have worked on the issue to various degrees, but at the end of the day multi-GPU setups have never proven to be as reliable as single-GPU setups, which is why our editorial position on the matter has been to always favor single powerful GPUs over multiple GPUs when at all possible. Consequently it’s impractical to fully solve micro-stuttering and achieve frame pacing consistency on level with single-GPU setups, but it’s still possible to improve on previous methods and achieve a level of frame pacing that is reasonably effective and “good enough” for most needs. This is what AMD has been focusing on for the past few months.
Moving on, how AMD ended up in this situation is effectively the combination of three factors. The first of course being the innate technical challenged posed by AFR, while the second and third factors have been a poorly realized position on lag vs. consistency and a failure of competitive analysis respectively.
On the former, AMD’s position up until now has been that they’ve favored minimizing input lag in their designs. If you need to hold back a frame to better pace it, then you are by definition introducing some input lag, a quality that is generally undesirable to a user base that usually avoids mechanisms like v-sync for that reason. AMD’s position hasn’t been wrong of course, but it has come at the exclusion of allowing a bit of input lag to better manage frame pacing. AMD’s decision then has been to lighten up on this position and dedicate the resources to deal with both approaches. AMD would introduce advanced frame pacing as an optional control, while leaving the simpler, less laggy approach as another option.
Meanwhile the story with competitive analysis is far less complex. Simply put, AMD wasn’t testing for frame pacing as part of their standard competitive analysis, so when these results first broke AMD was caught flat-footed. This is a business failure rather than a technical failure, which makes it easy enough to resolve. But it’s also the reason why AMD needed time to develop an advanced frame pacing mechanism, as they had never seen the need to develop one before.
Ultimately this is a problem that should have never happened, and it is unfortunate that AMD let it come to this. At the same time however we believe it’s never too late for redemption, and AMD has been making all of the right moves to try to achieve that. They have been clear about their failures and shortcomings, including their frustrations that they’ve left performance on the table by not looking for these issues, and they have been equally clear in laying out a plan for how they would go about fixing all of this. So today we will finally get to see first-hand whether AMD’s initial efforts for resolving frame pacing in multi-GPU setups has paid off.
Facebook
Twitter
RSS
102 Comments
View All Comments
anubis44 - Friday, August 2, 2013 - link
TheJian, you seem to be suffering from verbal diarrea. You might want to take some immodium for that.What you could have said in about 1/10th the space is: you harbour an inexplicable hatred for Ryan Smith, because he's ever said anything positive about an AMD product, and that you think that despite AMD's huge stride forward in one driver revision to address and fix a problem with multi-GPU crossfire smoothness (let's face it, a fairly obscure problem, too), nothing AMD will ever do will be good enough for you, because you harbour an inexplicable hatred of AMD, too.
There, I summarized your entire rant in one sentence. Short and sweet. Concision is bliss.
TheJian - Tuesday, August 6, 2013 - link
https://en.wikipedia.org/wiki/Ad_hominemPlease review the chart from Graham, then come back with something at the top of his pyramid instead of the bottom :)
You seem to be suffering from the inability to make a coherent response to a valid argument, thus attack me instead :) It's always amusing to see fanboys flounder when faced with the facts (no matter if I argue for or against a company, it happens on both sides).
I own a radeon 5850...ROFL. I don't care about NV and will state they suck when or if they do. There is a reason I bought the 5850 :) The only thing I hate about AMD is management taking a total dump on one of my favorite companies (probably mostly due to spending all the R&D on consoles, thus screwing my PC choices and driving them directly into the ground). Having said that, I'll buy maxwell next unless something is terribly wrong with it if only for money backing the drivers. You can google my posts here and see I've been begging people to STOP asking AMD for price cuts and free games so they will start making money. I have done this MANY times. I'm not looking forward to NV owning the gpu world and making it too expensive for me to upgrade as much as I please.
No fan of Ryan or Anandtech these days. I'd hope their alexa traffic numbers forces them to start acting like they did pre Sept last year (which are off by half, as people see the points I and others make). People are not being fooled.
transphasic - Tuesday, August 6, 2013 - link
Excellent points, and well said!The AMD fanboy sure is a grumpy one when their feelings get hurt at the fact that their beloved company has DROPPED the ball AGAIN for the hundredth time.
The loudly proclaim to the world that they finally recognize a problem with their CF setup, and have supplied a minor tweak here and there to get a few games fixed, and that we as AMD owners should start cheering loudly for all through the night.
LOL. It has taken them forever to finally see a problem, and then they take forever to fix it after all this time, even when it hasn't been fixed.
AMD reminds of me Kramer from Seinfeld, who tells Jerry that he will give back to him the pliers he borrowed (that he also broke into pieces and destroyed) but only when Jerry does what he wants him to do, and in so doing, makes Jerry feel like he should be happy about it.
Like Kramer, AMD tries in vain to make us very happy about something that should already had been fixed long ago- just like their Enduro nightmare which STILL after 18 months has not be fixed, and when they do come out with a PARTIAL and incomplete fix- as is THEIR obligation after these many years, we are told to feel happy that we at least got SOMETHING.
They took almost 2 1/2 months to come up with a minor fix for some games, and it's still only Beta, and those with single CPU setups got nothing, all the while Nvidia keeps cranking out the WHQLs, and improved Drivers for a wide variety of games- including AMD-based games.
There's a REASON why AMD is so much cheaper, and far less expensive than Nvida GPU's and Intel CPU's, and we all know why- crappy drivers that are slow to come out, poor attention to detail, weak performance across the board on all their product lines, and a lack of motivation about fixing the problems in a timely manner that they chose to ignore in the first place.
As the saying goes, you get what you pay for...
TheJian - Friday, August 2, 2013 - link
"Ultimately we have to give AMD the kudos they deserve. They have come forward about their issues"No we don't have to give kudos to a company with a beta product who hasn't even fully fixed it yet today. It's like shipping cars with 3 tires. Kudos to the company for putting on a 4th tire for the users today...Seriously? NO WAY. And since it doesn't fix everything (XP users see nothing, eyefinity again nothing), it's really just a 4th tire that is FLAT still...ROFL. They didn't come forward either. They were FORCED INTO THE LIGHT. See PCper's comments in my previous post. They told them he was wrong 1/2 dozen times...ROFL. That isn't coming forward, it's denying you have issues.
"For users who have a reasonable level of faith in Crossfire scaling and are satisfied with AMD’s frame pacing improvements, a $799 7990 is a very good deal at the moment."
If you're stupid enough to still believe BEFORE seeing, well you get what you deserve ;) He keeps printing stuff like this. We're not talking Jesus Christ here (whom I guess you need faith in forever right?), this is a company who can't seem to fix problems that have been dogging them for years (not just since april - they've been claiming they had no problem as hardocp shows this is why NV created FCAT to prove AMD wasn't stutter free for years). They still wouldn't discuss the issues with PCper that are ongoing.
"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."
Shouldn't they be coming forward with what is going on? Still hiding:
http://www.pcper.com/reviews/Graphics-Cards/Frame-...
" 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 2560x1600 pixels 60 times a second is taxing but 5760x1080 uses about 50% more pixels is where things seem to break down for AMD."
Of course 4K won't help this situation as he notes later (I pasted that previously). Be careful if you're just reading anandtech people. Read other sites when AMD vs. NV/Intel is the topic being discussed here. You should NOT buy a A8-5600 as Ian suggests in the 1440p articles for single gpu cards, over Intel. It is foolish as I pointed out in the comments on those articles (and I wasn't alone). I can't believe anyone would recommend AMD over Intel for all but extremely poor people. To recommend it for all single gpu people though is just ridiculous (Titan with a $100 cpu? 780, 7970 etc all show Intel running away BELOW 1440p). I listed the games and links to the articles showing this in the 1440p comments sections. GO LOOK then judge anandtech yourself. I'm not sure what they get promoting AMD, and giving them kudos but I hope it's a LOT of $$. Destroying your reputation isn't worth it IMHO.
DeviousOrange - Friday, August 2, 2013 - link
whine, whine, whine, whine, whine, whine, whine.... yeah you are like a broken record.TheJian - Tuesday, August 6, 2013 - link
But always right :) Thanks for verifying it.gi_ty - Friday, August 2, 2013 - link
Whoa there fella, you shouldn't go out on the internet with your stupidity showing like that.TheJian - Tuesday, August 6, 2013 - link
Personal attacks instead of anything about the data. Shocker.https://en.wikipedia.org/wiki/Ad_hominem
Might want to read that and look in the mirror when done. I don't think you made it out of the pink or orange bar in graham's chart ;)
Come back when you can at least crack the top 3. Then we can talk ;) I'm guessing none of you people took debate class (note I didn't just call you stupid, I'm insinuating you're ignorant) :) It's ok to throw in junk from the bottom of the chart (I'd ignore it anyway most likely) but at least give me something to think about. You know, a valid counterpoint backed by something...Otherwise why bother?
Slugbait - Friday, August 2, 2013 - link
Remember back in the day, when the ideal video setup was a Matrox card paired with a couple of Voodoo2 cards? We were all bashing ATi for their drivers back then.Remember when ATi released the Rage, and it didn't come close to the performance that was advertised? (well...Tom's loved it). ATi said it was because they shipped with beta drivers, because their customers really, really, REALLY wanted the hardware NOW. But every subsequent driver release was "beta", and then they cancelled driver development so they could concentrate on a new line called "Radeon". A lot of people here at Anand's (and FiringSquad, Rage3D, AGN3D, etc) were quite peeved.
Remember when you bought any ATi consumer card for your NT Server machine, only to find out that ATi has never written drivers for NT Server, and you had to use Windows generic drivers (no dual-monitor, etc). Want NT Server support? Buy a FireGL or FirePro.
Remember when your CAD program consistently crashed, but everything was perfectly fine after replacing your ATi card with a card from any other company?
"Catalyst" is often used as a dirty word on the forums here.
They have always known that they write poor drivers. This is not some revelation...this is a public spanking by one of their competitors.
Will they finally wake up and turn things around with their drivers? My confidence is...well, it's kinda low.
boozed - Friday, August 2, 2013 - link
If there're two things I've learned from the internet, it's that Nvidia drivers are terrible, and also ATi drivers are terrible.Meanwhile I've had little trouble with either. Am I doing something wrong?