AMD Tweaking Radeon Drivers To Reduce Frame Latency Spikes

crookedvulture writes "Slashdot has previously covered The Tech Report's exposure of frame latency issues with recent AMD graphics processors. Both desktop and notebook Radeons exhibit frame latency spikes that interrupt the smoothness of in-game animation but don't show up in the FPS averages typically used to benchmark performance. AMD has been looking into the problem and may have discovered the culprit. The Graphics Core Next architecture underpinning recent Radeons is quite different from previous designs, and AMD has been rewriting the memory management portion of its driver to properly take advantage. This new code improves frame latencies, according to AMD's David Baumann, and the firm has accelerated the process of rolling it into the official Catalyst drivers available to end users. Radeon owners can take some comfort in the fact that a driver update may soon alleviate the frame latency problems associated with AMD's latest GPUs. However, they might also be disappointed that it's taken AMD this long to optimize its drivers for the now year-old GCN architecture."

Comments

Give them credit

[Anonymous Coward]

taken AMD this long to optimize its drivers for the now year-old GCN architecture.

Give them some credit... they've acknowledged the problem and this isn't a simple tweaking/bugfix, this is a complete redesign and rewrite of the entire driver architecture.

Re:Give them credit

[hattig]

Well, it's in the memory manager portion of the driver. Memory management isn't easy at the best of times, and when you're dealing with a GPU that has thousands of cores, and each of those cores has its own local memory, and shared memory with a local cluster group, and then there are software controllable caches further up the hierarchy, I can see how writing this code could be fraught with difficulty.

And as many of us here have worked in professional software environments, I'm sure we can all see how something that was pretty hard to pin down like these latency spikes might not have been a top priority for development, even if they were aware of it at all - after all the FPS figures were great. You'd end up with a driver kernel that had some magic that nobody would want to touch, and most of the work would be game specific optimisations and higher level optimisations. A year sounds about right really.

Re:Give them credit

[Anonymous Coward]

More than that, this frame latency issue is new to the recent drivers - possibly even restricted to the 12.11 beta drivers, which one might expect to have the occasional issue. Further, other review sites that investigate the smoothness of gameplay, even if not by quite the same method (ex: Hard|OCP), have not found the issues that Tech Report has, which leads me to suspect that it might not be a universal issue.

I think that Tech Report has been very irresponsible in their handling of this issue. They've been quick to condemn, ignoring that the driver is _not_ a final version, and apparently unwilling to investigate the disparity between their older results with the same cards/older drivers (which had no such frame latency issues) and their new results. Frankly, I'm not willing to give them any regard as a review site any longer - their frame latency methodology certainly has merit, but they seem to be utterly incompetent in actually analyzing or investigating the results.

Re:Give them credit

[nanoflower]

Only one problem with what you suggest is that it is based on bad information. TR made the effort to look at different versions of the drivers and they've tested it on Win7 and Win8. Also only a couple of other sites have done the same level of testing frame rates that TR has been doing and they've found the same issues. Then you add in that AMD has looked into the issue and acknowledged there is a real issue that they need to address. So you are doing a disservice to Tech Report by misstating the situation and ignoring the other sites that have agreed with their findings.

Re:Give them credit

[Impy the Impiuos Imp]

One would think an automated test that looked at timestamped frame production would show the jumpy nature of it.

Re:Give them credit

[Anonymous Coward]

I don't doubt their findings, merely that the latency spikes are a universal problem, and that they weren't recently introduced in a _beta_ driver (or at the least, a newer one - we know the problem didn't exist back in earlier versions).

I wouldn't be surprised to find out that this is a result of the chipset/disk controller driver (which they updated in the interval between their older stutter-free results and their new stutter-heavy ones) interacting poorly with the newer graphics drivers, and possibly the older ones as well - disk reads can be a major culprit in stutter, particularly in all these console ports which use texture streaming heavily. It's unlikely, but since disk reads are one of the biggest sources of notable stutter, it merits investigation.

Or it could be a hyper-threading problem, as hyper-threading is known to cause stutter in some cases, and some, if not all, of the testers who haven't noticed stutter issues despite casting an eye towards smoothness of frame delivery are using processors without hyper-threading. This would be particularly easy to investigate, and they didn't even try.

Win7/Win8 was a reasonable thing to investigate, but that only serves to highlight their utter irresponsibility in not even commenting on the possibility in the original article, or actually investigating it before publication like any responsible reviewer should when faced with such remarkable results.

The problem here is _not_ the data, it's the lack of questions they've asked about the wildly different data, and the lack of any sort of scientific rigor applied to a very surprising result.

Re:I read: AMD updates video driver

[loufoque]

It's news because it's not a simple optimization. It's fixing something that was fundamentally broken.

I can confirm this

[Billly Gates]

I have the latest beta cataylst drivers from December 3rd and SWTOR is now fluid when I turn. Also video seems better too. Also worth mentioning is this and the last stable released fix the massive security hole by disabling protected mode in Vista/7 with ASLR.

I highly recommend ATI users upgrade their drivers as I found the beta more stable than the stable one.

Re:Reach for your wallet.

[Billly Gates]

My el cheapo laptop from 2007 has WDDM 1.1 support for Windows 7 accelerated. I was impressed with that as this box was designed for Vista not 7. Also the driver version is from last March so ATI supported it for well over 5 years.

Re:I thought..

[Billly Gates]

They make great dedicated video cards.

I have noticed this issue on my ATI 5750 dedicated card last year.

The CPUs are so-so unless you are looking for a cheap multicore system with virtual machines. Intels are better for single tasking and also multitasking if you have a large budget. Unfortunately, this is no market for cheap multicore heavy multitasking users except for a few geeks.

Thew new beta drivers work for me and it is a big improvement.

Technology sites give brainddead measurements

[PhrostyMcByte]

This only went on so long because tech sites use such poor, useless benchmarking methods. Minimum/Average/Maximum FPS, or often just Average/Maximum FPS, are worthless!

A game, or a video card, can average 100fps, but still have that one frame every second that performs some extra I/O and takes 3x longer than usual causing an annoying stutter effect.

A good first step would be to use frame latency percentiles.. i.e. 90% of frames are at least 60 FPS, 95% of frames are at least 50 FPS, 99% of frames are at least 40 FPS.

The next step is to measure spikes themselves -- low framerate sucks, but not nearly as much as a stuttering framerate. A sudden spike from constant 10ms/frame to 50ms/frame and back should be counted as far more detrimental than a smooth transition from constant 10ms to constant 25ms.

Re:Technology sites give brainddead measurements

[hattig]

You should be reading the Tech Report GPU reviews then, they have all that stuff included.

Re:Technology sites give brainddead measurements

[TubeSteak]

This only went on so long because tech sites use such poor, useless benchmarking methods. Minimum/Average/Maximum FPS, or often just Average/Maximum FPS, are worthless!

Umm... Tech Report has been doing frame latency benchmarking for over a year now.
And I'm not sure if they were the first ones to come up with the idea, but I know they're not the only reviewers using that benchmark.

Re:I read: AMD updates video driver

[citizenr]

Because this should have been done 9 months ago. Leave it to AMD to once again just drop the ball. At least they're consistent at failing.

Better 9 months for software patch than 5 years for process change and MASSIVE GPU die off Nvidia gave us starting with 8xxx models.

Ordered Nvidia

[watermark]

I had a newish card, AMD 4xxx HD, but they force me to use the fgrlx-legacy driver. The fgrlx-legacy drivers don't work with the newest xorg, and the ones that work with the older xorg are missing features essential for steam. AMD scaled down the Linux team recently. I just bought an Nvidia.

Re:Ordered Nvidia

[Sable Drakon]

The HD4000 series isn't a newish card by any stretch of the imagination. Newish is the 5000 or 6000 series, and yes, AMD's Linux support has always been pathetic at best.

Re:Ordered Nvidia

[cbhacking]

The Radeon HD 7xxx series cards, such as the 7950 in my PC, are manufactured using a 28nm process. It's actually smaller than what AMD uses for their CPUs, oddly enough (the current CPUs are 32nm). 40nm is a couple generations back.

Re:I read: AMD updates video driver

[citizenr]

Clearly you are unaware of Nvidia fiasco and following litigation. It wasnt "share of the units". It was Majority of them. Basically finding a working laptop with nv8xxx/9xxx GPU is considered lucky (they ALL die sooner or later, ticking bombs), and there are companies doing nothing else but fixing them.

Re:I read: AMD updates video driver

[LordLimecat]

AMD didnt have a class action lawsuit requiring them to purchase new laptops for end users involved in the claim.

Bullshit; OP is right.

[cbhacking]

"Fundamentally broken" is BS. It would be fundamentally broken if the cards didn't *work* with the old drivers. You *might* even get away with that claim if the cards were unplayably slow, or the spikes were serious enough to make Windows think the driver had stopped responding, or some other such problem. "Benchmarks show that there is a specific performance problem, even though the overall performance is quite acceptable, and the issue was traced to an un-optimized memory manager which nevertheless was working correctly" is not even to "broken" by any reasonable definition, much less "fundamentally" so.

Re:Bullshit; OP is right.

[loufoque]

You know nothing of writing a rendering pipeline.

Re:Yet another reused abbreviation

[dstyle5]

GCN stands for "GameCube Nintendo" and as far as I recall did not describe any particular hardware component of the console.