when is 9.1 cat gonna be here ?

[lharp]

Just wanted to know if anybody new a word ..
Using 8.10 now did´nt like 11 or 12 ..

Seen beta released by some people but want the real things ...

[redrumy3]

next week they said (beta ppl that is)

[lharp]

So what are the new drivers suppose to fix ?
Or make better?

[Wiggy Fuzz]

hopefully stop crashing - recently 8.12 drivers just keep crashing in games with "display driver failed" problems. twice today in far cry 2 (which i still play)

[lharp]

I was gettin crashes all over with 8.10´s but think that was just my other psu ..
thank you antec .. L4D is fun again

[mastrdrver]

Multi cpu core enhancements, support for mirrors edge, performance profiles

Those are the three I heard that apparently was said by one of the beta testers.

[Ric2L]

hopefully stop crashing - recently 8.12 drivers just keep crashing in games with "display driver failed" problems. twice today in far cry 2 (which i still play)

Did you try the 8.12 Hot Fix for crossfired 4850's?

I have a 4870 1g and I havent had a single crash with the 8.12's while playing Far Cry2 or any other game. (knock on wood)

[Zuzzz]

I just installed a HIS4870 and while the 8.12 drivers did not work well with my old X800 they play very nicely (so far) with my new 4870.

Z

[Bon3thugz43v3r]

Can't wait for these new drivers I feel like Im a little kid waiting for Santa!

[Mr.Guvernment]

hopefully stop crashing - recently 8.12 drivers just keep crashing in games with "display driver failed" problems. twice today in far cry 2 (which i still play)

for me they cause a grey screen when trying to use overdrive overclocking.

[yeh_boi]

Are there going to be and BIG fps improvements? Otherwise no need for me to change.

[mastrdrver]

Suppose to be huge improvements in Crysis.

[Mr.Guvernment]

these drivers are supposed to be a performance bump drive for ATI, they were apparently holding back performance a little, then NVIDIA jumped ahead with a driver release so apparently ATI is going to do the same, originally it was to be 8.12's that were to do it, but the boost wasnt huge.

[exe163]

Suppose to be huge improvements in Crysis.

If that's true, can't wait the see numbers. This might change my decision on 4870x2 or GTX280 SLI

[vixro]

They are definitely taking their time.

[freakdiablo]

Hope they fix my problem. For some reason Vista won't boot with 8.11 and 8.12 for me.

[Mr.Guvernment]

then it must be some compatiblity with something else in your rig... cause 8.11 and 8.12 work fine for me with vista.. and i am sure many other people.

[bda1967]

No problems with Vista 64 here with the 8.11 or 8.12 drivers. I'm hoping to get some fps increases on COD5 with 9.1.

[deathman20]

then it must be some compatiblity with something else in your rig... cause 8.11 and 8.12 work fine for me with vista.. and i am sure many other people.

Aye zero issues with Vista 64, and im running Crossfire which seems to have even more issues.

[gian84]

just want to ask, is Ati going to make a physics driver similar to nVidia?

[redrumy3]

just want to ask, is Ati going to make a physics driver similar to nVidia?
nope they said no on that, was an article somewhere about it i have to find it hmm

[Rich'[ard]]

hopefully stop crashing - recently 8.12 drivers just keep crashing in games with "display driver failed" problems. twice today in far cry 2 (which i still play)

OMG. ditto...ditto...

[gian84]

:( I hope they realize they're putting their gpu's at a disadvantage to nvidia...

[deathman20]

:( I hope they realize they're putting their gpu's at a disadvantage to nvidia...

Yes and no....

Yes from the aspect that certain games will be-able to do on board physics processing.

And no from the fact that it will give the games 100% GPU control for graphics and also no worrying about physics drivers as well inside the standard Driver package.

Its a double edge sword they know, nVidia offered it to them and they refused. As well not all game makers go with the physics that nVidia offers as well, a lot use the intel "Havok" version.

[gian84]

hmmm i suppose having the physics calculated in the GPU does have its disadvantages... but it shouldn't be as stressful to them as it is to the CPUs right? Like in 3dmark vantage, those with phyx enabled nVidia get like 30000+ in cpu score! while my CPU for example only does about 6000... so say 90%GPU control for graphics and 10%GPU control for physics should be more beneficial especially in the heat of battles in games right? i.e. less fps drops.... I thought I read somewhere that Ati would have Havok calculated in their GPU.. i suppose they refused that too, or intel didn't let them.... or i misread :p

[deathman20]

hmmm i suppose having the physics calculated in the GPU does have its disadvantages... but it shouldn't be as stressful to them as it is to the CPUs right? Like in 3dmark vantage, those with phyx enabled nVidia get like 30000+ in cpu score! while my CPU for example only does about 6000... so say 90%GPU control for graphics and 10%GPU control for physics should be more beneficial especially in the heat of battles in games right? i.e. less fps drops.... I thought I read somewhere that Ati would have Havok calculated in their GPU.. i suppose they refused that too, or intel didn't let them.... or i misread :p

No if ATI did go any route it is with Intel with there Havok. They support them and have for a long time.

While the GPU might be better at calculating physics, CPU is more than capable of doing so as well expecially when you get multiple cores that sit and do nothing.

While that CPU test, really its unfair at showing the strength of the CPU, since its not doing the work the GPU is. The point of a CPU test is to show how good the CPU is not the GPU/CPU combo.

[Firestrider]

No if ATI did go any route it is with Intel with there Havok. They support them and have for a long time.

While the GPU might be better at calculating physics, CPU is more than capable of doing so as well expecially when you get multiple cores that sit and do nothing.

While that CPU test, really its unfair at showing the strength of the CPU, since its not doing the work the GPU is. The point of a CPU test is to show how good the CPU is not the GPU/CPU combo.

Really? From what I've heard most calculations in games can be done a lot faster on the GPU. AMD has been working on getting AI calculations GPU accelerated. Ray tracing which will enhance all aspects of game experience will be largely done by the GPU.

With this there will be less and less importance of the CPU in a game, and the CPU will only be needed for the driver and operating system. Probably some things in the operating system can be done on the GPU now.

[deathman20]

Really? From what I've heard most calculations in games can be done a lot faster on the GPU. AMD has been working on getting AI calculations GPU accelerated. Ray tracing which will enhance all aspects of game experience will be largely done by the GPU.

With this there will be less and less importance of the CPU in a game, and the CPU will only be needed for the driver and operating system. Probably some things in the operating system can be done on the GPU now.

Yes certain aspects are defiantly accelerated by the GPU but as said at what cost? 10, 20, 30% of FPS? Then you have say today a typical dual core CPU that some games use fully, and others say they support it but one is pegged at 100% and the other is at 15% and the game is dragging. AI can be done and is benifical to be run on the GPU, as well as Physics as shown. Don't think there is many other things that is viable to be run on the GPU. But as said if your letting the CPU idle when it does have the power to use it, while it might be less energy effiecent at doing so, it still can be utilized.

If Ray Tracing is introduced to games, there won't be any spare GPU power to give up if you want reasonable frames. Ray Tracing in real time is a system drain.... huge system drain. While it gives definatly a more realistic look to everything, it takes that much more power to render each frame. I mean your talking about photo realistic images, yes even more so then whats shown in some games today which do look great. This just takes it to the next level.

As well while the CPU might be less important per say, there is still alot the CPU is doing. Drivers, Sounds, Input devices, OS, Other applications in the background, Level information. I mean still even like Supreme Commander for instance can really drag down a Dual core for instance, with AI calcs. Pop a quad core in and the AI has its own dedicated CPU and can see double the frame rates. While if the AI is taken off the CPU, then whats the point of manufactures coming out with even cored CPU's then there is now?

If anything though, AI being done on the GPU is IMO more important than running physics currently, especially if the AI becomes way smarter since they have more overhead. Curious how the workload will be done with CF systems though. Even with SLI I wonder how the physics are handled on the cards.

[freakdiablo]

then it must be some compatiblity with something else in your rig... cause 8.11 and 8.12 work fine for me with vista.. and i am sure many other people.

Never said it was Vistas fault, it just won't display anything after the startup loading bar. Just have no idea what it could be.

[Shiggity]

If Ray Tracing is introduced to games, there won't be any spare GPU power to give up if you want reasonable frames. Ray Tracing in real time is a system drain.... huge system drain. While it gives definatly a more realistic look to everything, it takes that much more power to render each frame. I mean your talking about photo realistic images, yes even more so then whats shown in some games today which do look great. This just takes it to the next level.

Nvidia has added a ray-tracing element into their drivers. It is called "ambient-occlusion"

http://www.ocforums.com/showpost.php?p=5925519&postcount=1

I am unsure as to whether it is built into PhysX or whether it is a stand alone type thing like AA or AF.

[Firestrider]

Yes certain aspects are defiantly accelerated by the GPU but as said at what cost? 10, 20, 30% of FPS? Then you have say today a typical dual core CPU that some games use fully, and others say they support it but one is pegged at 100% and the other is at 15% and the game is dragging. AI can be done and is benifical to be run on the GPU, as well as Physics as shown. Don't think there is many other things that is viable to be run on the GPU. But as said if your letting the CPU idle when it does have the power to use it, while it might be less energy effiecent at doing so, it still can be utilized.

If Ray Tracing is introduced to games, there won't be any spare GPU power to give up if you want reasonable frames. Ray Tracing in real time is a system drain.... huge system drain. While it gives definatly a more realistic look to everything, it takes that much more power to render each frame. I mean your talking about photo realistic images, yes even more so then whats shown in some games today which do look great. This just takes it to the next level.

As well while the CPU might be less important per say, there is still alot the CPU is doing. Drivers, Sounds, Input devices, OS, Other applications in the background, Level information. I mean still even like Supreme Commander for instance can really drag down a Dual core for instance, with AI calcs. Pop a quad core in and the AI has its own dedicated CPU and can see double the frame rates. While if the AI is taken off the CPU, then whats the point of manufactures coming out with even cored CPU's then there is now?

If anything though, AI being done on the GPU is IMO more important than running physics currently, especially if the AI becomes way smarter since they have more overhead. Curious how the workload will be done with CF systems though. Even with SLI I wonder how the physics are handled on the cards.

That's why we need more and more powerful GPUs, and better scaling between them if there will ever be real-time ray tracing. Ray tracing is not only useful for graphics but also for physics and audio too.

[Shiggity]

That's why we need more and more powerful GPUs, and better scaling between them if there will ever be real-time ray tracing. Ray tracing is not only useful for graphics but also for physics and audio too.

Well supercomputers that are setup for fusion / weather / nanotech research can do this now. What may ultimately happen is that there could be a giant graphical super computer run on a cloud network with all the computation being done on their side and then having the game streamed to you over a very high speed internet connection in real time.

Real-time raytracing on a PC in the mainstream is so far away. Not only because of the hardware requirement, but because they'd have to re-design most of the software, which can take way longer than any hardware problem.

[mattgmann]

i just ran a few benches with the beta....nothing to write home about, negligible differences.

Crysis Warhead Enthusiast settings 1680x1050

8.12 2xaa=31.24fps
4xaa=30.22fps

beta 2xaa=31.69fps
4xaa= 30.74fps

Far Cry 2 Ultra 1680x1050

8.12 2xaa=58fps
4xaa= 41.8fps

beta 2xaa=60.83fps
4xaa= 40.57fps


system is q6600@3.8ghz, 2x 4870 in CF @810/1190

[mattgmann]

Also just ran 3dmark06, bumped the score up a little to ~20.7k I'll give vantage a whirl, but I doubt there'll be any huge gains.

These drivers may have more of an effect at higher resolutions. I only have 22" monitors though. I'd take it to the 42" for testing, but I just moved, and my new office is in the basement. Not lugging this rig up the stairs for another crappy driver release.

[deathman20]

That's why we need more and more powerful GPUs, and better scaling between them if there will ever be real-time ray tracing. Ray tracing is not only useful for graphics but also for physics and audio too.

Well more powerful yes but we'll be waiting a few years for that unless you start buying 2, 3 or even 4 GPU's for your PC to run it. Let alone the driver and software re-writes that would be involved. Going to be a looong time before we see it in any games.

I'll just be happy when they release 9.1's and hopefully it will give even more speed.

[Firestrider]

What may ultimately happen is that there could be a giant graphical super computer run on a cloud network with all the computation being done on their side and then having the game streamed to you over a very high speed internet connection in real time.

I don't think this will ever happen. The bandwidth you would need would be like 30MB/s to display 60 FPS at gamer resolutions. Not only that but there would be so much stuttering because of the latency. I would be nice, but I think technology evolves faster than putting in new infrastructure for networks.

[AngelicPenguin]

support for mirrors edge

Already finished it...a bit late on that one eh? I just renamed the exe to MassEffect.exe and I got crossfire support.

[Mekks]

i just ran a few benches with the beta....nothing to write home about, negligible differences.

Crysis Warhead Enthusiast settings 1680x1050

8.12 2xaa=31.24fps
4xaa=30.22fps

beta 2xaa=31.69fps
4xaa= 30.74fps

Far Cry 2 Ultra 1680x1050

8.12 2xaa=58fps
4xaa= 41.8fps

beta 2xaa=60.83fps
4xaa= 40.57fps


system is q6600@3.8ghz, 2x 4870 in CF @810/1190

what beta? lol the 8.12 hotfix was NOT the beta. nobody has seen the 9.1 beta yet.

[NewbieOneKenobi]

Guys, in my view, the many 4850 problems are or may be caused by the following things:

1. TDR, which is itself a bottomless hole. It is basically everything that causes a GPU request to take more than two seconds to perform, upon which the system thinks something is wrong. After that, the driver either recovers or not. Much of the time (most in my system), the driver recovers but the game crashes. Much of the talk about what exactly causes this is speculation. I suppose no one wants to appear guilty of causing this situation. Fact stands that CCC and ATI's drivers have something to do with this because of problems with .NET Framework or possibly C++ runtimes. Sometimes closing CCC before launching a game helped in earlier drivers. In 8.12, it doesn't seem to. Vista obviously has something to do with this by enacting its TDR procedure, which isn't probably necessary, so it's something the operating system imposes on people and their hardware. It's probably beneficial in some ways, but I don't want it to crash my games.

2. Heat - obviously influencing the TDR because it's ultimately TDR fail which crashes the game. Contrary to what is sometimes claimed, the card is not unaffected by its high running temps. I doubt any GPU is adapted to over 100C temps and we are talking such. 4850 is capable of running over 100C in idle. The fact that 4870 has stronger power circuits or similar such components, is a clue as well.

3. The fan problem. You don't run a fan at 20% for power saving. It must have to do with noise. The noise of the fan above, say, 37%, is awful and almost anything above factory speed of 20% is distinctly audible. Someone should have picked a better fan.

4. Heatsink and its implementation. The stock heatsink is cleverly designed and covers what's needed, but it holds on tape. Not the old kind like you use for memory chips and bad CPU coolers, but a jelly kind. Thick. At any rate, I suppose a thin layer of paste would have done better. A single slot heatsink and a 70 mm fan is not enough to cool this kind of card, either.

5. Something is wrong with the way temp is handled, either in the sensors, or, more likely, the driver. The card agrees to work in greater heat with the stock cooler than with better replacement coolers, where 75C seems to be the deadline. I suspect I'm having the same problem with a fan that runs off a Molex tip and so the card driver thinks the card's running on passive.

6. As people say, multicore. I honestly don't know what gives, but it seems multicore support is a big problem to code properly and it was already like that when the first HT CPUs came in. Anyone remember the USB and HT problem in Windows XP? We don't have this kind of mess now, but I guess people are quite well off with old single core AMD 4000's in games, or they would be if they had the same kind of RAM and GPU's that we do in our multicore systems.

7. RAM stuff. People report problems with shared RAM. Some people claim fixes for RAM call erros fix the 4850 crash problem. Don't know. They don't for me. But my mobo sucks (Asus P5Q-E that can't even remember BIOS settings).

8. General hardware (read: driver, mostly, I guess) incompatibility. People say updating drivers for all sorts of things (mostly motherboard/chipset related) has helped them. Means stuff doesn't work together well these days, much like in the old times.

I suppose what can be done by the ATI team is multicore support, proper handling of temps and fans in the driver and there's no excuse here, something to escape game crashes due TDR. They should also most definitely make sure that the drivers don't hang games because of some errors with C++ or .NET runtimes. By the way, updating .NET (3.5, SP1) didn't help me. Memory calls can probably be helped somewhat, too.

By the way, I've made my mind that stuttering gameplay must have some connection with the occurence of errors, probably because of enabling TDR failures. OC'ing my card, which was very well able to run at 690/1148 per CCC's Autotune feature, improved the smoothness of gameplay and thus, I think, reduced the chance of errors due TDR failures caused by taking too long to perform a request. Perhaps people with highly OC'ed CPU's don't notice this problem for the same reason - their systems just don't get to stutter, meaning the lag that causes TDRs isn't there.

By the way again, the 8.12 hotfix didn't fix it for me, either.

And if we are at it, from scouting the AMD forums and googling, but mostly the AMD forums, I've found the information that uninstalling Windows Update KB952287 helps (it's an SQL server thing that has something to do with .NET, apparently). Turning off ATI External Event in services.exe also seems to help some people. Adjusting the Boot tab (advanced) in MSConfig to force the correct number of processors and 3072 MB RAM (unless you have less, because if you insert 4096, it forgets) seems to help people avoid BSoD's, but don't know about game crashes. And I've had no BSoD's for a loooong time anyway.

[deathman20]

NewbieOneKenobi

2. Heat - obviously influencing the TDR because it's ultimately TDR fail which crashes the game. Contrary to what is sometimes claimed, the card is not unaffected by its high running temps. I doubt any GPU is adapted to over 100C temps and we are talking such. 4850 is capable of running over 100C in idle. The fact that 4870 has stronger power circuits or similar such components, is a clue as well.

While might be true but its not sometimes claimed it is what the manufacture has set into the card. Fan speed ramp up by default doesn't start til around 69C and isn't fully ramped up to 100% til 105C. 120C if not mistaken is what the card is rated for. This is set by ATI itself or the card manufactures. Of course as we know, cooler is better but what says goes. Heck even people put there CPU's above max temperatures recommended for the CPU and they still work. Mind you it is a lot cooler but still applies.

3. The fan problem. You don't run a fan at 20% for power saving. It must have to do with noise. The noise of the fan above, say, 37%, is awful and almost anything above factory speed of 20% is distinctly audible. Someone should have picked a better fan.

Its not problem. Its not for power savings and who ever said it was power savings, its always temperature related. This is the way its been with a lot of single slot cards. While nVidia does take a different stance on it and fits dual slot coolers on the warmer cards, to help keep the heat under control to keep noises down. Now everyone has something different to say about fan noise of course, 35-45% is barely any difference but start creeping upwards to 50-55% is defiantly audible and 60%+ hopefully your playing a game because its out right loud (mind you I have a 2x120 hole with 2 fans on the side of the case right where the cards are).

Fan doesn't make any difference in this case, all little fans will be loud period especially trying to put the proper air out to help cool the card. If anything yes a better heatsink which would leave only a dual slot card.

Added note, think partial reason why the GPU's have low fan speeds as well is to reduce the variance in temperatures going from idle to load as much so it doesn't do alot of expanding/shrinking.

4. Heatsink and its implementation. The stock heatsink is cleverly designed and covers what's needed, but it holds on tape. Not the old kind like you use for memory chips and bad CPU coolers, but a jelly kind. Thick. At any rate, I suppose a thin layer of paste would have done better. A single slot heatsink and a 70 mm fan is not enough to cool this kind of card, either.

Its fine, as for the tape for the memory chips, what company doesn't use this method to transfer heat to a heatplate? nVidia sure does, ATI does. This has been like this since it was released, and im talking about stock coolers, not aftermarket or revamped board designs by manufactures.

As for the single slot heatsink with fan, it is enough to cool it, mind you depending on if you get a good card. I got 2 cards, 1 runs about 10C warmer than the other card (with fan speeds adjusted), its not the heatsink or fans fault, its the process of the die or the whole board in itself. Now idling yes cards idle at a nice 40-45C for me, load anywhere between 65-80C depending on the day, the game, and card. While yes to get ideal temperatures (say 60-70C) so it really doesn't heat up your case, yes the heatsink and fan is not proper, but then again most heatsinks/fans that come with said cards aren't designed the best to cool them, hence why there is aftermarket heatsinks.

5. Something is wrong with the way temp is handled, either in the sensors, or, more likely, the driver. The card agrees to work in greater heat with the stock cooler than with better replacement coolers, where 75C seems to be the deadline. I suspect I'm having the same problem with a fan that runs off a Molex tip and so the card driver thinks the card's running on passive.

Are you covering and sinking up all the proper items to give it the best temperature? You might be overheating a chip and causing the issues. Many people run without the stock heatsink even non-reference boards run and some run warm.

6. As people say, multicore. I honestly don't know what gives, but it seems multicore support is a big problem to code properly and it was already like that when the first HT CPUs came in. Anyone remember the USB and HT problem in Windows XP? We don't have this kind of mess now, but I guess people are quite well off with old single core AMD 4000's in games, or they would be if they had the same kind of RAM and GPU's that we do in our multicore systems.

People say multicore since the drivers run better this way. The way that drivers handle information mutlicore is better now, but its more so with Crossfire/SLI rigs. Not so much with a single core card. Multicore support has come a long way since way back when it was released. While it takes time, they've been making the drivers that seem to benefit from multi core rigs for 2 years now if not mistaken. While it might not be multi-threaded, it might of been just the way the code was written to help the OS to move the process to an unused core. Don't look at me im not a programmer, but the benefits of multi-cored rigs with drivers these days for the GPU's its defiantly a good step in the right direction.

7. RAM stuff. People report problems with shared RAM. Some people claim fixes for RAM call erros fix the 4850 crash problem. Don't know. They don't for me. But my mobo sucks (Asus P5Q-E that can't even remember BIOS settings).

Shared ram? That is more referred to with on board video cards, where it can have a pool of ram right next to the gpu, but as well use system ram. Its also known as hypermemory. Other then that I do not know what your calling out on this one.


8. General hardware (read: driver, mostly, I guess) incompatibility. People say updating drivers for all sorts of things (mostly motherboard/chipset related) has helped them. Means stuff doesn't work together well these days, much like in the old times.[quote]

There is hundreds of different system configs out there. Of course there is going to be incompatibilities. Each system is unique, just like I might not have an issue with my 4850's with any driver issues out there, but a friend might with the exact same system. Perfect example was a friend and I a while back had extremely close configs. He had a different monitor and different hard drives but otherwise exactly the same. Well I had my system in raid and he was having issues with his card in a non-raided system. While we got it fixed its the point still, not everyone is a like and there will always be little fixes to the code that can improve something. Think that would have to be in programing 101.


[quote]I suppose what can be done by the ATI team is multicore support, proper handling of temps and fans in the driver and there's no excuse here, something to escape game crashes due TDR. They should also most definitely make sure that the drivers don't hang games because of some errors with C++ or .NET runtimes. By the way, updating .NET (3.5, SP1) didn't help me. Memory calls can probably be helped somewhat, too.

It always updating what they have. But fans and temps is not a driver issue, its controlled by the bios, unless you manually adjust it.

By the way, I've made my mind that stuttering gameplay must have some connection with the occurence of errors, probably because of enabling TDR failures. OC'ing my card, which was very well able to run at 690/1148 per CCC's Autotune feature, improved the smoothness of gameplay and thus, I think, reduced the chance of errors due TDR failures caused by taking too long to perform a request. Perhaps people with highly OC'ed CPU's don't notice this problem for the same reason - their systems just don't get to stutter, meaning the lag that causes TDRs isn't there.

Stuttering is the time that it takes to process 1 frame and tell the other gpu hey I need the next frame now. There is a delay and certain games you can see it a lot more than others. GPU/CPU speed has nothing to do with it because people can have high clocked systems and still see it. Its partial mind game, as well it is an actually happening all the time in CF/SLI setups. Only way to improve it is to further develop and more generations of cards to help fix the issue which seems to be a delay of timing in communication and processing time between the separate GPUs.

[mattgmann]

deathman,

I don't understand why you're contradicting Newbie on all of his points. You sound like an ATI rep. I don't agree with everything he put, but at least half of it is valid. ATI, no matter what you say, has a big problem with cooling implementation, and driver support. Catalyst has a much more polished interface that Nvidia's program, but is constantly riddled with ridiculous little problems.

Just because the manufacturer says it's ok isn't a valid argument. So what if the fan keeps it "cool enough". Asking for higher quality is always alright, and the fact is, ATI cheaped out on the cooling solutions for the new cards......especially the 4850.

I'm about sick and tired of my 4870cf setup. I should have never switched to ATI cards. They're power hungry, and the performance is over rated. Sure, it'll beat down some nvidia cards in overall numbers, but it really sucks where it counts. FarCry 2 is unplayable because of stuttering. Period. I would have better crysis framerates with a gtx280...maybe with a 260. I don't need any game to play at 100fps. But it'd sure be nice if they all played at well above 30fps, and be able to use the same quality settings on every game. I have had NO compatibility issues, and can even deal with the noise of the cards. But when it comes down to it, the real world performance isn't there, and I'm not satisfied. The only number I'm concerned with, is the number of dissatisfied ATI customers. Any by reading online forums, the number is disproportional compared with that of Nvidia customers.

[deathman20]

deathman,

I don't understand why you're contradicting Newbie on all of his points. You sound like an ATI rep. I don't agree with everything he put, but at least half of it is valid. ATI, no matter what you say, has a big problem with cooling implementation, and driver support. Catalyst has a much more polished interface that Nvidia's program, but is constantly riddled with ridiculous little problems.

Just because the manufacturer says it's ok isn't a valid argument. So what if the fan keeps it "cool enough". Asking for higher quality is always alright, and the fact is, ATI cheaped out on the cooling solutions for the new cards......especially the 4850.

I'm about sick and tired of my 4870cf setup. I should have never switched to ATI cards. They're power hungry, and the performance is over rated. Sure, it'll beat down some nvidia cards in overall numbers, but it really sucks where it counts. FarCry 2 is unplayable because of stuttering. Period. I would have better crysis framerates with a gtx280...maybe with a 260. I don't need any game to play at 100fps. But it'd sure be nice if they all played at well above 30fps, and be able to use the same quality settings on every game. I have had NO compatibility issues, and can even deal with the noise of the cards. But when it comes down to it, the real world performance isn't there, and I'm not satisfied. The only number I'm concerned with, is the number of dissatisfied ATI customers. Any by reading online forums, the number is disproportional compared with that of Nvidia customers.

Im not doing to all of I was adding information in there as well. As well not an ATI rep, I've done the same thing before when I had an nVidia card. Mind you I've had more problems with nVidia hardware then ATI but both are great. Right now if I where to get a new card I'd suggest a GTX285 if it was in budget.

Cooling its not an issue from there prespective. ATI cards have always ran hotter than nVidia. Its been that way for ages and probably will be that way for a long time to come. Its not like I agree with it, I perfer my cards to run under 70C all the time but also I got to take into account that I've OC'ed my cards and I don't expect the stock cooler to cope with the increased heat output either. Some people expect it to, since nVidia typically oversizes there coolers to cope with heat very well with less noise, so OCing and getting good temps are still viable.

Each to there own, while I've had my share of issues with both companies, nVidia I've had the most issues with. Each company has there advantages and disadvantages and you'll have to weigh those out in your own preference in the end from experience, not from someone elses information since each system is not the same.

[mattgmann]

I guess we're on the same page. Nvidia isn't without it's flaws. I'm just disappointed that I've had these ATI cards almost since they came out, and they still don't have proper support (IMO).

[NewbieOneKenobi]

Deathman,

1. And why would ATI say that 120C is OK? If so, it still doesn't make the card take that heat well. The card overheats and has problems at the same time. I am going to link these two and ATI specs saying 105 or 120C is OK won't dissuade me here simply because I don't have this kind of trust in ATI. If they can be wrong on other points or make imperfect drivers, they can be wrong on the temps too.

2. I was thinking that perhaps some of the reason the fan was set to such a low rotation speed as to allow extremely high temperatures might have to do with the fact that the fan is so loud. As if correcting the loudness of the fan by turning rotation speed down. However, this is rather unlikely. It sounds too much like a conspiracy theory to me. I suppose it's just two bad things about the same card that aren't probably related. But why was fan speed control not enabled then? This feature is rather old already. It should be there. And the lack of it caused people to have errors until they manually edited the profiles as someone discovered possible to do and kindly shared.

3. Tape is substandard and companies have used paste. Most of the cards I've bared have had paste. I don't deny that modern cards might generally all use tape, although I wouldn't be sure of that, but paste is better and it's been around in video cards for ages.

The stock cooler is insufficient as a fact. It does not manage to prevent the card from overheating in PCs owned by people who know what they are doing. The stock cooler should keep the card operational in average circumstances and average uses for that card. This card is obviously not a HTPC product, it's meant for gaming. This means the average gamer should plug this card into a decently built (read: average) computer and be able to play for a couple of hours, ideally the whole night, without overheat issues. Try playing for a couple of hours without a CTD with some of the 4850 cards.

The point I was more directly referencing, however, is that the 75C barrier somehow manages to make more difference when you use a replacement cooler than when you use the stock one. This came up here on one occasion. I had the problem too. When I tried running Accelero passive, 75C would not be tolerated and games would hang with no hope of letting me play. However, with the stock cooler, 105C was the idle temp and while I generally turned the fan on 100% to reduce the temps before launching a game, I'm pretty sure more than 75C was tolerated... and in fact, I played games before discovering how to manage profiles.

4. I have the same issues as before replacing the cooler and the stock cooler, according to you, was sufficient. ;) Also, it covered everything. Not perfectly, I say, but better than leaving it bare...

I have a little heatsink on every memory chip and power unit, thanks to Accelero. :)

5. People have tons of compatibility issues, but when so many people begin to have problems with the same particular card, it's likely the card.

6. Fans are controlled by the driver and so is voltage. Responses to temp sensors are also a matter of the driver. If you take a look at an ATI CCC profile (an .xml file), you will see frequencies, voltage, fan speed and everything you need right there, in a driver settings file. :)

7. Stuttering and other laggy behaviour is, however, most often caused by insufficient horsepower under the case. I don't believe modern games make the best possible use of the hardware power they get since it's more acceptable these days to tell players to upgrade their hardware for a game, but in most cases, it will simply mean that the computer is too weak to run the game at a better, smoother, more fluent rate. Therefore, if you OC and achieve stability in that, your framerate grows as a result of the power increase. You are also less likely to spend more than 2 seconds performing a GPU request simply because you have more power. This seems to be the case with me now.

And I perfectly realise much of my post is guesswork or even conjecture when it comes to causes or links, but I believe ATI has a lot of explaining to do and a lot of fixing. The state of 4850 is unacceptable and I'm going to vote with my money as a result. Next card is nVidia unless nVidia screws up badly or ATI improves. I'm not an nVidia fanboy. My 7600GT, which was a great card, died on me, so all in all, nVidia should be seen as causing me a greater loss, but at least the card got RMA'd and I didn't have issues with games, like this not working, that not working (yes, I know gaming companies probably have an nVidia bias since you see, "the way it was meant to be played," everywhere), and I certainly didn't get half a year of grief. I wished I could afford selling the 4850 as defective and buying a different one.

[deathman20]

Deathman,

1. And why would ATI say that 120C is OK? If so, it still doesn't make the card take that heat well. The card overheats and has problems at the same time. I am going to link these two and ATI specs saying 105 or 120C is OK won't dissuade me here simply because I don't have this kind of trust in ATI. If they can be wrong on other points or make imperfect drivers, they can be wrong on the temps too.

Not saying its OK by any means. I don't vid cards should be tempting fate at above 80C at all it should reduce clock, voltages at that temp and 70C would be ideal. While they might be wrong on temps (look at the 4780X2 running at around 90-100C) cards can obviously take the temps, if its above and beyond the reference board design, its not ATI's fault but the manufactures if they deviate from the design.

2. I was thinking that perhaps some of the reason the fan was set to such a low rotation speed as to allow extremely high temperatures might have to do with the fact that the fan is so loud. As if correcting the loudness of the fan by turning rotation speed down. However, this is rather unlikely. It sounds too much like a conspiracy theory to me. I suppose it's just two bad things about the same card that aren't probably related. But why was fan speed control not enabled then? This feature is rather old already. It should be there. And the lack of it caused people to have errors until they manually edited the profiles as someone discovered possible to do and kindly shared.

Damn fan is loud hehe and I think thats why they kept it that was so idle and gaming its relatively cool but as well keeps close tempatures. While fan editing has been in profiles if you install CCC, you can enable them as well via the panel for a few versions now. Enough complain about it and it happens. But what amazes me more is the manufactures do not change there bios profiles to increase the fan speed from reference. While ATI is at fault originally they have added the option to up it, but also the card manufactures could increase the fan speed via bios and help out alot too.


3. Tape is substandard and companies have used paste. Most of the cards I've bared have had paste. I don't deny that modern cards might generally all use tape, although I wouldn't be sure of that, but paste is better and it's been around in video cards for ages.

The stock cooler is insufficient as a fact. It does not manage to prevent the card from overheating in PCs owned by people who know what they are doing. The stock cooler should keep the card operational in average circumstances and average uses for that card. This card is obviously not a HTPC product, it's meant for gaming. This means the average gamer should plug this card into a decently built (read: average) computer and be able to play for a couple of hours, ideally the whole night, without overheat issues. Try playing for a couple of hours without a CTD with some of the 4850 cards.

The point I was more directly referencing, however, is that the 75C barrier somehow manages to make more difference when you use a replacement cooler than when you use the stock one. This came up here on one occasion. I had the problem too. When I tried running Accelero passive, 75C would not be tolerated and games would hang with no hope of letting me play. However, with the stock cooler, 105C was the idle temp and while I generally turned the fan on 100% to reduce the temps before launching a game, I'm pretty sure more than 75C was tolerated... and in fact, I played games before discovering how to manage profiles.

Yes I agree companies use tape, and paste, but most commonly tape is found on nearly all Memory sinks on cards, and even on voltage moduals. Definatly paste is better but also depending on what paste they use, paste can wreak the card, and from an assembly line perspective its easier to use tape. There for if a machine screws up on putting paste in the proper places its not going to fry the card, if said paste is conductive of course.


While the stock cooler might be insufficient, it depends. Like I said I have 2 cards, with 2 vastly different in tempatures. Matter of fact I should RMA the one because it does run very warm by my standards and even compared to the other card. Now while I might be the on the lucky end, 10 people that I've personally known got a 4850, out of the 10, 1 had issues with it. While 10% might be small but as well it wasn't heat related as well. Most of them just game and could care less about the system just as long as they have a good experiance playing.

4. I have the same issues as before replacing the cooler and the stock cooler, according to you, was sufficient. ;) Also, it covered everything. Not perfectly, I say, but better than leaving it bare...

I have a little heatsink on every memory chip and power unit, thanks to Accelero. :)

Ah wish I could do that, CF and aftermarket heatsinks is just a no go :(

5. People have tons of compatibility issues, but when so many people begin to have problems with the same particular card, it's likely the card.

Agree with you on this to a point, your only catching part of the user base, the other side isn't having issues at all. What should be done to get an idea, is make a poll with 4850's in the forum seperated by card manufacture and see where all the issues are really coming from. It might be just a few manufactures that have the problems or a certain line of there cards. If anything it will narrow the area down to which cards have the true problems and which ones have a higher chance of no overheating issue.

6. Fans are controlled by the driver and so is voltage. Responses to temp sensors are also a matter of the driver. If you take a look at an ATI CCC profile (an .xml file), you will see frequencies, voltage, fan speed and everything you need right there, in a driver settings file. :)

As for temp responses its not driver related, 2D to 3D switching is driver related but actual temp and fan speed is controlled via bios. You can see it in dos even where fan speeds will change as the card gets hotter (limiting air flow). Yes you can modify the file, and that will over-ride settings thats in the bios, then you are really relaying on the drivers to handle it. I did that before, and after I found the right software to flash I went that route. No need for me to chance my cards via driver when if all else fails the card itself will protect itself.

7. Stuttering and other laggy behavior is, however, most often caused by insufficient horsepower under the case. I don't believe modern games make the best possible use of the hardware power they get since it's more acceptable these days to tell players to upgrade their hardware for a game, but in most cases, it will simply mean that the computer is too weak to run the game at a better, smoother, more fluent rate. Therefore, if you OC and achieve stability in that, your frame rate grows as a result of the power increase. You are also less likely to spend more than 2 seconds performing a GPU request simply because you have more power. This seems to be the case with me now.

There is so much on the "Stuttering" issue. At first I didn't believe it, heck didn't even notice it til 1 game, GRID. I could see it in that, barely but it was there. I mean getting 80-140FPS maxed out, figured I wouldn't see it but it is. Vsync defiantly helps at eliminating it though. Its a combination of things but as said it is hardware, and I'd point it more tword the graphics card end then anything, especially since they do have dedicated data lines going back and forth between the 2 or more cards, it should be able to communicate fast enough for the GPU to anticipate commands, instead of relaying on the CPU. Might be a bios or a driver thing too but there is so much speculation behind this anymore I don't know if anyone knows what to think of it or even if it can be solved. Hope it is solved though. But as said faster seems to eliminate it more often then not, but it can still happen, and probably does its just we don't notice it.

And I perfectly realise much of my post is guesswork or even conjecture when it comes to causes or links, but I believe ATI has a lot of explaining to do and a lot of fixing. The state of 4850 is unacceptable and I'm going to vote with my money as a result. Next card is nVidia unless nVidia screws up badly or ATI improves. I'm not an nVidia fanboy. My 7600GT, which was a great card, died on me, so all in all, nVidia should be seen as causing me a greater loss, but at least the card got RMA'd and I didn't have issues with games, like this not working, that not working (yes, I know gaming companies probably have an nVidia bias since you see, "the way it was meant to be played," everywhere), and I certainly didn't get half a year of grief. I wished I could afford selling the 4850 as defective and buying a different one.

I mean your have alot of issues with the card, might suck but I'd RMA the sucker. I mean if its crashing under stock cooling and im sure it would if you get a nice hot summer day and left it at default. When you get it back, if you have the cash sell it, if not hopefully that new one won't have the same issue your having now. Who knows like you said above, while ATI says one thing on the temps and it might be true for the cards, not everyone is alike and it could of caused damage, and just slap that aftermarket heatsink on her right away.

And speaking of stuttering, are you running CF at all? I just noticed this and think we are talking about 2 different things :) Your stuttering if 1 GPU, is defiantly a horsepower issue, mine is a time delay issue dealing with communication between 2 GPU's. Where arguing about 2 different things if thats the case on that part ;)

And thanks, think this is about the longest civil discussion I've had in a long time on this forums in this length without each other going at either heads :D

[mattgmann]

...on the stuttering issue. I've heard quite a few different explanations about it, and none seem to pan out. The games I've had issues with include Grid, Fallout 3, Crysis, and most notably Far Cry 2. I don't think the stuttering is caused by the same thing in many cases. I've run all of these games with single and cf setups, and different cards. I think the word stuttering is used to encompass too many different performance issues. The stuttering I'm most concerned about is in games where I know i'm getting 60+fps and still have issues with smoothness. Vsync is sometimes the answer...but in games like FC2, nothing works. Crysis, I believe, is poorly coded and that leads to separate issues. I have no explanation for problems in Grid or Fallout 3. All of these issues, for me, add up to a conclusion that there's an overall problem with "smoothness". A lot of little factors all seem to add up to lessen the experience, and I'm tired of it. It's like owning a ferrari, but the clutch sticks, and the knob on the radio is broken, and there's a crack in the windshield and a spring keeps poking me in the left nut.

[deathman20]

Well what im calling stuttering is called "micro stuttering" where you have good FPS but it seems like a hang or pause between frames. It only happens in CF/SLI setups. Single cards do not experience this type of issue.

[NewbieOneKenobi]

Not saying its OK by any means. I don't vid cards should be tempting fate at above 80C at all it should reduce clock, voltages at that temp and 70C would be ideal. While they might be wrong on temps (look at the 4780X2 running at around 90-100C) cards can obviously take the temps, if its above and beyond the reference board design, its not ATI's fault but the manufactures if they deviate from the design.

Yeah, that is true. I worry the power of the modern cards is basically equivalent to an overclock and possibly a taxing one.

Damn fan is loud hehe and I think thats why they kept it that was so idle and gaming its relatively cool but as well keeps close tempatures. While fan editing has been in profiles if you install CCC, you can enable them as well via the panel for a few versions now. Enough complain about it and it happens. But what amazes me more is the manufactures do not change there bios profiles to increase the fan speed from reference. While ATI is at fault originally they have added the option to up it, but also the card manufactures could increase the fan speed via bios and help out alot too.

Hah, so it looks like you see it as a plausible scenario that they kept the fan so slow because otherwise it would be too loud. I thought I was seeing things when I conceived this idea.

Yes I agree companies use tape, and paste, but most commonly tape is found on nearly all Memory sinks on cards, and even on voltage moduals.

True. Mostly just the GPU gets paste. But that's because memory gets separate little heatsinks if at all (e.g. the orb design provides cooling but doesn't use a thermal conductor there because there's no sticking area) and voltage units generally get forgotten about. However, with an all-card heatsink that gets screwed to the card, they could use paste because the heatsink can't fall off if screws hold it. I was shocked when I saw how thick the tape was, but because it's a different kind of tape, like a silicon thing, perhaps it's not as a bad as a thick layer of traditional tape. Maybe I just love putting ACS5/Zalman grease/Whatever under stock heatsinks on stuff.

Definatly paste is better but also depending on what paste they use, paste can wreak the card, and from an assembly line perspective its easier to use tape. There for if a machine screws up on putting paste in the proper places its not going to fry the card, if said paste is conductive of course.

True. But there's also that pre-applied paste that comes with boxed CPU's or some aftermarket graphics coolers. For example, I got some on my Accelero and decided to use it because I tought scrapping it off and applying a different paste wasn't necessarily the best idea. Perhaps I should do it some time, however...

While the stock cooler might be insufficient, it depends. Like I said I have 2 cards, with 2 vastly different in tempatures. Matter of fact I should RMA the one because it does run very warm by my standards and even compared to the other card.

Are they from the same manufacturer or at least look the same (cf. early Asus and early Gigabyte that were identical)? If so, the hotter one is probably RMA material.

Now while I might be the on the lucky end, 10 people that I've personally known got a 4850, out of the 10, 1 had issues with it. While 10% might be small but as well it wasn't heat related as well. Most of them just game and could care less about the system just as long as they have a good experiance playing.

Yeah, I wouldn't care so much if everything ran OK. Maybe I got a bad card because I bought it early. In fact, I bought it two days before the official release, meaning I probably got the earliest version after testing samples.

Ah wish I could do that, CF and aftermarket heatsinks is just a no go :(

:( But can't you take an upward facing heatsink on the first card and a downward-facing heatsink on the second one? I don't know if the cheap Accelero can do this (I think not, I'm just not 100% sure it's impossible), but there are some such coolers. Or you could buy an Accelero for the second card because it's cheap and as good as the expensive stuff, and something more expensive to face upward on the face card - as long as it doesn't clash with the Northbridge or CPU cooler, that is. Or the RAM. I suppose the RAM sticks block any serious attempt at an upward facing heatsink on any stronger card... unless you can find something short enough. I was thinking about using an old CPU cooler that way, actually. Some people have done that.

Agree with you on this to a point, your only catching part of the user base, the other side isn't having issues at all. What should be done to get an idea, is make a poll with 4850's in the forum seperated by card manufacture and see where all the issues are really coming from. It might be just a few manufactures that have the problems or a certain line of there cards. If anything it will narrow the area down to which cards have the true problems and which ones have a higher chance of no overheating issue.

True and the problems are of multiple kinds. TDR errors themselves are caused by a number of different things. What I see as a "stable feature", however, is CCC runtimes and things that result from following the specification - by the way, do the manufacturers even use different factory lines? I thought the cards came from the same factory line, just getting different stickers along the way and maybe different coolers when it comes to it.

As for temp responses its not driver related, 2D to 3D switching is driver related but actual temp and fan speed is controlled via bios. You can see it in dos even where fan speeds will change as the card gets hotter (limiting air flow). Yes you can modify the file, and that will over-ride settings thats in the bios, then you are really relaying on the drivers to handle it. I did that before, and after I found the right software to flash I went that route. No need for me to chance my cards via driver when if all else fails the card itself will protect itself.

I see. Yeah, obviously, those things are coded in the BIOS, but it's drivers that enable change. That's what I meant. But if drivers can change these parameters when you tell them to, they can also be made to react in some more desirable way to passive cooling or non-reference fans.

There is so much on the "Stuttering" issue. At first I didn't believe it, heck didn't even notice it til 1 game, GRID. I could see it in that, barely but it was there. I mean getting 80-140FPS maxed out, figured I wouldn't see it but it is.

Awch. What kind of monitor are you using? If I see above 60 fps, I turn more options on because that's the refresh rate of my HP in its native resolution (and I think I can see some ghosting when the 60 fps is actually used... 5 ns isn't ideal, but at least I have nice colours). Still, I doubt I'd see 140 in a modern game or even 80 with reasonable settings. Some good OC'ing there. :thup: Unfortunately, my issues-ridden baby makes me a bit too cautious to try anything extreme and looking for a job means I can't afford frying anything. :/

Vsync defiantly helps at eliminating it though.

Yeah. That was the first thing I turned off to increase fps when I had a weaker system, but later I noticed it wasn't worth it.

Its a combination of things but as said it is hardware, and I'd point it more tword the graphics card end then anything, especially since they do have dedicated data lines going back and forth between the 2 or more cards, it should be able to communicate fast enough for the GPU to anticipate commands, instead of relaying on the CPU. Might be a bios or a driver thing too but there is so much speculation behind this anymore I don't know if anyone knows what to think of it or even if it can be solved. Hope it is solved though. But as said faster seems to eliminate it more often then not, but it can still happen, and probably does its just we don't notice it.

Yeah. Cranking stuff up isn't a solution per se, but if everything is faster, communication tends to be faster too and commands get executed faster, so the wretched TDR leaves us alone. For this reason, I'll probably have to look at OC'ing my 7200 already, although thus far I was contenting myself with using two fans at 825 rpm each to remove noise. One can barely hear the computer with a case lid removed and that's how I like it. But if I play a game, the music deals away with most evil sounds anyway. :) It even made me forget about the 4850 stock fan running at 100% if the game was good! :D


I mean your have alot of issues with the card, might suck but I'd RMA the sucker. I mean if its crashing under stock cooling and im sure it would if you get a nice hot summer day and left it at default. When you get it back, if you have the cash sell it, if not hopefully that new one won't have the same issue your having now. Who knows like you said above, while ATI says one thing on the temps and it might be true for the cards, not everyone is alike and it could of caused damage, and just slap that aftermarket heatsink on her right away.

I would do that, but I can't now, having replaced the cooler. I should have RMA'd it, but I wouldn't have got a replacement card, so I'd be without a computer, basically, for a month. While it could help work and study, I was too much of a gaming junkie to do it. :o Or truth be said, I worked such long hours at the law firm that I wouldn't probably have noticed much. Anyway, can't do much now.

And speaking of stuttering, are you running CF at all? I just noticed this and think we are talking about 2 different things :) Your stuttering if 1 GPU, is defiantly a horsepower issue, mine is a time delay issue dealing with communication between 2 GPU's. Where arguing about 2 different things if thats the case on that part ;)

Hah! No, single-card for me. I don't believe in CF/SLI because I think single cards pay better in the end. I'd imagine CF stuttering relies much more on communication issues than a single card does, especially as the second card runs off a slower slot. As far as I know, 4850 has a special collection of problems specific to Crossfire. I likely won't experience them, however, because unless I get a strikingly good bargain, I'm more partial to allowing a family member to "inherit" my computer as it is now and buying a new box. Looks better to me than having unused spare parts or selling them off for dirt and especially if you consider OEM license on the OS following the mobo. I just might throw a second 4850 in if I get a used one cheap, though, although 500W PSU isn't particularly inviting.


And thanks, think this is about the longest civil discussion I've had in a long time on this forums in this length without each other going at either heads :D

Civil discussion is the basis of learning, hehe. Thanks as well. :)

[NewbieOneKenobi]

...on the stuttering issue. I've heard quite a few different explanations about it, and none seem to pan out. The games I've had issues with include Grid, Fallout 3, Crysis, and most notably Far Cry 2. I don't think the stuttering is caused by the same thing in many cases. I've run all of these games with single and cf setups, and different cards. I think the word stuttering is used to encompass too many different performance issues. The stuttering I'm most concerned about is in games where I know i'm getting 60+fps and still have issues with smoothness. Vsync is sometimes the answer...but in games like FC2, nothing works. Crysis, I believe, is poorly coded and that leads to separate issues. I have no explanation for problems in Grid or Fallout 3. All of these issues, for me, add up to a conclusion that there's an overall problem with "smoothness". A lot of little factors all seem to add up to lessen the experience, and I'm tired of it. It's like owning a ferrari, but the clutch sticks, and the knob on the radio is broken, and there's a crack in the windshield and a spring keeps poking me in the left nut.

Fallout 3 has issues with anti-aliasing, at least on ATI cards. Turning it off reduced the number of errors for me but didn't eliminate them and I thought the reduced visual experience (awful vs "pretty", i.e. as "pretty" as post-atomic wasteland can get) wasn't worth it. I turned it back on, although modestly (probably 2X or maybe 4X) and played for a while. Maybe the 8.10 release was good or whichever it was I used before 8.12. I had a large gap because I had a lot of private litigation writing to do in November and December. ATI claims 8.12 brings a 25% fps increase in Fallout 3 specifically. Let's finally see.

As for other causes, I generally associate stuttering with the HDD being busy and rather with something else than the game being played. I sometimes think the system is doing something and while I suspect also the anti-vir software, it doesn't prove to be the case. Also, as long as the DVD is being used, I would look into that one because we gamers tend to be an unhygienic lot and it's not like your typical drive has perfect accuracy, although things are better than they used to be, especially as pretty much every drive is a burner and burners have superior laser and read bad tracks better. I should probably make a HDD image of NWN2 or Fallout 3. If you play a CD game and have 4 GB RAM or more, or even a DVD game if you have 8 GB RAM or more, you could probably look into a RAM-drive like in DOS times.

EDIT: BTW, does anyone know if uninstalling CCC vs merely shutting it down makes any difference in terms of stability/compatibility?

[deathman20]

True. Mostly just the GPU gets paste. But that's because memory gets separate little heatsinks if at all (e.g. the orb design provides cooling but doesn't use a thermal conductor there because there's no sticking area) and voltage units generally get forgotten about. However, with an all-card heatsink that gets screwed to the card, they could use paste because the heatsink can't fall off if screws hold it. I was shocked when I saw how thick the tape was, but because it's a different kind of tape, like a silicon thing, perhaps it's not as a bad as a thick layer of traditional tape. Maybe I just love putting ACS5/Zalman grease/Whatever under stock heatsinks on stuff.

True. But there's also that pre-applied paste that comes with boxed CPU's or some aftermarket graphics coolers. For example, I got some on my Accelero and decided to use it because I tought scrapping it off and applying a different paste wasn't necessarily the best idea. Perhaps I should do it some time, however...

Aye all that paste is general crap. Its some of its better than the CPU paste they have on there heatsinks but some is exactly the same. It varies of course. The tape.... yum hehe, think they do that so they don't have to be nearly as accurate on how close its placed near the ram chips. Chips are different thicknesses? Im not sure but if it is, it might affect the heatsink design (full covered one) that it would be logical to keep it as is with tape. As well its not like the memory needs much cooling on it typically, even ram sinks on them are overkill unless you really OC the crap out of the memory in most cases. Course that also depends how hot the GPU gets as well since heat transfers on the board and affects memory tempature where heatsinks are nessisary.


Are they from the same manufacturer or at least look the same (cf. early Asus and early Gigabyte that were identical)? If so, the hotter one is probably RMA material.

Yeah, I wouldn't care so much if everything ran OK. Maybe I got a bad card because I bought it early. In fact, I bought it two days before the official release, meaning I probably got the earliest version after testing samples.

Powercolor identical (swapped both cards heatsinks to test it), same batch, same everything ordered at the same time, on or right around release day. So they are first batch cards as well. I should RMA it, but then again it runs fine and thats what they'd say and send it back, because no way im sending both mine in. Gaming nut as well so no GPU's = sad gamer. Mind you I do have a X-360 but still


:( But can't you take an upward facing heatsink on the first card and a downward-facing heatsink on the second one? I don't know if the cheap Accelero can do this (I think not, I'm just not 100% sure it's impossible), but there are some such coolers. Or you could buy an Accelero for the second card because it's cheap and as good as the expensive stuff, and something more expensive to face upward on the face card - as long as it doesn't clash with the Northbridge or CPU cooler, that is. Or the RAM. I suppose the RAM sticks block any serious attempt at an upward facing heatsink on any stronger card... unless you can find something short enough. I was thinking about using an old CPU cooler that way, actually. Some people have done that.

No room for that, and if I can fit one in, while it being close I can fit the new Thermalright low profile heatsink in there. Im a thermalright guy so if I go I'll do that :) Downside is that if I put the heatsinsk in there I loose my X-FI sound card which im getting use to again. And even then im not sure I could use the on board audio as well since the cable might be too short to put where there will be room for a free slot on the back. I've been contemplating long and hard about new heatsinks but keep running ito the same issue. Oh as well the ram hits the top slots card basically, so the heatsink would have to clear that, clear the nb, and let lone clear the massive CPU heatsink. Might need a board with 3 PCI-E x16 to get the job done. x58 :) just gotta wait though.

Awch. What kind of monitor are you using? If I see above 60 fps, I turn more options on because that's the refresh rate of my HP in its native resolution (and I think I can see some ghosting when the 60 fps is actually used... 5 ns isn't ideal, but at least I have nice colours). Still, I doubt I'd see 140 in a modern game or even 80 with reasonable settings. Some good OC'ing there. :thup: Unfortunately, my issues-ridden baby makes me a bit too cautious to try anything extreme and looking for a job means I can't afford frying anything. :/

I got a Dell 2405FPW, while I don't notice ghosting often on it I know its there, if not mistaken its a 12ms panel. Its not that type of stuttering it deals with the CF card communication.

Oh I know about the baby thing, just had a little one, but thankfully we haven't had issues. Just trying to keep the budget good in times like this thats all Im looking for. GL with your little one and hope all is well.

Yeah. Cranking stuff up isn't a solution per se, but if everything is faster, communication tends to be faster too and commands get executed faster, so the wretched TDR leaves us alone. For this reason, I'll probably have to look at OC'ing my 7200 already, although thus far I was contenting myself with using two fans at 825 rpm each to remove noise. One can barely hear the computer with a case lid removed and that's how I like it. But if I play a game, the music deals away with most evil sounds anyway. :) It even made me forget about the 4850 stock fan running at 100% if the game was good! :D

Nice and silent, I was that way once, they I added fans, more fans, cut holes in the case and added more fans :) Its a cool beast but I got fans out the ying yang (8x 120mm fans, 1x 80mm fan).



Hah! No, single-card for me. I don't believe in CF/SLI because I think single cards pay better in the end. I'd imagine CF stuttering relies much more on communication issues than a single card does, especially as the second card runs off a slower slot. As far as I know, 4850 has a special collection of problems specific to Crossfire. I likely won't experience them, however, because unless I get a strikingly good bargain, I'm more partial to allowing a family member to "inherit" my computer as it is now and buying a new box. Looks better to me than having unused spare parts or selling them off for dirt and especially if you consider OEM license on the OS following the mobo. I just might throw a second 4850 in if I get a used one cheap, though, although 500W PSU isn't particularly inviting.

I was same way as well originally for single cards but since this was the same price as getting a 4870 at the time I went with it. Definatly don't regret it, fun playing with 2 GPU's. :) Course they have to be more powerful than the $300 range for me but its all good, if I get a new card going single again I'd might go but nto sure.


Civil discussion is the basis of learning, hehe. Thanks as well. :)

Yeah but not everyone believes in that.


Wheres my 9.1's :p

[NewbieOneKenobi]

Aye all that paste is general crap. Its some of its better than the CPU paste they have on there heatsinks but some is exactly the same. It varies of course. The tape.... yum hehe, think they do that so they don't have to be nearly as accurate on how close its placed near the ram chips. Chips are different thicknesses? Im not sure but if it is, it might affect the heatsink design (full covered one) that it would be logical to keep it as is with tape. As well its not like the memory needs much cooling on it typically, even ram sinks on them are overkill unless you really OC the crap out of the memory in most cases. Course that also depends how hot the GPU gets as well since heat transfers on the board and affects memory tempature where heatsinks are nessisary.

I so don't feel like removing Accelero, but I might do it because I'll probably be placing the fan at a different angle (i.e. use a shorter cord here and a longer one there :p), so I'll need to take the card out anyway.

Powercolor identical (swapped both cards heatsinks to test it), same batch, same everything ordered at the same time, on or right around release day. So they are first batch cards as well. I should RMA it, but then again it runs fine and thats what they'd say and send it back, because no way im sending both mine in. Gaming nut as well so no GPU's = sad gamer. Mind you I do have a X-360 but still

Ouch. Well, I'd RMA it if it caused any errors, but I'm not sure they'd replace it just because it's hotter as long as it runs. Might be worth it, but I'm not sure.

No room for that, and if I can fit one in, while it being close I can fit the new Thermalright low profile heatsink in there. Im a thermalright guy so if I go I'll do that :) Downside is that if I put the heatsinsk in there I loose my X-FI sound card which im getting use to again. And even then im not sure I could use the on board audio as well since the cable might be too short to put where there will be room for a free slot on the back. I've been contemplating long and hard about new heatsinks but keep running ito the same issue. Oh as well the ram hits the top slots card basically, so the heatsink would have to clear that, clear the nb, and let lone clear the massive CPU heatsink. Might need a board with 3 PCI-E x16 to get the job done. x58 :) just gotta wait though.

Tough luck. Well, perhaps two heatsinks that face each other and a fan strapped between them in some criminal way to blow the air outside through the holes after removing PCI lids? But if your mobo and case are anything like mine, it won't work. If I wanted Crossfire now, I couldn't do it for anything in the world. The 4850 is the only slot card and it takes all.

I got a Dell 2405FPW, while I don't notice ghosting often on it I know its there, if not mistaken its a 12ms panel. Its not that type of stuttering it deals with the CF card communication.

From what I've heard 15 ms enables normal play.

Oh I know about the baby thing, just had a little one, but thankfully we haven't had issues. Just trying to keep the budget good in times like this thats all Im looking for. GL with your little one and hope all is well.

Awch, I'm sorry! I was referring to my computer, actually. This probably means I should go out more. :o

Good luck with yours, though!

4850 is not the only modern component I dislike. I bought an Asus P5Q-E too and it took the last BIOS update, the Nth version, for it to start remembering the CMOS settings and stop giving me Press F2's. I spent nearly all my salary on it that month and the job terminated a couple months later. Young lawyers have an awful situation getting a job here because of all the crisis.

Nice and silent, I was that way once, they I added fans, more fans, cut holes in the case and added more fans :) Its a cool beast but I got fans out the ying yang (8x 120mm fans, 1x 80mm fan).

My previous rig at some moment worked off a CPU fan, GPU fan and PSU fan. I used rubber pads and I even put some pads on the PSU to stabilise it against the walls of the case. It was well enough. :) Now I have two 92's on the CPU, the PSU has 2 80's and the GPU has one hanging (on cords, mind you) down the Accelero, a 120. I'm not using a system exhaust fan because it would be so close to the outward CPU fan it wouldn't make much sense. I often read or write on this computer and I'm not willing to listen to fans during that, as a price for not even so great a benefit in quickening the dissipation.

I was same way as well originally for single cards but since this was the same price as getting a 4870 at the time I went with it. Definatly don't regret it, fun playing with 2 GPU's. :) Course they have to be more powerful than the $300 range for me but its all good, if I get a new card going single again I'd might go but nto sure.

Yeah, price is the only factor to make it worthwile, I think, although the experience is surely great. There are so many things I like better in a single-card config.

Yeah but not everyone believes in that.

Sometimes it takes a fighter to know the value of peace. ;)

Wheres my 9.1's :p

Already online but apparently lacking the tweaks from the 8.12 multicore hotfix. I'm skipping it for now.

[gian84]

Incase some of you don't know 9.1 is out! :D

http://downloads.guru3d.com/ATI-Catalyst-9.1-Vista-32-bit-download-2172.html

I hope this driver kicks as5 :p

[NewbieOneKenobi]

I know it's out, but I fear to download it because of the reported issues and the fact the multicore fix didn't make it. :p

[gian84]

well i've installed it and run 3dmark06,vantage,crysis benchmark. No improvement in fps/performance... but at least no loss either :) works fine with my setup, no issues so far. I hope Ati comes out with the hotfixes fast to address those reported issues you said because I don't want to run into them :p