New AMD Drivers Coming July 31st

[KazeoHin]

I'm waiting for the mature driver that supports single-cards and eyefitity.

 

[sushiwarrior]

I'm waiting for the mature driver that supports single-cards and eyefitity.

Supports WHAT on single cards? Lawl, there aren't any real issues (besides maybe FC3?) on single cards, what do you want it to fix

 

[veedubfreak]

I actually don't know what's wrong with his setup- I'm expecting him to come back and tell us what's wrong.

I'm fairly sure there's a mismatch regarding outputs between a GTX690 and three 4MP panels. My only other point is that it's an awful idea- the card is ill-equipped for that setup, as one GTX690 is barely enough for a single 2560x1600 monitor, and the card only has 2GB per GPU.

That was exactly my point. My 7970 spans my 3 30" monitors just fine and doesn't fight me at all, the nvidia drivers tell me to fuck off because even though all 3 monitors are 25x16 @60hz "they are not the same resolution".

And as soon as the new drivers fix crossfire for eyefinity I'll be buying a 2nd 7970.

 

[Sheppard]

As far as testing on my 7990 goes, I have absolutely nothing to report. I can not notice a conceivable difference in 3dmark scores. I'd post screenshots but on my setup they would look almost identical.

Tomb Raider looks AMAZING though!

 

[Tanquen]

only certain games suffer from runt frames/micro stutter with CFX, and even then it's only with VSYNC disabled. Crysis 3 doesn't have any issues with VSYNC disabled, but Sleeping Dogs is absolutely terrible. Same with Far Cry 3, awful...

Far Cry 3 = Still bad.

 

[Unknown-One]

yeah, you have a single 6970. The drivers are for crossfired cards. It's all in the release notes.

So when are they going to fix the horrible stuttering that single-cards are still dealing with?

Supports WHAT on single cards? Lawl, there aren't any real issues (besides maybe FC3?) on single cards, what do you want it to fix

As I noted above, Fallout 3 and Skyrim both experience periodic framerate spikes on single-card AMD setups. These aren't the only games doing this, but they're the one's I have readily available at the moment.

 

[sushiwarrior]

This fix worked for this guy. Other people are saying it's a CPU problem, low CPU utilization causing the CPU to go into idle states, maybe disabling power states in the BIOS would fix it.

 

[DogsofJune]

I have zero issues with Skyrim, FC3, either Fallouts with or without crossfire anymore.
When I did have issues, I just un-enabled crossfire. I have never had any single card issues with any those games.
First I have heard, or at least paid attention to, a issue with a single card.

 

[Elmy]

This driver does not work for Eyefinity.

running 5400x1920 and this driver is a no go.

Waiting patiently for this driver to work for Eyefinity peeps.

 

[Unknown-One]

This fix worked for this guy. Other people are saying it's a CPU problem, low CPU utilization causing the CPU to go into idle states, maybe disabling power states in the BIOS would fix it.

Not a CPU problem, I can assure you of that. It's exclusive to AMD cards as far as I can tell (tested on an HD 5850, HD 5870, and HD 6970).

Didn't have these hitching problems on my old GTX 260, nor do I have them on my new GTX 780.

 

[sushiwarrior]

This driver does not work for Eyefinity.

running 5400x1920 and this driver is a no go.

Waiting patiently for this driver to work for Eyefinity peeps.

My god, are you capable of reading.

Not a CPU problem, I can assure you of that. It's exclusive to AMD cards as far as I can tell (tested on an HD 5850, HD 5870, and HD 6970).

Didn't have these hitching problems on my old GTX 260, nor do I have them on my new GTX 780.

See above ^^^ did you read the link? Disable AA in game, try forcing it through CCC instead.

 

[IdiotInCharge]

That was exactly my point. My 7970 spans my 3 30" monitors just fine and doesn't fight me at all, the nvidia drivers tell me to fuck off because even though all 3 monitors are 25x16 @60hz "they are not the same resolution".

And as soon as the new drivers fix crossfire for eyefinity I'll be buying a 2nd 7970.

But you still haven't said why. There's a reason for it other than 'Nvidia hates you'. I know people have done 3x2560x1600 on Nvidia.

 

[Nicterys]

Small sample from last night: Far Cry 3, Crysis 3 and Bioshock Infinite were vastly improved by frame pacing(even on Eyefinity) but Tomb Raider was very choppy.

 

[Unknown-One]

See above ^^^ did you read the link? Disable AA in game, try forcing it through CCC instead.

Tried turning AA off in-game, no difference on the 6970. Didn't really expect that to make a difference considering only FXAA was enabled (which is just a shader). MSAA was off.

Only thing I've found that helps AMD cards in Skyrim is using a framerate limiter (and this doesn't 100% eliminate the hitching, but it does reduce it massively). Unfortunately, using such a limiter causes internal engine sync issues that cause some seriously glitchy behavior after playing for a while.

 

[Tamlin_WSGF]

Tried turning AA off in-game, no difference on the 6970. Didn't really expect that to make a difference considering only FXAA was enabled (which is just a shader). MSAA was off.

Only thing I've found that helps AMD cards in Skyrim is using a framerate limiter (and this doesn't 100% eliminate the hitching, but it does reduce it massively). Unfortunately, using such a limiter causes internal engine sync issues that cause some seriously glitchy behavior after playing for a while.

Skyrim is pretty good these days. Both that and Fallout 3 is much smoother on my 7970 then my GTX 570 ever was. More GPU power, but still ...

Actually, it was pretty good, something got borked in the driver, then it got fixed again:

Skyrim frame times 2011:
http://techreport.com/review/22048/today-mid-range-gpus-in-skyrim/9

12.11 beta 8 (where skyrim was borked on AMD GPU's) and 13.2 beta where Skyrim is smooth again:
http://techreport.com/review/24218/a-driver-update-to-reduce-radeon-frame-times/2

Can't remember Techreport using skyrim later, but here's a recent benchmark with frametimes on several other games:

http://techreport.com/review/24996/nvidia-geforce-gtx-760-graphics-card-reviewed/10

For a time, Nvidia still held an edge in our latency-focused tests, until AMD addressed some issues with its drivers and recaptured the lead.

I haven't tried the 6970 series from AMD, but single 7970 is smooth as butter in games.

 

[Unknown-One]

Skyrim is pretty good these days. Both that and Fallout 3 is much smoother on my 7970 then my GTX 570 ever was. More GPU power, but still ...

I literally have a machine with an HD 6970 (and one 1920x1200 monitor) right next to a machine with a GTX 780 (and three 1920x1200 monitors in Surround) both running Skyrim.

The computer with the 6970 is hitching, the computer with the GTX780 is not.

Swapping the cards between machines = the problem follows the AMD card.

I haven't tried the 6970 series from AMD, but single 7970 is smooth as butter in games.

Quite possible it's only fixed for 7000 series cards. Wouldn't be the first time AMD has left older cards for dead

Edit: Looks like this driver doesn't help a single 7970 much either, check out its frame times in Tomb Raider and compare them the single GTX 680 in the table. The 7970 shows variance, even though it's not in a crossfire config. Looks like single-card systems are still in need of fixes.

 

[Tamlin_WSGF]

I literally have a machine with an HD 6970 (and one 1920x1200 monitor) right next to a machine with a GTX 780 (and three 1920x1200 monitors in Surround) both running Skyrim.

The computer with the 6970 is hitching, the computer with the GTX780 is not.

Swapping the cards between machines = the problem follows the AMD card.


Quite possible it's only fixed for 7000 series cards. Wouldn't be the first time AMD has left older cards for dead

Perhaps the computer, perhaps its like you say that this haven't been fixed for 6970. This is not true for 79X0 however, as I have shown with frame time tests from Techreport and also seen myself from my own observations between GTX 570 and HD 7970.

I've gone through 2-300 games the last months with different drivers and games are smooth with a single 7970. The FCAT tests on single cards supports my observations as well.

 

[Tamlin_WSGF]

Edit: Looks like this driver doesn't help a single 7970 much either, check out its frame times in Tomb Raider and compare them the single GTX 680 in the table. The 7970 shows variance, even though it's not in a crossfire config. Looks like single-card systems are still in need of fixes.

Look at the picture again. The light one is Fraps, the other one is FCAT. I think you are reading the info wrong.

And here is Tomb Raider from Techreport on an older driver (13.5 beta 2):
http://techreport.com/review/24996/nvidia-geforce-gtx-760-graphics-card-reviewed/6

Smooth

 

[Zuul]

So when are they going to fix the horrible stuttering that single-cards are still dealing with

I don't know about single cards but hardOCP said this:

Phase 2 of the driver will come later this month. In Phase 2 CrossFire in Eyefinity resolutions will be supported. There will also be support for DX9 and OGL games in Phase 2. So if you game in Eyefinity, or across multiple displays, or higher resolutions than 2560x1600 then Phase 2 driver is what you will have to wait for. Also, if you want support in DX9 games.

 

[Unknown-One]

Look at the picture again. The light one is Fraps, the other one is FCAT. I think you are reading the info wrong.

Nope, I'm comparing to the Nvidia results.

Nvidia shows no variance in either test, AMD shows pretty horrible variance in one of the two tests. There's still a huge difference in how these cards are getting frames rendered, even if AMD is trying to fix it or mask it.

 

[Tamlin_WSGF]

Nope, I'm comparing to the Nvidia results.

Nvidia shows no variance in either test, AMD shows pretty horrible variance in one of the two tests. There's still a huge difference in how these cards are getting frames rendered, even if AMD is trying to fix it or mask it.

Fraps captures in (ipresent). A lot of things happen after that. FCAT captures the output itself. Seriously, statements like "AMD is trying to fix or mask it" is crap that doesn't belong here. You are reading this like the devil reads the bible.

Read up on fraps vs. FCAT:

Above Hitman Absolution with the very same AMD GPU and drivers. Notice how all the weird spikes are gone? That's the advantage of FCAT, you record what you see on screen and this is in line with what we see on the monitor. So what does that say about FRAPS ? Yeah, it indeed can show a lot of stuff that really isn't there and perhaps vice versa. That doesn't mean though it is recording non-exsistant stuff, contrary. But is is inserting and leaving out stuff we can not observe visually.

http://www.guru3d.com/articles_pages/fcat_benchmarking_review,8.html

Being trendy - websites jumped onto it and are using FRAPS, but the thing is... if you want to expose game stutters you probably should not use FRAPS. Here's why:

http://www.guru3d.com/articles_pages/fcat_benchmarking_review,1.html

 

[Ozymandias]

Nope, I'm comparing to the Nvidia results.

Nvidia shows no variance in either test, AMD shows pretty horrible variance in one of the two tests. There's still a huge difference in how these cards are getting frames rendered, even if AMD is trying to fix it or mask it.

Going by the FCAT results, both show a frame time variance of ~10 ms. The 7970 actually looks like it's spending most of the time in the 30 ms range, with spikes into the 40 ms range. The 680 has a similar plot, but ~10 ms higher (ie, 10 ms worse). I don't think those graphs are really helping your point here.

 

[Unknown-One]

Read up on fraps vs. FCAT:

I know the difference between FRAPS and FCAT, just not seeing how it's relevant to the point I've made.

Nvidia's results still seem to be far better than AMD's here, considering Nvidia's results don't show a problem ANYWHERE in the chain, where as AMD's do.

And I'm still getting hitching on the newest AMD hardware I have access to. Kinda understandable why I'm not happy with them -_-

 

[sushiwarrior]

I know the difference between FRAPS and FCAT, just not seeing how it's relevant to the point I've made.

Nvidia's results still seem to be far better than AMD's here, considering Nvidia's results don't show a problem ANYWHERE in the chain, where as AMD's do.

Lol, the only part of the chain which is showing problems is buggy, irrelevant, and not visible to the end user. pleaaaaaase, troll harder

 

[Unknown-One]

Lol, the only part of the chain which is showing problems is buggy, irrelevant, and not visible to the end user. pleaaaaaase, troll harder

Not trolling, pointing out a pretty obvious difference...

Once again, I'm sitting at a machine with an HD 6970 in it RIGHT NOW... and I'd like it to work properly.

Seriously, statements like "AMD is trying to fix or mask it" is crap that doesn't belong here.

If you'd look at what I was pointing out it might make a bit more sense. There's obviously something still causing high levels of variance the pipeline on AMD and it's masked by the time FCAT gets hold of it.

 

[Tamlin_WSGF]

I know the difference between FRAPS and FCAT, just not seeing how it's relevant to the point I've made.

Nvidia's results still seem to be far better than AMD's here, considering Nvidia's results don't show a problem ANYWHERE in the chain, where as AMD's do.

I don't think you do understand the difference. You are speaking of a variance that doesn't even show up on output. A lot of shit happens after "present" where fraps captures its info. All that shit is done differently in Nvidia drivers and AMD drivers. How they do it, doesn't matter, since its just the method before final output. That you call it a problem in the chain, when FCAT clearly shows that this wasn't a problem, shows you didn't understand what you were looking at.

To quote another passage from the link at guru3d and please read this:

So above two example FRAPS Frametime recordings. You'll agree with me that when looking at the Radeon HD 7790 you'd expect a downright horrible game experience. The problem with FRAPS is this, it can show stuff that is not relevant - most of these weird spikes you see do not relate to what you see on-screen. So the interesting thing is that once you play and visually look at the games being rendered by say AMD, you (for the bigger part) just might not notice or see the latency spikes.

 

[Tamlin_WSGF]

If you'd look at what I was pointing out it might make a bit more sense. There's obviously something still causing high levels of variance the pipeline on AMD and it's masked by the time FCAT gets hold of it.

There is variation early in the pipeline, which obviosly is not a problem for the driver as shown by FCAT, so frame times are smooth. Perhaps Techreport shouldn't use fraps as well, since people get confused and think that this is a problem even though it clearly shows the driver can handle it pretty well.

 

[Flopper]

drawing conclusions without understanding the software =user error.

 

[Unknown-One]

I don't think you do understand the difference. You are speaking of a variance that doesn't even show up on output.

I never said it showed up in the output, I simply said it was there... which it is.

drawing conclusions without understanding the software =user error.

Only conclusion I drew, and pointed out, is that AMD's hardware is still showing high levels of variance in 1 out of 2 tests, while Nvidia shows it in 0 out of 2 tests. Nothing more, nothing less.

 

[Tamlin_WSGF]

I never said it showed up in the output, I simply said it was there... which it is.

But you did say:

Looks like this driver doesn't help a single 7970 much either, check out its frame times in Tomb Raider and compare them the single GTX 680 in the table. The 7970 shows variance, even though it's not in a crossfire config. Looks like single-card systems are still in need of fixes.

When it clearly shows this driver makes the output very smooth on the single 7970 and doesn't need to be fixed.

You saw a variance, assumed this was a problem (even though the same image shows you this wasn't a problem for the driver to handle) and went on a spree saying the driver needs to be fixed.

I don't think you understand the difference between FRAPS and FCAT. If you did, you wouldn't have posted that.

 

[Unknown-One]

But you did say:

Looks like this driver doesn't help a single 7970 much either, check out its frame times in Tomb Raider and compare them the single GTX 680 in the table. The 7970 shows variance, even though it's not in a crossfire config. Looks like single-card systems are still in need of fixes.

When it clearly shows this driver makes the output very smooth on the single 7970 and doesn't need to be fixed.

Again, nowhere there did I say it appears in the output, just that I still think single-card configurations need some fixes.

I also offered up my own 6970 as an example, where I'm still seeing hitching (not that the hitching is related to the above charts, just that it's still a problem, and is something else in the "single cards still need fixes" category).

You saw a variance, assumed this was a problem (even though the same image shows you this wasn't a problem for the driver to handle) and went on a spree saying the driver needs to be fixed.

I see variance and still assume there's a problem, one AMD has masked in software by the time FCAT results are generated (which means smoother visuals, but doesn't necessarily mean everything is all rosy). Treating the symptoms is a good start, fixing the root cause is better.

I'd like to see the entire pipeline operate better, even if they've managed to make the final output appear smoother.

 

[Tamlin_WSGF]

Again, nowhere there did I say it appears in the output, just that I still think single-card configurations need some fixes.


I see variance and still assume there's a problem, one AMD has masked in software by the time FCAT results are generated. Treating the symptoms is a good start, fixing the root cause is better.

I'd like to see the entire pipeline operate better, even if they've managed to make the final output appear smoother.

Fix what? As explained to you by me and also by the links I gave you from Guru3d, FRAPS captures very early in the pipeline. What happens there may or may not be relevant to the actual gameplay. On both Nvidia and AMD cards, you see variation and spikes from FRAPS that you might not find on FCAT.

On the other games in the same review you linked to, there is less varation also for AMD cards. What you need to look at, is FCAT. Fraps only confuses you as seen in this thread and makes you see problems that are not there (kinda like guru3d explained FRAPS captures vs. FCAT).

There is no masking going on. The driver handles this game very well and gives a great and smooth output.

Even between drivers of the same vendor, you see variations on how it handles different games. Don't worry about that, unless you are creating drivers yourself. Its the output that matters.

 

[Unknown-One]

Fix what? As explained to you by me and also by the links I gave you from Guru3d, FRAPS captures very early in the pipeline.

You answered your own question. I'd like to see them fix this early in the pipeline rather than late.

We're already dealing with some of the detriments to solving a problem like this late in the pipeline. AMD's implementation appears to be API-specific (only works on DX10 and DX11 games at the moment). Means they're going to have a more complicated solution overall.

Nvidia's frame metering seems to be low-level enough that it "just works" on everything. That's what I'd like to see from AMD as well. That's a totally valid concern, is it not?

What happens there may or may not be relevant to the actual gameplay.

Never said it was relevant to gameplay... where are you getting this stuff from? I'm pointing out a difference, I find the current solution to be sub-optimal, and I'm hoping AMD improves their implementation. The fact that it doesn't work in DX9 titles or in high resolutions / eyefinity is already a major downside that they have yet to address.

On the other games in the same review you linked to, there is less varation also for AMD cards. What you need to look at, is FCAT.

I'm looking at the results from both FRAPS and FCAT to point out a pretty obvious difference in implementation. I'm talking about lower-level issues than simple "gameplay experience"

Fraps only confuses you as seen in this thread and makes you see problems that are not there (kinda like guru3d explained FRAPS captures vs. FCAT).

FRAPS isn't confusing me, I'm using its results compared against FCAT to determine that AMD solves the metering problem much later in the pipeline than Nvidia. Simple as that.

Not sure why you're having such a hard time wrapping your head around the concept...

There is no masking going on. The driver handles this game very well and gives a great and smooth output.

Of course there's masking going on, you can see it in the results. There's high variance early in the pipeline and there's low variance late in the pipeline. The variance has been masked.

Nvidia might be masking it (earlier) or preventing it from the get-go (by actually delaying the rendering of a frame rather than delaying the display of a rendered frame). I'd be curious to know how they've handled it.

Even between drivers of the same vendor, you see variations on how it handles different games. Don't worry about that, unless you are creating drivers yourself. Its the output that matters.

Like I said at the top of this post, solving this late in the pipeline is asking for additional complexity. Additional complexity means the potential for more bugs and additional difficulty in supporting features (we're already seeing API limitations as is...)

 

[Tamlin_WSGF]

You answered your own question. I'd like to see them fix this early in the pipeline rather than late.

Lots of micro-quoting, so I'm cutting this short (and I'm a bit tired of this discussion too).

There is nothing to fix as long as FCAT shows a smooth output. That you OCD on some difference between a fraps capture of present and FCAT output, is getting tiresome. If they would be similar, the game would not be smoother. It doesn't make any difference other then it worries you. Try to forget you saw the FRAPS output.

AMD and Nvidia doesn't approch every single game the same way, not even between drivers. As long as the mojo works, it works and there is nothing to fix.

Move along.

 

[Unknown-One]

There is nothing to fix as long as FCAT shows a smooth output.

Sure there is, because AMD's solution seems so software-driven that it doesn't work AT ALL in most games... Nothing OpenGL or DX9 is helped as of yet.

They're working on it, but if they have to fix this per-API (or worse, per-game) then it's sub-optimal from an implementation standpoint. They've made a lot of additional work for themselves.

Like I said earlier, just because FCAT comes out clean (in the games that AMD's solution actually works on) doesn't mean their solution is optimal. I'm not complaining about the on-screen results (when it works), I'm complaining about the implementation.

That you OCD on some difference between a fraps capture of present and FCAT output, is getting tiresome. If they would be similar, the game would not be smoother. It doesn't make any difference other then it worries you. Try to forget you saw the FRAPS output.

I never said the game would be smoother if they were similar... again, not sure where you're getting that from.

I said I don't agree with AMD's implementation and pointed out several issues I take with it. Nothing more.

AMD and Nvidia doesn't approch every single game the same way, not even between drivers. As long as the mojo works, it works and there is nothing to fix.

Except it doesn't work in most games on the AMD side of things... at least not yet. That's highly troubling, and tells me AMD's results are never going to be as consistent with how they've implemented their fix.

 

[Tamlin_WSGF]

Sure there is, because AMD's solution seems so software-driven that it doesn't work AT ALL in most games... Nothing OpenGL or DX9 is helped as of yet.

They're working on it, but if they have to fix this per-API (or worse, per-game) then it's sub-optimal from an implementation standpoint.

Like I said earlier, just because FCAT comes out clean (in the games that AMD's solution actually works on) doesn't mean their solution is optimal. I'm not complaining about the on-screen results (when it works), I'm complaining about the implementation.


I never said the game would be smoother if they were similar... again, not sure where you're getting that from.

I said I don't agree with AMD's implementation and pointed out several issues I take with it. Nothing more.

Except it doesn't work in most games on the AMD side of things... at least not yet. That's highly troubling, and tells me AMD's results are never going to be as consistent with how they've implemented their fix.

The frame pacing driver doesn't have anything to do with single cards as we dicussed and not the Tomb raider results we were discussing either. Anyway, its part 1. Part 2 brings DX 9 frame pacing for crossfire.

Both AMD and Nvidia have game spesific tweaks and profiles. Also for single cards. Tomb raider as we discussed is an AMD title with TressFX hair from AMD. They didn't use it in the review from TR, but AMD probably have its own mojo for this game, since they took part in the developement.

Don't OCD on the Fraps difference vs. FCAT. Its pointless and as mentioned several times from me and guru3d, fraps can sometimes show things that are not relevant to the output.

This is not an issue and we have both waisted a lot of time discussing something you found irregular and didn't understand, making a big fuzz over nothing.

I'm done discussing this.

 

[Unknown-One]

Don't OCD on the Fraps difference vs. FCAT. Its pointless and as mentioned several times from me and guru3d, fraps can sometimes show things that are not relevant to the output.

...one more time. I never said it was relevant to the on-screen end-result. How many times do I have to say it?

I'm talking implementation, not end-result. Two very different things.

This is not an issue and we have both waisted a lot of time discussing something you found irregular and didn't understand, making a big fuzz over nothing.

It's obviously an issue, because we're still waiting for them to actually make it work, at all, on most games... and I'm fairly positive it's because of how high-level AMD's solution is. If the lack of a low-level fix is what's holding them up, that's hardly "nothing"...

I understand perfectly, you're missing the point I'm making completely. You keep arguing that all that matters is the end result, which it totally false. I'd much rather see AMD pursue an alternate means to their ends, one that isn't API-specific in any way, that works universally.

 

[Iching]

Is there an option to edit out all those whiny people? It really annoys when I have to sieve trough all that mumbo jumbo.

 

[The Mac]

yes, you can block posts being displayed from designated people...

i have a couple blocked from the old P67 thread....

 

[solidnickel]

@ the unknown one, what games are you having issues with micro stutter now outside of directx 9?

I think it should come down to what your brain processes when looking at the game on screen vs fixating on numbers not visible to the naked eye, get my point?