Do AMD / ATI Drivers really suck??

[Unknown-One]

I thought frame pacing was working on the R290/R290X even in CrossfireX.

A quick Google search returns no results suggesting that the R290 and R290X allow frame pacing in Eyefinity, 4k, DX9, and/or OGL...

Even so, if it was ONLY fixed on the latest cards... that sends a pretty crappy message. "there's a bug? Yeah, you'll have to buy new hardware to get the fix for that"

Unfortunately, I wouldn't be surprised if that really was their answer. It wouldn't be the first time:

1. HD 58XX series:
UVD clock bug (card refuses to enter 3D clocks if an application is engauging UVD mode)
Never fixed, had to upgrade to a 69XX series card.

2. HD 68XX and 69XX and 78XX and 79XX series:
Tearing persists on one monitor in an Eyefinity group (even with V-Sync enabled) when using DVI + DVI + DP outputs.
Never fixed, had to upgrade to the 290 or 290X


AMD eventually gifted me an HD6970 after troubleshooting HD58XX cards to hell and back. "Fixed" the UVD clock bug, but landed me with tearing in Eyefinity. That's honestly the biggest reason for dumping my HD6970 into a single-monitor system. After I heard reports that the 78XX and 79XX cards were still experiencing the SAME bug, I gave up on it. Nice that they finally fixed it on the R290 and R290x, but that doesn't help any previous customers (and doesn't help the resale value of our cards, either).

 

[Final8ty]

A quick Google search returns no results suggesting that the R290 and R290X allow frame pacing in Eyefinity, 4k, DX9, and/or OGL...

The results also dont suggest that they do not either.

And try to learn the distinction between bugs and improvements.

 

[Unknown-One]

The results also dont suggest that they do not either.

Most of them just list the same thing as usual, that 4k, eyefinity, DX, and OGL still aren't supported by frame pacing.

Even so, with a lack of proof in BOTH directions, I'd fall back to the principal of occam's razor. Namely that the explanation with the fewest assumptions tends to be the correct one.

If it's still a problem on the 290 and 290x, we're assuming:
1. That AMD has done nothing special for the 290 and 290x (as in, frame pacing works the same way as it does on all GCN-based cards).

If it has been fixed on the 290 and 290x, we're assuming:
1. AMD has fixed the issue.
2. The fix only works on the 290 and 290x even though older cards are also GCN-based.
3. This fix hasn't been shouted from the rooftops for unknown reasons.

And try to learn the distinction between bugs and improvements.

What, among the two things I listed, was not a bug?

The card getting stuck at 2d clocks when UVD mode is active is most certainly a bug. If you so much as leave a YouTube video paused in a minimized browser window, and you have an HD 58XX card in your machine, it will outright refuse to run 3D clocks no matter what application you open. It wont return to normal operation until the video window that's holding it up is closed completely. Anything that uses hardware accelerated video decoding triggers this behavior, and the bug persists to this day. Only way to fix it is to get a new card.

Getting tearing on one display in Eyefinity (even with V-Sync enabled) is also a pretty obvious bug... it also wasn't an "improvement" because the 58XX cards DID NOT suffer from the problem. It was a new issue introduced with the 68XX and 69XX cards.

 

[BababooeyHTJ]

The results also dont suggest that they do not either.

And try to learn the distinction between bugs and improvements.

They sort of do. Techreport and I think pcper have done some benchmarks that may indicate that it may still be an issue in those scenarios.

I only noticed microstutter in a couple of corner cases at 2560x1440. That includes quite a few DX9 titles that i tried.

 

[Final8ty]

Most of them just list the same thing as usual, that 4k, eyefinity, DX, and OGL still aren't supported by frame pacing.

Even so, with a lack of proof in BOTH directions, I'd fall back to the principal of occam's razor. namely that the explanation with the fewest assumptions tends to be the correct one.

If it's still a problem on the 290 and 290x, we're assuming:
1. That AMD has done nothing special for the 290 and 290x.

If it has been fixed on the 290 and 290x, we're assuming:
1. AMD has fixed the issue.
2. The fix only works on the 290 and 290x even though older cards are also GCN-based.
3. This fix hasn't been shouted from the rooftops for unknown reasons.

If you like to assume the worst is up to you but its another thing to perpetuate it to the point that its put across as fact when you have nothing to go on and when there is a known fact that changes have been made to CF on the new cards where it would be more understandable if changes haven't been made.

 

[HFBallad]

Supposedly, the fix for older cards (280x/7900s) is being worked on. AMD just prioritized the flagship cards for obvious PR reasons.

Kind of made me wonder just how different Hawaii are from other cards, since the fix for one can't be applied to another (even though they share a similar enough architecture for mantle).

 

[Unknown-One]

If you like to assume the worst is up to you

I'm not assuming the worst. I listed the assumptions that need to be made for each explanation, and concluded that the issue NOT being fixed involves the fewest assumptions.

but its another thing to perpetuate it to the point that its put across as fact

Please, read more carefully. I did not put it across as fact, I listed the ENTIRE thought-flow from end-to-end...

You yourself said there is a lack of evidence for both, and I agreed with you... I agreed with you to the point that Occam's razor actually entered the conversation legitimately. It's a tool for evaluating explanations when there is no evidence to support either one, nothing more.

when you have nothing to go on and when there is a known fact that changes have been made to CF on the new cards where it would be more understandable if changes haven't been made.

Ok, so add to the list of assumptions that the these changed made to Crossfire may have somehow inherently fixed the problem.

You're not adding any evidence, you're just adding more assumptions...

 

[Final8ty]

I'm not assuming the worst. I listed the assumptions that need to be made for each explanation, and concluded that the issue NOT being fixed involves the fewest assumptions.


Please, read more carefully. I did not put it across as fact, I listed the ENTIRE thought-flow from end-to-end...

You yourself said there is a lack of evidence for both, and I agreed with you... I agreed with you to the point that Occam's razor actually entered the conversation legitimately. It's a tool for evaluating explanations when there is no evidence to support either one, nothing more.


Ok, so add to the list of assumptions that the these changed made to Crossfire may have somehow inherently fixed the problem.

You're not adding any evidence, you're just adding more assumptions...

The lack of evidence means that i would of personally chosen to say nothing at all to whether the changes have fixed it or not unless a situation arose because there is nothing to go on, i was simply pointing out changes have been made, so the point is not to assume either way which i have not but you had.

And as you have a reputation to ramble on endlessly we are done.

 

[exlink]

Do they suck? Not quite.

But I think that they still have a ways to go until they get to Nvidia's level on driver quality (not saying there haven't been a few bad drivers with the green team either).

 

[maboblivion]

http://www.pcper.com/reviews/Graphi...-290X-CrossFire-and-4K-Preview-Testing/Skyrim on cat 13.8 4k runt frames in dx9.

 

[TFchris]

If AMD/ATI drivers really do suck... I'd hate to see how much they would crush the holy hell out of Nvidia if they had "good" drivers.

 

[Final8ty]

http://www.pcper.com/reviews/Graphi...-290X-CrossFire-and-4K-Preview-Testing/Skyrim on cat 13.8 4k runt frames in dx9.

Ouch, a return of the runt/dropped frames! Skyrim is our only DX9 title remaining in the test suite and it shows us that AMD has only improved frame pacing on the R9 290X for Eyefinity and 4K for DX11 and DX10 titles. Notice the difference in the orange line between the FRAPS FPS and Observed FPS images - that is the problem we saw across the board with CrossFire that is slowly being fixed, product by product.

So they tested one DX 9 game and one that is notorious for stuttering on multi AMD cards, its not like there are no other Dx9 games or that none of the other games that they tested didn't have a Dx9 mode.
Saying that it seems everything else has been fixed on the 290.

 

[exlink]

If AMD/ATI drivers really do suck... I'd hate to see how much they would crush the holy hell out of Nvidia if they had "good" drivers.

Performance wise, their drivers were always fine for me. It always just seemed that for every one thing they fixed in a new driver revision, they broke two other things: texture flickering, black screens, microstuttering, poor/negative crossfire scaling, game crashes, etc.

So no, I doubt they would crush Nvidia with "good" drivers, but rather they would just stop pissing off some of their consumers with problems.

 

[Final8ty]

Performance wise, their drivers were always fine for me. It always just seemed that for every one thing they fixed in a new driver revision, they broke two other things: texture flickering, black screens, microstuttering, poor/negative crossfire scaling, game crashes, etc.

So no, I doubt they would crush Nvidia with "good" drivers, but rather they would just stop pissing off some of their consumers with problems.

I have seen the exact same issues on the Official NV forum.

https://forums.geforce.com/default/board/33/geforce-drivers/

 

[Final8ty]

They sort of do. Techreport and I think pcper have done some benchmarks that may indicate that it may still be an issue in those scenarios.

I only noticed microstutter in a couple of corner cases at 2560x1440. That includes quite a few DX9 titles that i tried.

Is that with lashing of Supersample

 

[exlink]

I have seen the exact same issues on the Official NV forum.

https://forums.geforce.com/default/board/33/geforce-drivers/

Yes, but I'm speaking from personal experience as I've mentioned in both posts. I've also mentioned in my first post that Nvidia also suffers from driver issues, never said they didn't. The thing that separates the two is how frequent these problems are, and from my experience, AMD drivers tend to suffer from these problems quite a bit more.

 

[Final8ty]

Yes, but I'm speaking from personal experience as I've mentioned in both posts. I've also mentioned in my first post that Nvidia also suffers from driver issues, never said they didn't. The thing that separates the two is how frequent these problems are, and from my experience, AMD drivers tend to suffer from these problems quite a bit more.

That's all a matter of opinion because personal experience differs and no one can claim theirs is the one that all others should be judged from.

 

[maboblivion]

So they tested one DX 9 game and one that is notorious for stuttering on multi AMD cards, its not like there are no other Dx9 games or that none of the other games that they tested didn't have a Dx9 mode.
Saying that it seems everything else has been fixed on the 290.

ppl were saying frame pacing doesn t work in dx9 above 1440p?
Link shows crossfire works on dx9 but not above 1440p.

if the skyrim problem was just stuttering, the frame times graph would be different.
it clearly shows a frame pacing issue at 4k.

next month amd will release the dx9, eyefinity fix and hopefuly we can put this debate behind us.

 

[exlink]

That's all a matter of opinion because personal experience differs and no one can claim theirs is the one that all others should be judged from.

Did I say that my personal experience or opinion is the final say in the matter? No.

I clearly stated in all my posts in this thread that they are based on my personal experiences and never said that it was the be all and end all in the subject matter.

 

[samuelmorris]

Methinks I'll wait and see what actually happens before upgrading
If there's no frame pacing fix for MST 4K by february-march time a pair of Geforces really does seem like the most logical choice. So expensive though :/

 

[BababooeyHTJ]

I have seen the exact same issues on the Official NV forum.

https://forums.geforce.com/default/board/33/geforce-drivers/

Not so much the texture flickering and artifacts. That for me has always been a far larger issue with AMD cards. Especially with non-AAA games.

You might see more issues reported on AMD's forums if they were actually useful or if AMD bothered to respond on their own forums. Thats one other huge issue that I had with AMD.

Why in the hell should they need to respond to the black screen issue on OCN instead of on their own forum? Seriously, Nvidia doesn't have to do that for a reason. They built up a support community on their forums over time. AMD could and should do the same. Not being able to get any status updates on open issues is annoying to say the least.

I don't see why you would point to users communicating with Nvidia and getting a response as some sort of bad thing. AMD needs to do the same. That right there is my #1 issue with AMD and it has been for years. I am sure that you've seen me blast them for just that on the forums.

 

[Final8ty]

Not so much the texture flickering and artifacts. That for me has always been a far larger issue with AMD cards. Especially with non-AAA games.

You might see more issues reported on AMD's forums if they were actually useful or if AMD bothered to respond on their own forums. Thats one other huge issue that I had with AMD.

Why in the hell should they need to respond to the black screen issue on OCN instead of on their own forum? Seriously, Nvidia doesn't have to do that for a reason. They built up a support community on their forums over time. AMD could and should do the same. Not being able to get any status updates on open issues is annoying to say the least.

I don't see why you would point to users communicating with Nvidia and getting a response as some sort of bad thing. AMD needs to do the same. That right there is my #1 issue with AMD and it has been for years. I am sure that you've seen me blast them for just that on the forums.

Indeed.

 

[samuelmorris]

The only support I ever got out of AMD was by phone, which is outsourced to a tech support firm in the country you live in. Unfortunately at the time they refused to give much help as I wasn't running a PSU big enough for their 'recommended PSU requirements' table, even though it was ample for the job. Fortunately in that instance I could fix the issue for myself. If you want customer support from the manufacturer, definitely go with nvidia. With AMD it's down to the partner brands (i.e. no support at all), the phone line, and the retailer.

 

[Final8ty]

Did I say that my personal experience or opinion is the final say in the matter? No.

I clearly stated in all my posts in this thread that they are based on my personal experiences and never said that it was the be all and end all in the subject matter.

Indeed but it is a rarity for people to keep it personally so yes i did overlook that fact in error, we are the minority.

 

[Unknown-One]

The lack of evidence means that i would of personally chosen to say nothing at all to whether the changes have fixed it or not unless a situation arose because there is nothing to go on, i was simply pointing out changes have been made, so the point is not to assume either way which i have not but you had.

One more time, read carefully. I did not assume one way or the other...

All I said is that (in the absence of evidence) the conclusion that requires the smallest number of assumptions tends to be the correct one. I then proceeded to list the assumptions required for BOTH conclusion currently in-play. Nothing more.

"The issue being resolved" requires more assumptions than "the issue remaining unresolved," ergo "the issue being unresolved" will TEND to be the correct conclusion.
Now, this is where you need to read carefully; the above does NOT assume that it IS the correct conclusion, only that the odds are in favor of it begin the correct one (as that is the best you can do in the absence of evidence).

And as you have a reputation to ramble on endlessly we are done.

A reputation? Really?

I'm happy to re-explain my point until what is being said is properly understood, and that is all. Attempting to shut people down with phrases like "we are done" doesn't help anyone.

Saying that it seems everything else has been fixed on the 290.

Still doesn't look 100% fixed, and still doesn't help the majority of users with AMD cards. Still doesn't look good :-/

I'm still not happy about it being a driver-based software solution, either. I would have much preferred a hardware frame-rate metering setup like Nvidia has been using for the last few generations (which doesn't have any rendering engine or resolution dependencies as far as I'm aware). I have doubts that AMD's frame pacing implementation will ever support 100% of titles (again, not saying it wont happen, but their driver team has a long road ahead).

next month amd will release the dx9, eyefinity fix and hopefuly we can put this debate behind us.

That'd be nice... hopefully said drivers turn up soon. I'd love to put my HD 6970 through its paces with them.

 

[Final8ty]

Still doesn't look 100% fixed, and still doesn't help the majority of users with AMD cards. Still doesn't look good :-/

Originally Posted by Final8ty View Post
So they tested one DX 9 game and one that is notorious for stuttering on multi AMD cards, its not like there are no other Dx9 games or that none of the other games that they tested didn't have a Dx9 mode.
Saying that it seems everything else has been fixed on the 290.

Which means not 100% fixed already and NVs solution 100% in all titles either even though it more compatible at this stage,

Whats good enough or not is for the individual user to decide.

 

[BababooeyHTJ]

Last few generations? Nvidia has only been using a hardware solution since Kepler.

They've been working on it for a few generations and it shows with Fermi but I don't think that had any hardware solution.

Have you used 290x crossfire? Read [H] comments? They're well on their way with that issue. I'm not so sure at this point that microstutter would be a determining factor for me in my purchasing decision.

 

[Final8ty]

Last few generations? Nvidia has only been using a hardware solution since Kepler.

They've been working on it for a few generations and it shows with Fermi but I don't think that had any hardware solution.

Have you used 290x crossfire? Read [H] comments? They're well on their way with that issue. I'm not so sure at this point that microstutter would be a determining factor for me in my purchasing decision.

Indeed as they are just too close and neither solution is 100% and they never will be.

 

[irishtea]

Hello

Would like to know if AMD / ATI Drivers really suck, I have seen people say they suck, others say they are fine. I have not used a ATI card in many years so do not know. But might buy one of there new cards soon or just grab a gtx 770 not sure yet.

One game I am worried about is Skyrim. I have read of many people having problems with ATI cards and Skyrim. Want to play a heavy modded game. Right now I am just runing out of vram so yea need a new card lol. I posted here because I would think this forum would be a little more neutral than the AMD or Nvidia forums

Personally, running 6970's in xfire with newest catalyst beta drivers and I still haven't had any major issues. The most problematic thing I've ever encountered is that ATI/AMD drivers do not like to work with a lot of forced hardware acceleration runtimes in mods for games like Skyrim and Final Fantasy. However, I've also noticed that depth of field modding in any game also instantly destroys any performance or visibility that would normally be present. But, outside of that, I've been getting my moneys' worth playing games at their highest quality. Can run Witcher 2 Enhanced Edition with Ubersampling and max settings, so I'm not upset especially considering I spent 1/2 as much on my ATI setup as I would have on an intel rig.

food for thought.

 

[koolaidkitten]

As an owner of a 7950 7770 5950 4870 in various boxes. I dont see a problem. All my cards play games,movies,webpages just fine.

 

[Unknown-One]

Btw, can anyone get the latest version of Plex Home Theater to work correctly on an AMD setup?

Trying to run it on my HD 6970 results in horrible pink and green corruption within the Plex app that gets worse the more I navigate around. There are also issues with Z-order (objects layered in the wrong order). When full-screened, the UI scrunches up against the top of the screen leaving a huge black bar along the bottom of the UI and causing controls to mis-align badly.

Tried 3 different drivers (including the latest betas) and it happens with all of them. The Nvidia system sitting 4 feet away doesn't experience any of these issues when Plex is loaded, so I'm fairly sure this is AMD-specific.

Edit: And just so all bases are covered, the HD6970 checks out as stable. Handles looping Heaven / Valley without a problem.

 

[wonderfield]

AMD drivers are okay, in my experience, though I frequently encountered issues trying to install them. The control panel, however, is an entirely different story. AMD cards practically require third-party applications for configuration.

No one called them out on the microstutter until recently. Nvidia only fixed it this gen as well.

No, NVIDIA's been doing frame pacing since the 5-series, if memory serves. They only recently moved to a hardware solution.

 

[Unknown-One]

No, NVIDIA's been doing frame pacing since the 5-series, if memory serves. They only recently moved to a hardware solution.

GeForce 5 (aka, the GeForce FX series) doesn't even support SLI... First cards to support SLI were all GeForce 6 series.

 

[Jeffredo]

I've been using Nvidia exclusively since 2007. During Black Friday I caught a very good deal on an MSI HD 7850 2GB Twin Frozr and installed it on a secondary PC I have. The Catalyst drivers installed without a hitch and have been performing perfectly in all my games. I haven't noticed any difference between it and my GTX 670 main PC.

 

[wonderfield]

GeForce 5 (aka, the GeForce FX series) doesn't even support SLI... First cards to support SLI were all GeForce 6 series.

5xx-series.

 

[ebduncan]

I haven't really experienced a driver related program with AMD graphics cards as of late.

Crossfire is hit or miss, some times a game will support it, sometimes it will not. This doesn't apply to the AAA titles, more like green light games and small studio games. I cannot blame AMD on this because usually the game developers didn't bother to support multiple gpu setups.

I have had the drivers crash on me, but this is my fault for overclocking the cards, this only happens when pushed to far.

So no the drivers don't suck.

 

[Dan_D]

If AMD/ATI drivers really do suck... I'd hate to see how much they would crush the holy hell out of Nvidia if they had "good" drivers.

Raw frame rates weren't one of the problems associated with AMD drivers. They wouldn't crush the holy hell out of anything. They would have stronger sales as people like me would be more likely to buy their cards as we wouldn't have been burnt by their drivers so many times. Overall my Crossfire experiences have been mixed:

• X850XTX Crossfire - Didn't work at some resolutions, loud, unavailable cards, hack job feature with loop back monitor cable. Didn't buy.
• X1800XT Crossfire - Used some of these on the test bench. These were loud and slower than their equivalent NVIDIA cards. You only think the R290X's are loud until you've heard these.
• X1950XTX - Absolutely fantastic. This led to further ATI/AMD purchases.
• X1950Pro - Thrown in another machine for the girlfriend. Worked FLAWLESSLY.
• Radeon HD 4870 X2 CrossfireX - Couldn't get it to work properly. Lack of frame pacing, worse performance than a single card in most games, horrendous tearing in CrossfireX.
• Radeon HD 5970 - Great card. I'd have bought another but AMD never did get CrossfireX sorted out so that it would run worth a shit.
• Radeon HD 69xx series - Avoided for fear CrossfireX would suck. They actually had good scaling that generation but I had felt too burned by ATI/AMD to chance that generation.
• Radeon HD 7970 Fantastic card on it's own, but CrossfireX was a complete failure for me. Horrible performance much of the time. Hard locks only in multimonitor mode, second card wouldn't always kick in for Crossfire, compatibility problems with X58 and X79, so I said screw that and my 7970 is now sitting on a shelf. I returned the second one I had for a full refund which I used to buy a GTX 680 4GB card at the time.

But frame rates (aside from Crossfire configurations where you would sometimes see worse performance with multiple cards) were never one of their problems. NVIDIA hardware always felt smoother to me and it is because frame pacing has been in their drivers for some time now.

 

[schizrade]

My first dealing with a Discreet ATi card was the Nexus 32. It had 32MB or ram onboard... of which only 16MB were enabled by the drivers... FUCK THAT SHIT. lol.

Went out and got a 3DFx card the next day. Problem solved.

Since then, I have not had that kind of piss poor ATi/AMD experience, but it certainly taints it for me.

 

[xoleras]

I can state one thing for certain. Having been nvidia for the past year, their drivers and software are way more polished and functional than AMD's. They also have far more features.

With 7970CF, the performance was always great but i'd get a strange stupid issue once per 1-2 months. Including but not limited to forced vsync not working, CF causing crashes in Witcher 2, CF being broken in assassin's creed brotherhood, beta AMD drivers destroying windows installations and forcing me to reinstall or use a system restore, among other things. And then there's the microstutter which I had to workaround using a framerate limiter and vsync, which I should not be required to do. You don't have to do workarounds like this on the green side. Hell, AMD still hasn't fixed CF/eyefinity stutter for 79xx cards more than two years later...

I liked those cards a lot and thought weird issues were the norm for multi GPU. Enter nvidia SLI. Have not had stupid shit like that. Ever. Now some may find AMD to be useful in terms of value, but it would take a hell of a lot for me to take AMD over nvidia now that they're comparably priced. NVidia just has way more polished software, and if something breaks they're QUICK to fix it. The 7970CF crashing in Witcher 2 took 5 months to resolve. Talk about piss poor software fixes. In retrospect, you deal with less bullshit on the green side. That's what i've found, having mulled the situation from using 7970CF extensively and then using SLI extensively. I thought stupid bullshit issues now and then were status quo for multi GPU. Well, it isn't. Nvidia generally gets you a good experience from day one and if something is broken, they fix it fast. AMD on the other hand, is super slow to fix anything IF they fix it at all.

In fact, it usually takes websites pestering them non stop to fix issues at times. For example, CF microstutter. 290/X press vs retail variance. Would AMD have bothered if it weren't for the press breathing down their neck? I highly suspect they would not bother. It takes a miracle for their software team to do their job sometimes, it seems like. Or perhaps they don't have the manpower or money that nvidia does to be on their level. Whatever the case, I only care about the end result - I find nvidia far better in terms of software in general. Maybe something good can come from this mining craze - maybe AMD will have enough money now to actually hire some additional software engineers. I certainly hope they do, because the situation now is that nvidia is just better in every way in terms of polishing their software and ensuring issues don't happen. Now nvidia isn't perfect. But they're a hell of a lot better than AMD, particularly with multi GPU configurations.

 

[Final8ty]

[*]Radeon HD 5970 - Great card. I'd have bought another but AMD never did get CrossfireX sorted out so that it would run worth a shit.

Maybe for you but QuadFire 5970 was fantastic for me and my favorite so far.

2560x1600 1x5970 Ares CF 2x5970 Ares quadfire Ares 5970+5870 TriFire

The review is old and so are the drivers, much has changed since then, but that's basically my Quadfire setup.
http://www.techpowerup.com/reviews/ASUS/ARES_CrossFire/1.html


19 games tested some of them with no AA which does not help eliminate CPU bottlenecking which is running at 3.8Ghz., the more GPUs you have the greater the potential driver overhead.

Out of 19 games tested Quadfire is ahead of TriFire in 14 of them and about equal in the rest.

Out of 19 games tested Quadfire negative scales to worse than CF in 2 by a tiny margin.

Quadfire negative scales to 3 frames worse 138.6 fps than CF 141.9 fps in CODMW.

Quadfire negative scales to 2 frames worse 224.3 fps than CF 226.3 fps in Quake4 No AA