Thursday, January 3rd 2013

AMD to Fix GCN Latency Issues with Driver Updates

Last month, an investigative report by The TechReport found out that despite being faster, AMD's Radeon HD 7950 graphics card isn't "smoother" than NVIDIA GeForce GTX 660 Ti, in that shows signs of higher frame-delivery latency, a theory proven by high-speed camera recordings. Over the holiday, AMD's David Baumann responded in discussions around the web talking about the issue, in which he put AMD's stand.

Apparently, AMD Catalyst drivers still have refinement left in working perfectly with GPUs based on the Graphics CoreNext (GCN) architecture. Baumann explained that GCN, and AMD's older Very Long Instruction Word (VLIW) architectures feature fundamentally different memory management, and drivers that make the most of it are still a work in progress. Baumann stated "Over the early part of the year you'll see a few driver updates help this across a variety of games."

He continued "Additionally, when we switched from the old VLIW architecture to the GCN core there was a significant updates to all parts of the driver was needed – although not really spoken about the entire memory management on GCN is different to prior GPU's and the initial software management for that was primarily driven by schedule and in the meantime we've been rewriting it again and we have discovered that the new version has also improved frame latency in a number of cases so we are accelerating the QA and implementation of that."Source: The TechReport
Add your own comment

95 Comments on AMD to Fix GCN Latency Issues with Driver Updates

#1
Calin Banc
To easily detect stutter in Skyrim (or any other game, more so with v-sync on), in MSI Afterburner use the frametime also, not just the FPS counter. You'll notice how in a room, looking in a certain direction, your render time should be 16,6/7ms or so; looking a little bit to the left or right, the counter should show some odd number by 4 numbers or so, changing rapidly - in a screenshot the frametime number would appear as it was in that instance 30+ms in my case. A solution is to set a FPS limit of 58 in MSI Afterburner. A solid 17,2ms should stay for most part of the game (with some micro second burst, mostly unnoticeable). Maybe someone can test that this limit from MAB could eliminate stuttering from other games also?
Posted on Reply
#2
Calin Banc
It looks like limiting your FPS to 59FPS eliminates the stuttering in the games I've played, but of course, it doesn't get rid of the lag. BF 3 at a constant 59FPS and framerendering time is butter smooth. Still, HDAO gives a small amount of lag; not a problem really, it makes the game look worse - SSAO being more than enough.
Posted on Reply
#3
...PACMAN...
Can the CPU used in conjunction with a particular gfx card also have a very detrimental effect with regards latency also? I remember with the same Nvidia card (560 ti) and an AMD processor (phenom II) it had all kinds of weird stutter going on in Mafia 2. With an i5 2500k setup, same HDD, same PSU, same memory it was butter smooth with PhysX enabled throughout the whole game.

Does the CPU memory bandwidth have a major effect? I know it certainly affects Crossfire and SLI setups with regards AMD.
Posted on Reply
#4
Calin Banc
If the CPU is slow enough, I suppose it can hinder a card's performance even on that department. As far as I know, Mafia 2 had some special optimization for PhysX to better run on the CPU. i5 being faster (perhaps on some set of special instructions need for that), it may give better results. You should also take into consideration what other programs run in the background.
Posted on Reply
#6
blibba
by: BigMack70
AMD has made good on their promise
Skyrim was released 14 months ago, and they've now (somewhat) fixed frame latency issues for one architecture in that game. I'm still not impressed.
Posted on Reply
#7
qubit
Overclocked quantum bit
by: blibba
Skyrim was released 14 months ago, and they've now (somewhat) fixed frame latency issues for one architecture in that game. I'm still not impressed.
Agreed. If they can get on top of problems like this and then stay on top of them, then I'll be impressed.
Posted on Reply
#8
BigMack70
Well, it doesn't sound like they were even caring to look at these sort of metrics, so I don't know that I blame them for bad performance in them.

Nvidia definitely beat them to the punch on this one (as often seems to be the case with anything driver-related), but it's good to see them "fixing" it so quickly.
Posted on Reply
#9
qubit
Overclocked quantum bit
I don't really like the test methodology that the tech report uses.

Manual playthroughs, even when averaged over several runs are not as consistent as scripted tests, as TR admits. Variations in frame rendering times like we see are exactly the kind of thing you'll get from this way of testing. Still, the result doesn't seem a million miles off.

Must be a pita to manually play the same bit of game over and over, lol.
Posted on Reply
#10
blibba
by: qubit
I don't really like the test methodology that the tech report uses.

Manual playthroughs, even when averaged over several runs are not as consistent as scripted tests, as TR admits. Variations in frame rendering times like we see are exactly the kind of thing you'll get from this way of testing. Still, the result doesn't seem a million miles off.

Must be a pita to manually play the same bit of game over and over, lol.
Manual playthroughs vs. automated tests have their advantages and disadvantages, and I think arguments can be made for either.

However, I am entirely sold on frame time testing rather than FPS. I've always considered FPS a misleading metric, and I'm delighted that someone is actually measuring something more meaningful. I no longer need to bother with FPS-based reviews, which is ideal.
Posted on Reply
#11
3870x2
by: blibba
Manual playthroughs vs. automated tests have their advantages and disadvantages, and I think arguments can be made for either.

However, I am entirely sold on frame time testing rather than FPS. I've always considered FPS a misleading metric, and I'm delighted that someone is actually measuring something more meaningful. I no longer need to bother with FPS-based reviews, which is ideal.
It is not more meaningful. FPS along with frame latency is useful, frame latency on its own is almost useless.

Edit: In that test, it shows the 660 getting a higher FPS, I thought the 7950 had a higher FPS originally. it was actually originally less.

This is a very interesting discrepancy:

Tech report:


TechPowerUp!:


How did they get a higher FPS?
Posted on Reply
#12
qubit
Overclocked quantum bit
by: 3870x2
How did they get a higher FPS?
Running at a lower resolution and/or quality settings perhaps? TPU tests at 1920x1200 and I'll bet TR tested at 1080.
Posted on Reply
#13
blibba
Just different in game environments could easily make that difference.

I disagree that frame latency on its own is useless.

Say (plausibly enough) all I'm interested in is whether the game is stuttery to play

If there are no frames that take more than 30ms to render, I know that the game is not stuttery. I don't need any more information.

If, on the other hand, the minimum FPS is 30 but I have no latency metrics, I have no guarantee either way.
Posted on Reply
#14
3870x2
by: qubit
Running at a lower resolution and/or quality settings perhaps? TPU tests at 1920x1200 and I'll bet TR tested at 1080.
by: blibba
Just different in game environments could easily make that difference.

I disagree that frame latency on its own is useless.

Say (plausibly enough) all I'm interested in is whether the game is stuttery to play

If there are no frames that take more than 30ms to render, I know that the game is not stuttery. I don't need any more information.

If, on the other hand, the minimum FPS is 30 but I have no latency metrics, I have no guarantee either way.
That is quite a bit of a difference for different environments. There is too little difference between 1080p and 1200p to make for such a discrepancy.

Also fame latency is great, but what if you have excellent frame latency but only 15 FPS?

I do agree however that frame latency is a great thing to add to FPS benchmarks.
Posted on Reply
#15
blibba
by: 3870x2

Also fame latency is great, but if you have excellent frame latency but only 15 FPS?
That's impossible. If I'm only getting 15FPS, I can't be getting a minimum frame latency lower than 66.6 (1000/15), which is much too high.

On the other hand, say I render 1 frame in the first minute, and 23,999 frames in the second minute. That's 200FPS across a two minute benchmark. An extreme example, but just to illustrate my point.

Regarding different environments, in a game like Skyrim the load varies hugely from rolling plains to animation-intensive combat to lighting-intensive indoor areas.
Posted on Reply
#16
3870x2
by: blibba
That's impossible. If I'm only getting 15FPS, I can't be getting a minimum frame latency lower than 66.6 (1000/15), which is much too high.

On the other hand, say I render 1 frame in the first minute, and 23,999 frames in the second minute. That's 200FPS across a two minute benchmark. An extreme example, but just to illustrate my point.

Regarding different environments, in a game like Skyrim the load varies hugely from rolling plains to animation-intensive combat to lighting-intensive indoor areas.
Did TR say specifically why they do free roam rather than a benchmark? A benchmark would bring these numbers closer together with other benchmarks.
Posted on Reply
#17
blibba
by: 3870x2
Did TR say specifically why they do free roam rather than a benchmark? A benchmark would bring these numbers closer together with other benchmarks.
The trouble with benchmarks imo is the risk of drivers being optimised for them. I don't know what TR's specific justification is - ask them :P

by: 3870x2
A benchmark would bring these numbers closer together with other benchmarks.
I don't see that this is necessarily a good thing.
Posted on Reply
#18
Calin Banc
Skyrim, with some mods. Going from Riverwood to Whiterun, looking around, talking with some people, inventory management and level up, and also some messing around in Dragonsreach. I don't know from where those spikes came up, the game is smooth capped at 59FPS to bypass the silly in-game v-sync. Going under that, to 40+FPS in more heavy scenes, of course it will loose some of the fluency, but that is true for every game out there. Oh, and this is with the old beta driver, 1680x1050, 7950@1170/1600, 2500k@4,5GHz

PS: That site I think it tested the game at 2560x1600 or somth.
Posted on Reply
#19
blibba
I suspect a lot of those spikes are from things like opening your menus.
Posted on Reply
#20
jihadjoe
I have to agree with blibba here.

Frame latency, even by itself gives you everything you need to know. Aside from the obvious laggy frame, you can easily compute your FPS from the frame time.

For instance, if all frames are rendered within 16ms, then you know that there's no stutter, and you also know that the minimum frame rate is 60fps. If the frame time is 8ms, then you also know the minimum frame rate is 120fps.

Frame time just simply a much more comprehensive measure of performance than FPS. TR is doing something right.
Posted on Reply
Add your own comment