Slashdot Mirror
Multi-Display Gaming Artifacts Shown With AMD, 4K Affected Too
Vigile writes "Multi-display gaming has really found a niche in the world of high-end PC gaming, starting when AMD released Eyefinity in 2009 in three-panel configurations. AMD expanded out to six-screen options in 2010 and NVIDIA followed shortly thereafter with a similar multi-screen solution called Surround. Over the last 12 months or so, GPU performance testing has gone through a sort of revolution as the move from software measurement to hardware capture measurement has taken hold. PC Perspective has done testing with this new technology on AMD Eyefinity and NVIDIA Surround configurations at 5760x1080 resolution and found there were some substantial anomalies in the AMD captures. The AMD cards exhibited dropped frames, interleaved frames (jumping back and forth between buffers) and even stepped, non-horizontal vertical sync tearing. The result is a much lower observed frame rate than software like FRAPS would indicate and these problems will also be found when using the current top-end, dual-head 4K PC displays since they emulate Eyefinity and Surround for setup."
AMD also seem to have some serious problems, which seem to be worsening with each new driver, on their premium workstation cards when driving multiple displays. We've seen numerous video playback issues, including glitches away from the video area itself, on multi-display configurations. The most likely culprit at the moment seems to be changes in the GPU memory timing. I really hope they fix this soon, because our "professional" workstations are giving our professionals headaches right now.
If you disagree, post your argument. (-1, Overrated) isn't your personal censorship tool for views you don't like.
I started off with AMD cards maybe 6 years ago before I tried an NVIDIA card. It really is just a smoother experience overall. I don't know what it is, but I've been shying away from building any new systems with an AMD lately.
first world problem. I mean, really - we should care because a company's tech won't perfectly display game images on multiple monitors?
I was playing flight sims on my Quadra 900 in the late 80s/early 90s with 4 displays. The resolutions and detail may be higher today, but I never had any issues or failures of the system. FA/18 Hornet was my favorite.
Article X: The powers not delegated... by the Constitution...are reserved...to the people
That pretty much hints at a driver issue, or bad GPU sync.
Still waiting on Serviscope_minor to wake up to fucking reality and realize that Jessica Price isn't going to fuck him.
I remember when we had those Matrox cards to go with our video editing workstations. Those things were stable as hell
Back then there were more vendors competing fiercely in the market, and all of them were on their toes as they knew even one slip could turn out to be totally fatal.
Nowadays, other than AMD and Nvidia, what other serious players do we have ?
None.
With the market turns into duopoly both the players no longer have the urge to bring new and innovative features into their new products.
How many times we have heard of the horror stories brought on by their crappy drivers ?
Other than lamenting online, the users (no matter if they are casual gamers or professional users) have no other option but to wait for a newer version of the drivers, or roll back the drivers to one that worked.
ps. I still have several of those Matrox cards with dual video outputs.
Muchas Gracias, Señor Edward Snowden !
I might want multiple monitors, but why does it cost 250 watts of extra energy? I'm not rendering 3d models or getting to play BF3.
Huh?
The reported anomalies at first glance look to be it is hard to shove lots of data around (dropped frames), microcode issue ie binary blobs that they load into the card... a level below the "driver" (interleaved frames) and sync or timing issues (even stepped, non-horizontal vertical sync tearing). But that's just a first impression. Have I done any testing? Uh no. The fact of the matter is it is impressive that they move 4k of data around as fast as they do. But it could be an architecture issue, it could be a hardware issue, it could be software in the card communicating to the card or sending data to the OS to be relayed to the card. Heck it could be an architecture issue of the mainboard, the task scheduler or the age old bus just ain't fast enough to feed the card.
So we have a problem. Now the hard work of narrowing the problem down can begin. My money is on all of the above. Subtle errors all over the place that nobody could test for and thus couldn't know they needed finding and fixing.
fristwoldporblem
Hmm...AMD is about to drop Hawaii and retake the single GPU crown and GUESS WHAT! Another round of FUD from the company that cancelled their next premium GPU in the name of funding also-ran ARM devices and niche handheld gaming devices. Fuck off, Jens. You played your hand now reap what you sow.
Note that the words "driver" and "version" don't occur on the page. There is a know issue that AMDs been working that sounds a lot like this issue. It's been known for months, they've got a "two phase" plan to attack it, the first of which is implemented in the current beta driver-set.
The timing of this article is very suspect. They're either reporting on a new problem (and totally failing at providing any relevant data on their configuration), or they're simpy regurgitating an already know issue, like doing a big splashy article about a bug report.
But the old PA-RISC, Sun Microsystems, and SGI boxes all had high end video options that beat out personal computers in resolution, bit depth, memory speed, and reliability as often as not.
Now mind you since PCs began dominating the market due to much more rapid turnover, R&D, and generational increases in performance, yeah I would agree the driving force has become games. However somewhere between 96 and '00 was when that really started to happen, rather than 'in the monochrome/cga/ega' days.
So we have a problem. Now the hard work of narrowing the problem down can begin. My money is on all of the above. Subtle errors all over the place that nobody could test for and thus couldn't know they needed finding and fixing.
You mean narrowing down the problem that is already known and already being worked on?
Perhaps the problem is rather, why does this article, which pretends nothing of this is already known, exist? If this is a new issue, they totally failed to show it.
Yes AC a lot of good people will be testing. Lets hope its not the CPU, motherboards, OS or games. If this can be fix by one company- great. :)
Subtle errors all over the place would have hopefully been picked up by the OS people, Intel, Nvidia, game testers, hardware makers over time?
Guess we might need a next gen card buy up for 4k
Domestic spying is now "Benign Information Gathering"
http://eliteownage.com/desk/cc082013preview.jpg
The Official Site of 1337 Pwnage
The biggest problem with multi-monitor gaming is that it's just plain garbage in any kind of "surround" configuration. Apart from Fisheye-Quake and some fancy pants flight sims and racing games, arcing three or more monitors does nothing but waste power and processing capability to render a smeared-out mess on every monitor but the one in the center. Most games aren't even mathematically capable of producing a 180-degree FOV. I've never been quite sure who should get the ball rolling in that department, but I've just decided it should be Valve. I don't have a good reason. Get on it, guys! Ubiquitous support for rendering games to multiple-viewports.
These days, you can't get more features out of your gamer card by tricking the card or driver software into believing it is a "Pro" card any more. If you buy a Pro card now, it's usually based on the previous generation of chipset, with a well stabilized and thoroughly tested driver, compared to the very short time to market that top range gamer cards get. The big problem is that newer chipsets often are run on the same driver and iterations between the chipsets are often nothing more than a die shrink size and maybe some optimization in memory path, controllers and such. This means that drivers are essentially the same for all chipsets and the single code base requires both stability for Pro cards and bleeding edge features for the latest and greatest gamer stuff. Essentially, the Pro users of GFX cards like the CUDA and engineering people, suffer from the big pull of the gamer market demanding ever increased high resolution and frame rates because the manufacturers work with a single code base for both lines of product.
I was promised a flying car. Where is my flying car?
Ran U2711 + the better Samsung led screen @ 2560x1440 and tearing was horrible on a 1.3GHz 7970 with almost any moving task about 2 years ago.
This was early days when drivers are often 'flakey' to say the least though.
See it rarely with 2560x1440 + 1080p tv today, running under clocked. At standard clocks don't notice it usually.
Since early days of crossfire and sli I've always been a fan of one big card clocked hard on a good screen with good colour.
Multi screen needs a while to catch up. Used both camps for a dizzying array of different projector/screen setups and AMD has better drivers for irregular screen size/setups. Nvidia often threw a shitty when it wasn't a certain resolution, or had issues detecting longer cable runs with powered splitter boxes etc over multiple cards. ATi/AMD flew through it majority of the time.
AMD has long had driver performance issues, compared to nVidia. Their hardware started really kicking some ass with the 4000 series and was just dominant with the 5000 series, but the software side has had some issues. I'm not sure what the issue is, maybe they need more people, maybe they need better people, maybe they need a better process. Whatever the case they end up having more issues. Stuttering and rendering partial frames has been one (that they have largely cleared up with single display setups), issues with Endur would be another (a year after my laptop release, it still has big issues).
So it just is what it is. If you get an AMD card you know you may have some issues like this, but you also know they will work on it, though any resolution may not be fast.
However this is a big issue for AMD fanboys. They, like all fanboys, put their identity, their ego, in AMD being the One True Way(tm) and being better than those other guys. So when something makes their chosen toy look bad, they have to steer the discussion away. They'll find a bug in nVidia drivers and say "BUT LOOK NVIDIA HAS T3H PROBLEMS!!!" to try and deflect the argument.
Rather than simply buying the card they want and enjoying the purchase, they have to make it an "us vs them" kind of thing.
An attitude which I never understood. Games designed to enforce a 90 degree FOV fail to take into account that on average, our peripheral vision encompasses about 150-160 degrees for most people.
Well, that's sort of the point of peripheral vision, isn't it? There's an easy test that I was taught in junior high that quickly demonstrates this. Hold your arms out in front of you, thumbs up. Move them to the edges of your vision on both sides until you can just see them. Stop, and take a quick look left and right. If you're like most people, you'll find that you're arms are now almost straight out from your sides.
Games which take into account this awareness tend to to do one or both of two things. The first is to allow an FOV up to some arbitrary limit somewhat greater than 90 degrees, say 110 or 120 degrees. Anything after that tends to get so distorted as to be useless on a single monitor anyhow.
The second option is to show some sort of indicators on the side of your monitor and/or allow a quick free look around of just your head. The best implementation of this model belongs to an FPS series that emphasizes realism in its player model to an extent that I've seen nowhere else. I'm speaking of course of the Operation Flashpoint/ArmA I-III series. This game series has been working on this basic model since, what? 1998? The ArmA branch of that series has also provided native support for multiple monitors and TrackIR since the first iteration.
If a FPS this fanatically dedicated to realism (OK, as long as you forget the brain dead AI and concentrate on everything else!) thinks this is OK, then why can't other games at least acknowledge the issue?
I'm running what's called a "5x1P" (that's 5 portrait monitors arrange horizontally) array and have been using Eyefinity since it was released with the Radeon 5870 HD. Multiscreen is kind of the thing over at the Widescreen Gaming Forum. I gotta say, I've been experiencing little issues here and there for a while. Problems with tearing seem to originate more from using alternating port types. When grouped in, for instance, all DisplayPort, there is no tearing. I imagine a lot of the issue arises from having a buffer that's too small to handle the higher screen resolutions. Personally, I will drop multiscreen gaming as soon as 21:9 4K TVs come out. High resolutions coupled with an accessory Oculus Rift are going to kill multiscreen tech in the next few years. For the time being, though, it's worth the glitches and the bezels.
Also, having an Oculus Rift in my possession (borrowed from the wonderful WSGF), I can NOT, for whatever reason, get it to work from my desktop. All I get is a black screen on the Rift. I'm not at all sure if it relates to using Eyefinity or from having 5 other devices plugged into the card, but I don't care to tinker with more than the software as it was annoying enough to get it set up. I'm just running it from my laptop and putting up with the performance limits. The screen door effect is way too obvious to be fixed by upping to 1080p screens, but in the next few years it may be feasible to do away with external monitors altogether.
What day is it? Could you please tell me?
Were you running two different port types? Switching between the ports seems to be the cause of tearing. I was having MAJOR issues running across 5 screens and finally tracked it down to that. Which I had known so I could have looked for a card with all the same port type.
What day is it? Could you please tell me?
this is exactly why I continue to read /. and I hope to see more of these types of articles instead of the crappled slashvertisements
Mod me up/Mod me down: I wont frown as I've no crown
Other than lamenting online, the users (no matter if they are casual gamers or professional users) have no other option but to wait for a newer version of the drivers, or roll back the drivers to one that worked.
No, I think we have at least one other option: next time we're specifying new workstations, we can just use (relatively) cheap gaming cards, instead of paying a factor-of-several premium for workstation cards. The latter are often the same basic hardware, but cost more because their "certified" drivers supposedly have better performance and guaranteed compatibility with major content creation applications. Why pay the premium if the reality is that the premium drivers are no better (or, in this case, much worse) than what you could get a few years ago with a basic gaming card?
If you disagree, post your argument. (-1, Overrated) isn't your personal censorship tool for views you don't like.
As far as I know no other article anywhere has published what problems actually exist with Eyefinity, as they are very different than the problems that exist with CrossFire on single display configurations.
See, people say AMD has the driver issues, but the only issues I've ever had with drivers in the past 7 years or so was with an Nvidia card (and, actually, that was more a problem with the game). OTOH, the only video card hardware failure I've had was also Nvidia. Really, I think it's a case of YMMV. Some people have no problems with AMD, some have tons. Some have no problems with Nvidia, some have lots. I personally buy AMD stuff in part to help keep competition alive (and because their stuff is usually pretty inexpensive). I've never had a problem with any of their stuff (except for a really strange issue with playback of WMA files which seemed to be tied to my Athlon X2 CPU... but I've always avoided WMA anyways).
"None can love freedom heartily, but good men; the rest love not freedom, but license." --John Milton
I have tried a small handful of ATI/AMD cards over the years and have never had good results. Performance in Windows was always mediocre and I had massive stability or features not working on Linux. I will admit my last attempt was several years ago (less than 4 though). I'm not talking about embedded or mobile chipsets either, but dedicated AGP or PCIe.
(before you go all AMD fanboy on me (I'm hoping you wouldn't but just in case), I should mention I've been using AMD processors since I abandoned my old Pentium 3 back in the day, and my first few processors were cyrix)
For large sets, this will be our guide even unto death, for the LORD will work for each type of data it is applied to...
I played Quake Team Fortress (the original, for Quake 1) quite competitively. So there was no zoom key for sniping and the like, you just had to play with FOV. You made some binds to toggle FOV leves as you saw fit. This lead me to try bigger FOV numbers, and that worked too. So I had 4 FOV buttons, 10, 30, 90, and 160. 90 was where I played most of the time, 30 and 10 were for sniping, which I did rarely. 160 was for flag defense, which is often what I was assigned to. I could watch an entire flag room from one corner. Then, when I saw someone, I'd drop the FOV to 90 and go after them.
"AMD has long had driver performance issues, compared to nVidia."
Nvidia has had serious, serious hardware issues and also some driver issues like this minor multi-card glitch that will be resolved, idiot.
OTOH, the only video card hardware failure I've had was also Nvidia.
With video card failures, be they AMD/ATi or nVidia, you have to realize that the only part those companies make is the actual GPU, they don't make the actual card so the cooling system, caps, RAM and all the other components are made by others and assembled by an OEM so there are many points of failure that are completely unrelated to AMD/ATi and nVidia.
Hello, everybody, the good shoping place, the new season approaching, click in. ( http://www.sheptrade.com/ ) (Discount Air jordan shoes) $36, (Air Max shoes) $35, (Nike shox shoes) $36, (Handbags) $39, (Sunglasses) $16, (wallet) $18, (Belt) $17, (T-shirts) $20, (Jeans) $37, (NFL/MLB/NBA)Jerseys $25, ( http://www.sheptrade.com/ )