AMD Catalyst 11.2 Now Available

[Blacklash]

These are working great on my 5870s in my AMD rig. I've got back MLAA and the ability to cap tess if needed.

As far as crashing goes I get occasional crashing to desktop with Mass Effect 2 and my GTX 480 with the 266.58 driver. My AMD cards don't do that for whatever the reason in that title.

 

[Unknown-One]

Oh yea.. Not the driver, must be the configuration that caused it... Yea right..

Yup, that's one possible scenario.

So nVidia driver is sensitive enough to make shit happens, then it still driver issue..

I am done with arguing with you, pointless and you continuing pull the same thing over and over again without even acknowledge it is still the driver issue that caused it..

Nope, it could be another issue on the system causing the driver to crash through no fault of the driver itself. Just because Nvidia finds a way to work around the issue from their end does not mean the problem itself exists on their end.

I did acknowledge that it could be the driver itself (obviously, that's always a possibility). The thing is, the video driver can turn into a catch-all, crashing when something else goes wrong. Assuming that the problem is the video driver itself just because you see the video driver crash is very short sighted.

Nope it's not difficult at all, but you can't seem to grasp that for a lot of people the problem was/is nothing to do with hardware.

I've mentioned that the misconfiguration or conflict could be hardware or software related.

As for the flickering issue with dual monitors, doesn't nvidia have the same issue? And don't they get around this issue by running the card at full clocks when using two monitors?

Nvidia's issue is kinda similar, yeah. They use a higher idle clock with multiple monitors attached.

ATi uses a clock well below full 3D clocks for idle with multiple monitors (400 / 1000 is the idle speed on the HD5850), which work just fine.
The problem is, when you enable ATi Overdrive, the drivers glitch and cause the card to start idleing at 157 / 300. Those clocks are way too low, and cause secondary displays to flicker constantly.

 

[MRAB54]

I went from 10.0c? to this and I've had a weird problem thrice now with L4D2. FPS get really choppy and screen keeps wanting to go back to the desktop but doesn't quite get there, accompanied by long pauses. I alt-tab to see what the hell is interrupting things and there is nothing there. A reboot fixes the problem for the time being.

 

[MacLeod]

Upgrades from 10.10e only because this finally had upgrades that affected the 5000 series. Dont really notice any improvements anywhere but Ive not had any crashes or issues so it seems to be working so far. 10.10e was rock solid so I was leary of going with 11.2 but so far so good.

 

[LEVESQUE]

We're up to 1 Year, 5 months, and 20 days since the launch of the 5800 series...and we're still waiting for a non-bugged driver.

I'm using a Tri-Fire set-up: 5970+5870 since over 1 year already, and can't remember the last time I had a BSOD/crash/bug/freeze in the around 50 games I've played fully to the end since then, and also with every software I've used in the same period.

Rock-stable, rock-solid, great scaling. ... 10.11, 10.12, 11.1, 11.2. Not a single problems.

PEBKAC.

 

[MrLonghair]

The secret behind my system which is so stable I need to break something on purpose in order to mess it up is that instead of on-board Realtek sound I use a proper sound-card, and instead of a Realtek on-board NIC, I use a PCIe Intel Gigabit one.

Last year it was discovered that Realtek and Nvidia drivers clash, no tests were made with (then) ATI drivers, but chances are that's what going on with so many people in the world.

Asked a guy with a game that performs completely wrong to disable sound and whaddya know, it now runs at a steady 60+ framerate rather than 10-20 and crashing every other minute. Get proper soundcards and NICs, solve your problems and fix your unstable rigs.

(99% sure a real soundcard will help more than a NIC, but it feels good to dump everything Realtek in your system)

 

[samuelmorris]

No real issues with Realtek audio that I can see here, have been using it for a couple of years in my Radeon systems without incident. I do agree that Realtek stuff does leave a fair bit to be desired though. I'll be grabbing a proper PCI sound card again before long, just not dealing with all the driver BS from Creative again.

 

[Unknown-One]

I'm using a Tri-Fire set-up: 5970+5870 since over 1 year already, and can't remember the last time I had a BSOD/crash/bug/freeze in the around 50 games I've played fully to the end since then, and also with every software I've used in the same period.

Rock-stable, rock-solid, great scaling. ... 10.11, 10.12, 11.1, 11.2. Not a single problems.

PEBKAC.

It's not PEBKAC, the drivers have well documented bugs, and those bugs exist for everyone with a 5830 and up (at the very least).

Start a video that uses a hardware accelerated codec, your video card will drop to 400 / 900, secondary displays will flicker momentarily because the RAM clock changed, and the card will not clock up to full 3D clocks when you start a 3D application with that video open.

I'm not sure if running crossfire will effect this or not, but you also probably have the idle clockspeed bug when you enable ATi Overdrive (you start idling at 125 / 300 instead of 400 / 1200) which causes secondary monitors to flicker as well.

 

[refraxion]

Awesome, I'm in need of new drivers. My SCII units selected don't show up at all.

 

[samuelmorris]

It's not PEBKAC, the drivers have well documented bugs, and those bugs exist for everyone with a 5830 and up (at the very least).

Start a video that uses a hardware accelerated codec, your video card will drop to 400 / 900, secondary displays will flicker momentarily because the RAM clock changed, and the card will not clock up to full 3D clocks when you start a 3D application with that video open.

I'm not sure if running crossfire will effect this or not, but you also probably have the idle clockspeed bug when you enable ATi Overdrive (you start idling at 125 / 300 instead of 400 / 1200) which causes secondary monitors to flicker as well.

I do have the clock speed issue with a hardware acceleration, but I just turn it off if I want to play a game. No big issue. I have a second system next to me if I want hardware-accelerated video playing whilst I game. If I recall correctly though, isn't this an issue for geforces as well as radeons, if the video is actually playing while you game?

I do not have the idle clock speed issue with overdrive enabled.

 

[Unknown-One]

I do have the clock speed issue with a hardware acceleration, but I just turn it off if I want to play a game. No big issue. I have a second system next to me if I want hardware-accelerated video playing whilst I game.

So you've gone from "not a single problem" to "I'm ignoring the problems"

It is a big issue. These cards are pushed as media powerhouses, what with the 3D BluRay decoding and HDMI Audio capabilities. A key portion of the graphics card is malfunctioning.

And your solution, to a chunk of your video card not working, is to haul in an entirely separate computer? Come on, that's just ridiculous... A modern video card should have no trouble decoding video, it shouldn't immediately become a hobbled under-performing wreck the second you ask it to do some decoding (something which is was supposedly designed to do).

If I recall correctly though, isn't this an issue for geforces as well as radeons, if the video is actually playing while you game?

Not as far as I can tell. Clockspeed scaling works normally on my GTX260 with hardware accelerated video. I can't speak for their other cards, though.

 

[samuelmorris]

Well no, it's not a problem for me in the slightest. I don't see why you need to have something hardware-accelerate open in the background that you can't see. If you do run two displays and want to see a hardware-encoded video while playing a game you put up with reduced clock speeds for the privilege, I was pretty sure it was this way round for both brands.

The second PC is not my 'solution', the PC is there for several other reasons.

 

[Unknown-One]

Well no, it's not a problem for me in the slightest

That doesn't discredit it as a legitimate issue that a lot of people are pretty pissed about.

I don't see why you need to have something hardware-accelerate open in the background that you can't see. If you do run two displays and want to see a hardware-encoded video while playing a game you put up with reduced clock speeds for the privilege

Yes, I'm using multiple monitors, allowing me to watch video on one while I game on the other. The thing is, this doesn't just effect playing videos. Just leaving a paused video minimized in the background will prevent your card from running any faster than 400 / 900.

So if you so much as leave Firefox open with a YouTube video paused, your card will run slower than 2D mode even if you start a 3D application.

And what do you mean by "privilege" and why is it a requirement that we put up with a glitch? There's absolutely no reason for ATi to reduce the clockspeed to values lower than normal 2D clocks while playing video. This is a bug, not how the card or drivers should be behaving. All of the other PowerPlay states are handled in software by the driver, the UVD state isn't (slipped AMD's mind?), which causes the card to glitch and stick at the incorrect clockspeeds defined in its BIOS until the video is closed completely (allowing the card to exit the bugged UVD state).

 

[samuelmorris]

Lower than normal 2D Clocks? At idle my cards run at 250mhz, in game they should be 880mhz. If I leave a video open I get 500mhz. It's not ideal, but it only cuts my frame rate by about half. For a lot of games that's actually no issue, especially with two cards running, as crossfire still works.

 

[Unknown-One]

Lower than normal 2D Clocks? At idle my cards run at 250mhz, in game they should be 880mhz

Then you ARE seeing the idle clockspeed bug. A single HD5870 should idle at 400mhz core / 1200mhz RAM (the added HD5970 in your setup might change that slightly)

Attach a secondary display with your card(s) idling like that, and you'll get flickering.

If I leave a video open I get 500mhz. It's not ideal, but it only cuts my frame rate by about half. For a lot of games that's actually no issue, especially with two cards running, as crossfire still works.

I really can't fathom how you can discount this as not a big deal.

Your cards are crippling themselves for no good reason, cutting your frame-rate in half. That's a serious bug effecting key advertized functionality of the graphics card. Crossfire (eg, throwing more GPUs at the problem to offset the clockspeed glitching out) is not a legitimate solution.

You can mod your video BIOS so the drivers glitch to 3D clocks instead of sub-2D clocks when a video is playing. Still not a great solution (not to mention you risk bricking your card), but better than the massive performance hit (this also keeps your secondary monitors from flickering, since it keeps the memclock from being adjusted).

 

[samuelmorris]

I don't run a 5870+5970, I run two 6970s.

 

[Unknown-One]

Ah, sorry, that was LEVESQUE with the tri-fire setup.

Regardless, those still look like incorrect idle speeds. The core clock being low isn't really an issue, but the memory should stay at 1375mhz at all times in all power states on a 6970.

 

[samuelmorris]

Memory's 1375mhz at the moment, but the core's 500 as I have youtube open. I was hoping that a program like afterburner could be used to force a particular state, but apparently not it seems.

 

[Unknown-One]

Yeah, the glitched UVD PowerPlay state even ignores MSI Afterburner.

Only thing I've been able to do to work-around it is mod my video BIOS so that it sticks at 725 / 1000 (full 3D speed on my card) instead of sticking at 400 / 900.

 

[LEVESQUE]

It's not PEBKAC, the drivers have well documented bugs, and those bugs exist for everyone with a 5830 and up (at the very least).

Start a video that uses a hardware accelerated codec, your video card will drop to 400 / 900, secondary displays will flicker momentarily because the RAM clock changed, and the card will not clock up to full 3D clocks when you start a 3D application with that video open.

I'm not sure if running crossfire will effect this or not, but you also probably have the idle clockspeed bug when you enable ATi Overdrive (you start idling at 125 / 300 instead of 400 / 1200) which causes secondary monitors to flicker as well.

Took me 5 minutes to edit my BIOS, flash my 5970 and my 5870 with new clocks, and got rid of all those problems you just listed 5 minutes later... And I'm just using one 30'' screen BTW...

But I agree that for some, it's ''seems'' complicated to do... flash a BIOS... but the name of the forum is HARDforum. Not ''Sissy the Newbie'' forum.

We could argue all day between ''it's AMD's drivers fault'' or ''it's PEBKAC''. But since it took me 5 minutes to flash my 3 BIOS, I stay with the PEBKAC version of the story.

IMHO, flashing a video card BIOS is not different, and not more ''complicated'', then flashing your motherboard BIOS with latest version, and/or overclocking a CPU with cranking V core and tweaking BIOS settings... things everyone here are doing regularly everyday.

HARDforum...

 

[samuelmorris]

Yeah but you shouldn't HAVE to flash the BIOS of a card really, should you?

 

[Unknown-One]

But I agree that for some, it's ''seems'' complicated to do... flash a BIOS... but the name of the forum is HARDforum. Not ''Sissy the Newbie'' forum.

As the poster above me mentioned, we shouldn't HAVE to flash a hand-modified BIOS and risk bricking our cards and voiding our warranties in order to make them function more like they're supposed to.

And we aren't the only people that use AMD cards, you know. Normal users have them in their systems too, and are having the same issues. You can't just write off these problems because you personally have a work-around that you're happy with.

We could argue all day between ''it's AMD's drivers fault'' or ''it's PEBKAC''. But since it took me 5 minutes to flash my 3 BIOS, I stay with the PEBKAC version of the story.

How the heck is it PEBKAC when the problem is clearly within AMD's stock drivers and BIOS, and is in no way brought about user error?

You had to flash your card to work-around an issue that exists by default as part of the hardware and/or software. That is not PEBKAC.

What really bugs me is, on the surface, this seems like it would be a really simple bug for AMD to get resolved. They control the clockspeed of the card just fine the rest of the time, what makes UVD so problematic? But then you see the UVD clock screw up even when tools like MSI Afterburner are directly controlling clockspeeds...which makes me wonder if this is really a hardware issue that AMD can't resolve without recalling the entire 5800 series.

 

[samuelmorris]

I'm pretty sure it's a software issue, or else it wouldn't be in the 6 series as well.

 

[Unknown-One]

Or they just never bothered to resolve the issue for the 6 series either...you'd hope they wouldn't be that complacent, but you never know...

Edit: Oh joy, a new bug. At random, display mode switching will glitch, causing mode switches to take up to around 5 minutes to complete (during which time, all three displays are blank). The only way to make mode switches work correctly again is a reboot, but the glitch eventually resurfaces. I just added this to The AMD Driver Bug Thread as Bug #9

 

[bboynitrous]

Just an update

First 20 minutes of 11.2: Two blue screen memory dumps just doing normal desktop crap.

5hrs and lots of gaming and desktop stuff later after rolling back to 11.1a: Zero blue screen memory dumps.

Love ATI cards but jesus man that's one extreme to the next in driver quality and for a point release at that.

I had a bluescreen as well on my system. I had been perfectly stable for such a long time that almost immediatly after I update the drivers, I bluescreen? Had to have been the drivers. It's been 3 days i've installed the drivers and I haven't had one since and wasn't able to reproduce the other. Strange.

 

[Bbq]

unknown-one: So if I'm not mistaken, the process to flash the bios on the card is:
1. selecting the location in radeon bios editor to point to the directory with ati winflash, to download the bios from the card (in my case, XFX black edition 5850, clocked at 765/1125)
2. loading the bios into radeon bios editor
3. under clock settings, changing all the memory clocks to the full power speed (in my case, 1125mhz)
4. Leaving everything else alone (including voltages, etc)
5. using the flash option in radeon bios editor to re-flash the newly modified bios
6. profit ?

 

[Unknown-One]

unknown-one: So if I'm not mistaken, the process to flash the bios on the card is:
1. selecting the location in radeon bios editor to point to the directory with ati winflash, to download the bios from the card (in my case, XFX black edition 5850, clocked at 765/1125)
2. loading the bios into radeon bios editor

So far so good...

3. under clock settings, changing all the memory clocks to the full power speed (in my case, 1125mhz)
4. Leaving everything else alone (including voltages, etc)

Not quite. On the "clock settings" page in RBE, you want to edit "Clock info 03" to match the GPU clock, RAM clock,a nd voltage from "Clock info 00."

Leave the other clock info columns alone, otherwise ATi Overdrive will break and you wont be able to overclock.

5. using the flash option in radeon bios editor to re-flash the newly modified bios
6. profit ?

You got it

 

[RaiseR RoofeR]

So MFAA seems to now be an option for 5000-series cards. Has anyone tried it? Impressions?

 

[LEVESQUE]

As the poster above me mentioned, we shouldn't HAVE to flash a hand-modified BIOS and risk bricking our cards and voiding our warranties in order to make them function more like they're supposed to.

Not different then everyone else and their mothers overclocking their Intel or AMD CPUs everywhere. What's the difference? Same risk, void the warranty also.

Probably that 98% (or 99%) of the people posting on HardOCP are overclocking their CPU. Or else they would NOT come here. And they are also overclocking their GPU.

''Normal'' users don't post here. It's an enthusiasts site. If you are a ''normal'' user (and you are NOT, since you are posting here) then go do your whining somewhere else.

Why would you brick your card if you follow the instructions on RBE. Not different then following a guide to overclockcing your shiny new CPU by tweaking settings in your motherboard BIOS: voltages, etc. NO difference. Same risk.

 

[LEVESQUE]

You had to flash your card to work-around an issue that exists by default as part of the hardware and/or software. That is not PEBKAC.

So. Intel are selling the i7 2600K at 3.4Ghz stock. It's a DEFAULT stock value existing in my chip, part of the hardware. But, I can easily run it at 5.4Ghz with a voltage of 1.48v. It's my choice, after reading some random guides on the net.

Am I not doing the same thing? No difference. Intel default for the hardware is 3.4 @1.25v. But I made the choice to MODIFY those pre-existing settings in my BIOS and overclock my chip like crazy, voiding my warranty.

Just like 98% of the people coming here probably.

No difference. Be honest here... Did you try overclocking your CPU?

BTW, there is now 2 BIOS and a switch on the AMD 6xxx serie for a reason, and proving my point. They want us to play with those, and they know we will! There is 2 BIOS and a switch to flash the BIOS back on the 6xxx! I think it's clear why...

 

[Unknown-One]

Not different then everyone else and their mothers overclocking their Intel or AMD CPUs everywhere. What's the difference? Same risk, void the warranty also.

Probably that 98% (or 99%) of the people posting on HardOCP are overclocking their CPU. Or else they would NOT come here. And they are also overclocking their GPU.

What are you talking about? There's a HUGE difference.

Those CPU's work fine at their rated speeds out-of-the-box. You do not need to modify the chip firmware in any way to make them operate as they are supposed to.

These graphics cards, on the other hand, DO NOT work fine out-of-the-box. Since no driver update currently corrects the UVD clock issue, your only choice to get the card operating (closer to) normally is to modify your video BIOS.

''Normal'' users don't post here. It's an enthusiasts site. If you are a ''normal'' user (and you are NOT, since you are posting here) then go do your whining somewhere else.

Who cares weather normal users post here? That's entirely beside the point.

We ALL payed for a card that is supposed to work correctly out-of-the-box. None of us got what we payed for.

Why would you brick your card if you follow the instructions on RBE.

Because RBE really doesn't have a comprehensive instruction manual. I bricked my HD5850 once while attempting to implement a work-around for the UVD clock issue. Thankfully, I was able to recover the card (required the use of a second graphics card while the HD5850 essentially sat there dead).

If that hadn't worked, there's no way ASUS would have allowed me to RMA it and get it replaced. I would have had a $300 paperweight, for something that absolutely should not have been an issue in the first place.

Not different then following a guide to overclockcing your shiny new CPU by tweaking settings in your motherboard BIOS: voltages, etc. NO difference. Same risk.

Again, there's a HUGE difference here.

A simple setting change in something like a motherboard BIOS can always be reset. You flip the CMOS reset jumper and you're back to default values. There is minimal risk involved. More importantly, you were not REQUIRED to tweak with those settings in order to make the chip operate correctly.

When you actually have to flash the ROM, all kinds of things can go wrong. The risk is two-fold since you have to modify the BIOS as well. If the data isn't written correctly for any reason (power outage, incorrect modification, corrupt file) then you could be left with a paper weight. This process will also likely void your warranty.

You should NEVER have to void the warranty to get the card to run correctly in its stock configuration. That's called a DEFECTIVE PRODUCT.

BTW, there is now 2 BIOS and a switch on the AMD 6xxx serie for a reason, and proving my point. They want us to play with those, and they know we will! There is 2 BIOS and a switch to flash the BIOS back on the 6xxx! I think it's clear why...

Just because there's fail-safe on the 6000 series does not mean they can release a product that doesn't function correctly out-of-the-box.

And what about all of us with 5800's? We don't have that handy BIOS switch, but we do have malfunctioning cards. What's worse is, we could RMA them all we want, but we'll never get a properly working card in return since AMD seemingly hasn't even acknowledged the issue

 

[NathanP2007]

Unknown, whats the UVD problem? Iv never heard of it.

 

[Unknown-One]

Unknown, whats the UVD problem? Iv never heard of it.

Bug #2 from The AMD Driver Bug Thread

Symptoms:
a. GPU clockspeed is set below normal 2D clocks when a hardware accelerated video is playing or paused (includes DXVA accelerated and Flash video).
b. GPU clockspeed will not scale up to 3D clocks when such a video is playing or paused.
c. Being stuck below 2D clockspeeds leads to a massive performance hit.
d. Secondary displays flicker or jitter every time you open or close such a video.

Causes:
When the HD5000 series was first released, PowerPlay scaled vRAM clockspeed. It was discovered shortly thereafter that secondary displays flicker or jitter momentarily while vRAM clockspeed stabilizes at a new value. To remedy the flickering, recent drivers are supposed to keep the vRAM at full 3D clocks at all times. (If the clockspeed never changes, the secondary displays don't flicker). Core clock is still allowed to scale up and down normally.

Unfortunately, they forgot to handle one of the PowerPlay states. Specifically, the one that kicks-in when you launch any kind of hardware accelerated video. The card reverts to its old behavior and sets the lower vRAM clockspeed defined in its video BIOS for that PowerPlay state. The change in clockspeed causes displays to flicker momentarily. Performance drops harshly in 3D applications because the card isn't allowed to use 3D clockspeeds with that video open.


Workaround:
The only current way to work-around this driver bug is to modify the BIOS on your graphics card. The basic procedure involves dumping your current BIOS with ATi WinFlash, editing UVD Clock manually using Radeon BIOS Editor (Match UVD clocks and voltage to 3D clocks for your card), then flashing the modified BIOS back to your card with ATi WinFlash again.

I've gone ahead and pre-modified the BIOS for the HD5850 and HD5870. These are for reference model cards ONLY; if you have a non-reference design, you'll have to modify your video BIOS yourself.

Reference HD 5850 pre-mod BIOS - Updated 1/25/2011
Reference HD 5870 pre-mod BIOS - Updated 1/25/2011

- These BIOS are directly from ATi, they already have the DisplayPort voltage fix.
- UVD clock has been modified to full 3D clocks. This resolves issues with hardware accelerated video causing monitors to flicker, as well as 3D performance issues.


Official fix:
None. Unresolved as of Catalyst 11.2

 

[vulcZ]

I read on another forum today that a current work-around to the weird lag/stutter on the top/bottom right corner of the screen is to disable SHADOWS for your mouse cursor. People are stating that this fixes the problem. Another user didn't want to disable shadows, so he set his cursor profile to INVERTED and it seemed to fix the problem as well.

I would test this immediately except I wont be home for a few hours, and since I haven't actually verified myself that this is an effective workaround then YMMV.

If any has success with this please let us know!

 

[Ocellaris]

My 6850 no longer clocks down to UVD mode when playing accelerated videos using 11.2. I suspect it SHOULD be clocking down since other people are still having the problem, so I am undecided if this is a good thing or not.

 

[Unknown-One]

My 6850 no longer clocks down to UVD mode when playing accelerated videos using 11.2. I suspect it SHOULD be clocking down since other people are still having the problem, so I am undecided if this is a good thing or not.

What does UVD mode cause you card to do now, jump to full 3D clocks immediately (without a BIOS mod)?

I read on another forum today that a current work-around to the weird lag/stutter on the top/bottom right corner of the screen is to disable SHADOWS for your mouse cursor.

Woah, that did make it significantly better. Added it to the driver bug thread as a possible work-around. Thanks!

 

[Ocellaris]

What does UVD mode cause you card to do now, jump to full 3D clocks immediately (without a BIOS mod)?

Normally I am playing EVE in a window, the card is humming along at 775. Previously when I opened video the card would down clock to 300 (and EVE would run like shit obviously until I turned off video acceleration and relaunched everything). With 11.2 the card stays running at 775 when I open video. I have no explanation for it. No BIOS mod.

 

[Unknown-One]

That's really interesting. I'm still seeing the the drivers glitch-out on 5800 series cards; as usual, the card is dropped to whatever clockspeed is defined in the video BIOS for UVD, and it refuses to enter 3D mode as long as the video is open.

Anyone else with a 6800 series card seeing the same thing as ocellaris?

 

[shansoft]

That's really interesting. I'm still seeing the the drivers glitch-out on 5800 series cards; as usual, the card is dropped to whatever clockspeed is defined in the video BIOS for UVD, and it refuses to enter 3D mode as long as the video is open.

Anyone else with a 6800 series card seeing the same thing as ocellaris?

no problem on 6970.

 

[Unknown-One]

And we're sure these are hardware accelerated videos, right guys?

Sorry, I have to make sure. AMD has left this issue open so long I'd be genuinely surprised to see them address it now, but leave out the 5800 series where the problem originated.