ATI Deliberately Retards Catalyst for FurMark

[btarunr]

It is a known flaw that some models of the Radeon HD 4800 accelerators fail oZone3D FurMark, an OpenGL based graphics benchmark application that has found to stress Radeon HD 4800 series far enough to result in over-heating, artifacts or even driver crashes. The Catalyst 8.8 drivers have found to treat the FurMark executable differently based on its file-name. Expreview tested this hypothesis by benchmarking a reference design HD 4850 board using Catalyst 8.8 driver, with two runs of FurMark. In the first run, the test was cleared at a low score, much lower compared to those of whatever successful runs on older drivers could churn out. Suspecting that the driver could be using some sort of internal profile specific to the FurMark executable, Expreview renamed the furmark.exe file, thereby not letting the driver know it's FurMark that's being run. Voila! the margin of lead the renamed FurMark executable gave over "furmark.exe" shows the driver to behave differently. A shady thing since Radeon HD 4800 almost became infamous for failing at FurMark, and at least passing it with a low score seemed better than failing at it altogether.

Expreview caught this flaw when testing the PowerColor Radeon HD 4870 Professional Cooling System (PCS+) when odd behaviour with the newer driver was noted. Successive BIOS releases didn't fix the issue, in fact, it only got worse with erratic fan behaviour caused due to a "quick-fix" BIOS PowerColor issued (covered here).




View at TechPowerUp Main Site

 

[wolf2009]

hmmm ..... trying to save the poor components used in the VRM from blowing .

 

[newtekie1]

Not a good thing, they should work more on fixing the cards so they can actually handle heavy load instead of grippling them with drivers.

 

[jbunch07]

Im trying to make sense of this, ATi instructed the card to perform poorly on furmark so it wouldn't stress out and overheat?
This is odd to say the least.

 

[jydie]

Actually, all they need to do if force companies like Sapphire, Vision Tek, etc. to focus more on keeping the GPU at a safe temperature. Instead, they try to keep the fan running as slow and quiet as possible. These are high end cards... bring on the noise!! How cool would a motorcycle or sports car be if they were always silent?? Sure would not turn many heads now would they?

 

[alexp999]

So if I'm getting this right, the latest cat drivers cripple the performance of furmark cus the 48xx series cant cope with sustained 100% load even at stock?

I'm so glad I didnt get a 4870...

Are these same problems evident with the 4870X2?

And is this a heat or crappy component issue?

 

[mikek75]

AH HA!!! I thought I was going mad....On 8.7 I sucessfully completed Furmark with scores in excess of 7000, with 8.8 it dropped to 3800 or so. Bastards, I thought my card was going kaputt!

 

[Bull Dog]

Well frankly I wouldn't be too concerned about it. My 4870 handled furmark just fine with the stock cooler and stock bios.


And seriously we all know that the fur causes the GPU to generate more heat that anything else. So if the card can still run every game out there with 100% reliability and there is some silly, relatively obscure benchmark that some cards seem to be having problems with, who cares?

 

[newtekie1]

Well frankly I wouldn't be too concerned about it. My 4870 handled furmark just fine with the stock cooler and stock bios.


And seriously we all know that the fur causes the GPU to generate more heat that anything else. So if the card can still run every game out there with 100% reliability and there is some silly, relatively obscure benchmark that some cards seem to be having problems with, who cares?

This issue is that, as newer games come out that stress the GPU more, they will be pushed to their failure point just like Furmark does. I would certainly care if my GPU was unable to handle a few minutes of max load, there is bound to be a game that will eventually do the same.

I believe this is mainly an issue with the single slot coolers on the HD4850, not the dual slot ones used on the HD4870.

 

[SpookyWillow]

im not sure that its just this though, my soft shadow scores have dropped considerably too and that happened on my 3870 when they introduced the new OGL AA in the 8.6 drivers.

OpenGL Adaptive Anti-Aliasing Custom Filters support
This release of Catalyst™ introduces Adaptive Anti-Aliasing Custom Filters support for OpenGL applications for the ATI Radeon™ HD 3000 Series, ATI Radeon™ HD 2000 Series of products. Using the Edge Detect Custom Filter, users can enable 12X and 24X Anti-Aliasing. Selecting 4X Anti-Aliasing plus selecting the Edge Detect filter delivers the equivalent of 12X Anti-Aliasing. Selecting 8X Anti-Aliasing plus selecting the Edge Detect filter delivers the equivalent of 24X Anti-Aliasing.

 

[acperience7]

Actually, all they need to do if force companies like Sapphire, Vision Tek, etc. to focus more on keeping the GPU at a safe temperature. Instead, they try to keep the fan running as slow and quiet as possible. These are high end cards... bring on the noise!! How cool would a motorcycle or sports car be if they were always silent?? Sure would not turn many heads now would they?

I agree with you on the noise issue. These are not the days of 9800Pro's or Pentium 3 CPU's anymore, and air cooling tech can only go so far in the noise vs heat battle. On high end cards like the 4870 and GTX 280 people should expect them to make noise with the kind of power they have. Then everyone complains about the weight of these cards. Considering the abilities that they have and the heat they put out, the high levels of noise and the weight and size of these cards make sense. Maybe it's time for OEM graphics and CPU companies to start developing water cooling solutions for these cards for people that do want them quiet. I'm all for air cooling, but you can only go so far air. Technology is getting to a point where it's time to start integrating water cooling solutions as a mainstream method of cooling. As far as I'm concerned; Crank the fans, keep em cool, keep em alive longer.

 

[R_1]

It is easy for ATI partners to made a silent but hot card, then silent and cool card. The hot one will degrade and eventually get toasted, but it will live long enough to become obsolete in games.

 

[DarkMatter]

This only confirms something that I already suspected and that was almost confirmed already by the temperatures of all R7xx cards and the low OC capabilities. Ati clocked this cards very high in order to own this round and also used very cheap cooling to make the cards cheaper, for the same reason.

It is easy for ATI partners to made a silent but hot card, then silent and cool card. The hot one will degrade and eventually get toasted, but it will live long enough to become obsolete in games.

That's something hard to tell, but as of now it seems that the failure rate is not very high (=excesive to the point to annoy consumers), even though it's definately high above the average.

 

[AsRock]

Im trying to make sense of this, ATi instructed the card to perform poorly on furmark so it wouldn't stress out and overheat?
This is odd to say the least.

Maybe there worried about people bitching about fan noise( stop your bitching about fan noise), if there's one thing i am sick off is people wanting very fast video cards and expect them not to have fan noise.

If some thing works hard it's going produce a load of heat until we make some major break though which is not going be any time soon due to how fast things are moving.

Hence the reason i got the PSU i got as silent has bigger disadvantages IMO. All though never heard the PSU fan still lol.

If any thing they should make these cards easier for people adjust so they can make they work to there fullest

 

[btarunr]

Just to clarify, the cards aren't "failing" (as in going kaput) as such, it's just this benchmark that seems to be stressing the RV770, it's not clearing it. On the other hand, the card isn't "failing" in Crysis bench / game so it's not like the card is a failure when it comes to handling stress. Sources tell it could be because FurMark stresses the memory bus a lot, asserted by the MEM_IO temperatures.

 

[alexp999]

Just to clarify, the cards aren't "failing" (as in going kaput) as such, it's just this benchmark that seems to be stressing the RV770, it's not clearing it. On the other hand, the card isn't "failing" in Crysis bench / game so it's not like the card is a failure when it comes to handling stress. Sources tell it could be because FurMark stresses the memory bus a lot, asserted by the MEM_IO temperatures.

But on stock settings, it should never fail 100% stress/load. Where is the built in safety overhead which people usually exploit for OCing if it cant even sustain what it is theoretically capable of.

 

[DarkMatter]

Just to clarify, the cards aren't "failing" (as in going kaput) as such, it's just this benchmark that seems to be stressing the RV770, it's not clearing it. On the other hand, the card isn't "failing" in Crysis bench / game so it's not like the card is a failure when it comes to handling stress. Sources tell it could be because FurMark stresses the memory bus a lot, asserted by the MEM_IO temperatures.

But they do have some erratic performance when they overheat on some games. Just google HD4870 overheating issues and see. I have seen many threads about that here on TPU too.

As I said it's not something that hasn't happened in the past. It's very common in the industry, and something the average Joe shouldn't care about, but it does affect many people and it's good to know that it could happen. What it is safe on some regions, on others will make the card get fried up in less than 2 mins. Radeons are around 90C when ambient temps are 25-30C, but what about when the ambient temps are above 40C?

 

[alexp999]

This is getting more and more like the RROD scenario on the 360's. I just hope it doesnt take AMD 3 years to fix it.

 

[DarkMatter]

This is getting more and more like the RROD scenario on the 360's. I just hope it doesnt take AMD 3 years to fix it.

It's not that dramatic.

 

[EastCoasthandle]

Edit: The results are not the same...

 

[btarunr]

That's Temp1 (that CCC and Everest use to denote "GPU Temperature").

 

[R_1]

The hottest parts in 4800 series are GPU and power mosfets. When mosfets are heated up their on-state resistance rises with temperature, if the load is approximately a constant-current load then the power loss rises correspondingly, generating further heat. The effect is known as "Thermal runaway". It can result in destruction of the device.
GPU overheating can reduce it's life-span drastically to 3 years and few months.

 

[btarunr]

Is this post directed to me? If so, I have no idea what you are talking about.

Unless that's 100% fan throughout, ~40 C looks unreal.

CCC and Everest don't display DISP_IO, MEM_IO or SHADER_core values, the "GPU Temperature" value is somehow derived out of those three values.

 

[mdm-adph]

Are we sure this isn't just for the PowerColor 4870 PCS+ card? I've heard nothing but trouble with that thing and it's terrible BIOS.

 

[EastCoasthandle]

Unless that's 100% fan throughout, ~40 C looks unreal.

CCC and Everest don't display DISP_IO, MEM_IO or SHADER_core values, the "GPU Temperature" value is somehow derived out of those three values.

In this case they are not necessary to show that the amperage between the 2 are different. The fan speed between the 2 are within margin.