SweClockers: Geforce GTX 590 burns @ 772MHz & 1.025V

[Seero]

Fried cards are hardly fear mongering. And Nvidia is now preventing any voltage adjustments for a reason don't you think?

To prevent 3 year old kids or dumb people who doesn't know how to, but OC to fry their new cards may be?

 

[HyperMatrix]

Fried cards are hardly fear mongering. And Nvidia is now preventing any voltage adjustments for a reason don't you think?

Well I have it set to create an alarm event if the gpu temp ever hits 85c, which hasn't happened after a few hours of hardcore gaming.

I don't think pushing 1v is all that bad. The main problem is that people don't take cooling into consideration. As some reviews I read stated, the automatic fan control only bumps up another 300rpm when it's burning hot. Asus's board recommends anything past 675-ish to be set to a manual fan speed of 60%. I'm running it at 91%. Another issue might be that the software, on reboot, has occassionally forgotten to increase the fan speed and it's dropped back down to 40% manual. I noticed this thanks to evga's "evga precision tuner" software which outputs all the details to my G19's keyboard regardless of the program I'm in.

I'm probably going to take heat for running my dimm's at 1.7v too. At the end of the day, as long as you're not going overboard, it all comes down to heat. With my CPU, for example...OC'ing from 3.4ghz to 4ghz and running at 85c is worse than OC'ing from 3.4ghz to 5.2ghz and running at 75c.

 

[notty22]

To prevent 3 year old kids or dumb people who doesn't know how to, but OC to fry their new cards may be?

That and probably the same reasons, AMD puts a bright yellow sticker on the 6990 that tells you to check AMD.com/warranty before you remove it
You have to put a little knowledge/effort in to the o/c.

 

[Seero]

Well I have it set to create an alarm event if the gpu temp ever hits 85c, which hasn't happened after a few hours of hardcore gaming.

I don't think pushing 1v is all that bad. The main problem is that people don't take cooling into consideration. As some reviews I read stated, the automatic fan control only bumps up another 300rpm when it's burning hot. Asus's board recommends anything past 675-ish to be set to a manual fan speed of 60%. I'm running it at 91%. Another issue might be that the software, on reboot, has occassionally forgotten to increase the fan speed and it's dropped back down to 40% manual. I noticed this thanks to evga's "evga precision tuner" software which outputs all the details to my G19's keyboard regardless of the program I'm in.

I'm probably going to take heat for running my dimm's at 1.7v too. At the end of the day, as long as you're not going overboard, it all comes down to heat. With my CPU, for example...OC'ing from 3.4ghz to 4ghz and running at 85c is worse than OC'ing from 3.4ghz to 5.2ghz and running at 75c.

Sometimes it isn't just heat (well, it is but you really can't cool it in time as it spikes out too quickly and too suddenly), but voltage and blah blah blah.. you probably know all those.

 

[Seero]

That and probably the same reasons, AMD puts a bright yellow sticker on the 6990 that tells you to check AMD.com/warranty before you remove it
You have to put a little knowledge/effort in to the o/c.

Yup. I don't see anything wrong with that.

 

[AnandThenMan]

To prevent 3 year old kids or dumb people who doesn't know how to, but OC to fry their new cards may be?

So reviewers are 3 year old kids now? And what changed? I don't see Nvidia attempting to prevent over volt and overclocking scenarios on other cards. So Nvidia caved into the "3 year old" overclockers only for the GTX590, interesting.

Seems there is no limit to the excuses some people can come up with.

 

[HyperMatrix]

Out of curiosity, what time are you guys getting when running this:

http://ie.microsoft.com/testdrive/Performance/Preschool/Default.html

Seems off-topic but as it uses hardware acceleration in ie9, it's a nice little benchmark, possibly.

 

[Seero]

So reviewers are 3 year old kids now? And what changed? I don't see Nvidia attempting to prevent over volt and overclocking scenarios on other cards. So Nvidia caved into the "3 year old" overclockers only for the GTX590, interesting.

Seems there is no limit to the excuses some people can come up with.

Hey not everyone can overclock you know? I know my skills, and the amount of work needed to put in to see 10% increase in FPS. That is what? 3 FPS?

Only a 3 years old will CRY OUT LOUD when they break something by doing something wrong.

Okay, kids does that, not only 3 years old.

 

[HyperMatrix]

So reviewers are 3 year old kids now? And what changed? I don't see Nvidia attempting to prevent over volt and overclocking scenarios on other cards. So Nvidia caved into the "3 year old" overclockers only for the GTX590, interesting.

Seems there is no limit to the excuses some people can come up with.

Honestly some of them may as well be. I was reading a review of some site who said their card at 700mhz was running 94c so they wouldn't recommend going any higher. Lol. Honestly...there's no excuse for that. -_-

 

[AnandThenMan]

Sure. Would a gpu-z screenshot be good enough?

Can you do a 3DMark run, both at stock and overclocked?

 

[96Firebird]

But Doesn't Powercolor, XFX and Sapphire does offer warranty if the switch is flipped? It isn't like the card isn't able to run at that speed, otherwise AMD would had disabled overclocking or placing an AUSUM switch right? Just like nVidia did with the GTX 590.

XFX is the only company to release a statement on the issue.

 

[thilanliyan]

That is my question on the 6990. How can there be a cold spot colder than the intake of the video card?

Dunno why my img links keep disappearing, I did referenced them...

Do you mind linking those again. I want to see exactly what you are saying.

EDIT:
Okay here are some more things to note from the Hardware.fr article:

6990-->"The Radeon HD 6990 has two different bios, one with a limit of consumption to 375W and the other with a limit of 450W."
http://www.hardware.fr/articles/747/page22.html
GTX590-->"With an extreme consumption, announced at 365W in standard use, but at 450W into very heavy loads, it is important to observe what may be the impact on other system components."
http://www.hardware.fr/articles/747/page39.html

So the real power consumption in both cases is not known for both cards in this article. They state "375w" and "450w" for the 6990 but that is the TDP as reported by AMD correct? And then they mention that the GTX590 can also consume "450w" at high loads. Too bad xbitlabs didn't do their usual power consumption tests.

So 3 things IMO:

1)The real power consumption is unknown so the 590 could in fact be consuming more than the 6990 (even with the AUSUM switch flipped), hence producing the higher temperatures in the HDD bay. And a small point but there appears to be more cable clutter in the 590 case in the HDD bay area, which could affect airflow, but I imagine this effect would be small.

2)The higher fan speeds of the 6990, coupled with the type of fan it is, could be pushing a lot of air out of the case hence it is also cooler UNDERNEATH the graphics card (which it appears to be in that case).

3)If you look at the thermal images, the components above the 6990 are warmer. This could be heat pulled into the large aluminum backplate and radiated upwards. The 590 does not have such a large backplate to pull heat into quickly and dissipate upwards. That means more of the heat could be expelled through the heatsinks rather than radiated through other parts. This can also explain some of the extra heat in the HDD bay.

Most importantly:
"The system is closed and we go directly to support measures. To do this we run the test 3DMark 2006 Pixel Shader and Prime for the CPU load on a consistent basis because we are comparing graphics cards here at very different levels of performance and therefore the CPU bound potentially very different. It is indeed more CPU to display 500 fps to display 50. After 30 minutes we briefly interrupt the test of 3DMark to take measures for which we do not log and then we relaçons early. 15 minutes later we open the box to quickly take the thermal image. We note then the maximum temperatures recorded before opening the case in the statements."
http://www.hardware.fr/articles/747/page2.html

The thermal images were taken after OPENING THE CASE. The only way to do it but I think it renders those thermal images not completely accurate.

 

[dust]

Honestly some of them may as well be. I was reading a review of some site who said their card at 700mhz was running 94c so they wouldn't recommend going any higher. Lol. Honestly...there's no excuse for that. -_-

Funny how some guys point at the reviewers being crybabies, when that's exactly what you guys have been doing here.

IIRC the 480 reaches that temp in some cases and I haven't heard anything about dieing 480's.

I wish you guys would stop trying so hard to prove there's nothing wrong with the 590's. The reviewers found issues, some users found issues, Nvidia itself admitted there "might" be an issue (driver/ users/ launched on the wrong planet, take your pick) and yet you act like there's "nothing to see here". Just relax and wait for the issues to be sorted out and the 590 will finally be the card it was meant to be. It's not like the 6990 is perfect either!

 

[HyperMatrix]

Can you do a 3DMark run, both at stock and overclocked?

Good call, Anand. The new "safe" driver's totally buther OC'ing. I did some tests and this is what I came back with on gpu points with 3dmark vantage. It's a dx10 one. Not sure if there's a newer one. But here are the results:

1) Stock = 19909
2) Stock +10% (no voltage upper) = 21602 (8.5% increase)
3) 0.938v, 703mhz clock, 3728mhz mem = 22438 (12.7% increase)
4) 0.95v, 738mhz clock, 3946mhz mem = 22935 (15.2% increase)

I stopped at that point and decided that with these current drivers, that's as much as I'm going to get from the card while maintaining stability). Going higher than that, even while technically it could be stable, was throttled. As a side note...it only throttles 3d performance, not 2d. So if I ran a test with both 2D and 3D components, the 2d ones would gain the proper amount increase from the OC'ing, and 3d tests actually would drop "below" the original stock values.

Honestly...for most people, I'd recommend not touching the voltage for now and just going with about a 673mhz clock. You'll get a decend increase without any of the additional problems that currently exist.

The question now, is whether running 4) 0.95v, 738mhz clock, 3946mhz mem = 22935 (15.2% increase) is slower or faster than running the new beta 270 drivers at stock +10% levels. Experiment for tomorrow.

 

[HyperMatrix]

Couldn't wait. Updated to the new beta drivers. Did the standard +10% OC on it. And got 22455 points. Which means the new drivers, with +10% OC, provide a 4% increase over the driver's Asus gives you which allows for a voltage increase. Basically, upgrading to the new drivers is like a free +25mhz clock increase without increasing the voltage.

Recommendation: Do that.

 

[Keysplayr]

Couldn't wait. Updated to the new beta drivers. Did the standard +10% OC on it. And got 22455 points. Which means the new drivers, with +10% OC, provide a 4% increase over the driver's Asus gives you which allows for a voltage increase. Basically, upgrading to the new drivers is like a free +25mhz clock increase without increasing the voltage.

Recommendation: Do that.

Hmmm.. Well, my card has no warranty, so now you've gotten me curious. I'll see what clocks I can reach without any voltage adjustments (can't anyway).
I don't have quite as fast of a system as you though, Hyper. Very nice GHz on the 2600k.

 

[Seero]

Do you mind linking those again. I want to see exactly what you are saying.

I am under an impression that those image are removed by mods as linking images directly from another site may have violated rules of the forum. Yet, go back to hardware.fr and dig those image yourself is the same thing.

Edit: I fixed one of them, the others magically appears again.

So the real power consumption in both cases is not known for both cards in this article. They state "375w" and "450w" for the 6990 but that is the TDP as reported by AMD correct? And then they mention that the GTX590 can also consume "450w" at high loads. Too bad xbitlabs didn't do their usual power consumption tests.

I know that and my focus isn't about the amount of heat released, but where does the cold spot formed. If there exists extra cooling, than those numbers don't matter. I am trying to prove that there is extra cooling on 6990 when the IR is taken, but not the 590. If that can be proven, then the numbers acquired through those IR graphs can not be used as comparison between the 2 cards.

So 3 things IMO:

1)The real power consumption is unknown so the 590 could in fact be consuming more than the 6990 (even with the AUSUM switch flipped), hence producing the higher temperatures in the HDD bay. And a small point but there appears to be more cable clutter in the 590 case in the HDD bay area, which could affect airflow, but I imagine this effect would be small.

Again, my focus is heat transfer, not the amount of heat released. Theoretically speaking, the amount of heat released doesn't change the distribution or the way heat is being radiated or transferred. I'm aware that the dimension of the card differs and that may impact distribution, but that can't be used to explain the formation of the cold spot. Some argued that the formation of the cold spot is due to no heat gets transferred to that particular spot, but how can this be possible where there is a vent blowing hot air towards it? When the cables got heat up, heat transfer on the cable towards any colder spot it travels through, the IR image of 6970CF shows that, but 6990 shows that the cables that passes through the cold spot serve as a cooler medium thoughout the case. Not take it from me, see for yourself.

2)The higher fan speeds of the 6990, coupled with the type of fan it is, could be pushing a lot of air out of the case hence it is also cooler UNDERNEATH the graphics card (which it appears to be in that case).

Compare the IR graph of 6990 and 6970CF, it is clear that 6970CF doesn't release huge about of hot air into the HDD bay.

It isn't about the performance between HS, it is the existence of the cold spot. First you have a vent at the front, and an intake and the bottom. Doesn't it make sense that there will be a clockwise cyclone forming at the bottom left of the card? If so, why will there be a spot at the path of the cyclone that is far cooler than its immediate environment?

3)If you look at the thermal images, the components above the 6990 are warmer. This could be heat pulled into the large aluminum backplate and radiated upwards. The 590 does not have such a large backplate to pull heat into quickly and dissipate upwards. That means more of the heat could be expelled through the heatsinks rather than radiated through other parts. This can also explain some of the extra heat in the HDD bay.

Heat that are radiated cannot travel towards the HDD bay because there are obstacles blocking its path. Radiation can't bend and will only travel at a straight line. The thermal reading of the HDD bay can't be a result of thermal radiation. Look at the bottom of the case in the 6970CF's IR graph, look at the heat distribution and compare it with the heat distribution at the HDD bay of the 6990 IR graph.

Most importantly:
"The system is closed and we go directly to support measures. To do this we run the test 3DMark 2006 Pixel Shader and Prime for the CPU load on a consistent basis because we are comparing graphics cards here at very different levels of performance and therefore the CPU bound potentially very different. It is indeed more CPU to display 500 fps to display 50. After 30 minutes we briefly interrupt the test of 3DMark to take measures for which we do not log and then we relaçons early. 15 minutes later we open the box to quickly take the thermal image. We note then the maximum temperatures recorded before opening the case in the statements."
http://www.hardware.fr/articles/747/page2.html

The thermal images were taken after OPENING THE CASE. The only way to do it but I think it renders those thermal images not completely accurate.

I missed the dischaimer. That doesn't change any of my claims though.

 

[Keysplayr]

Couldn't wait. Updated to the new beta drivers. Did the standard +10% OC on it. And got 22455 points. Which means the new drivers, with +10% OC, provide a 4% increase over the driver's Asus gives you which allows for a voltage increase. Basically, upgrading to the new drivers is like a free +25mhz clock increase without increasing the voltage.

Recommendation: Do that.

What settings are you running in Vantage? Res?

 

[SirPauly]

Good call, Anand. The new "safe" driver's totally buther OC'ing. I did some tests and this is what I came back with on gpu points with 3dmark vantage. It's a dx10 one. Not sure if there's a newer one. But here are the results:

1) Stock = 19909
2) Stock +10% (no voltage upper) = 21602 (8.5% increase)
3) 0.938v, 703mhz clock, 3728mhz mem = 22438 (12.7% increase)
4) 0.95v, 738mhz clock, 3946mhz mem = 22935 (15.2% increase)

I stopped at that point and decided that with these current drivers, that's as much as I'm going to get from the card while maintaining stability). Going higher than that, even while technically it could be stable, was throttled. As a side note...it only throttles 3d performance, not 2d. So if I ran a test with both 2D and 3D components, the 2d ones would gain the proper amount increase from the OC'ing, and 3d tests actually would drop "below" the original stock values.

Honestly...for most people, I'd recommend not touching the voltage for now and just going with about a 673mhz clock. You'll get a decend increase without any of the additional problems that currently exist.

The question now, is whether running 4) 0.95v, 738mhz clock, 3946mhz mem = 22935 (15.2% increase) is slower or faster than running the new beta 270 drivers at stock +10% levels. Experiment for tomorrow.

More data is welcomed. Does the Asus cards allow any voltage modification at all with the new drivers?

 

[HyperMatrix]

What settings are you running in Vantage? Res?

1920x1200. Extreme settings. It's a preset configuration. It's build 1.1.0. All tests were dx10.

 

[HyperMatrix]

More data is welcomed. Does the Asus cards allow any voltage modification at all with the new drivers?

Sadly, no. I've tried through a few applications that are designed to allow voltage increases but apparently it's been blocked in the new beta driver. At least for those apps. I haven't tried all software options.

The voltage adjuster is just completely removed from smart doctor with these new drivers.

I am running at 678mhz and 3708 memory without any voltage increases at the moment. That's 11.7% better results than the original drivers without OC'ing. 11.7% without upping voltage...while fully stable...isn't all that bad, to be honest with you.

 

[Keysplayr]

1920x1200. Extreme settings. It's a preset configuration. It's build 1.1.0. All tests were dx10.

Ok, thanks. I'll give that a spin.

 

[thilanliyan]

Heat that are radiated cannot travel towards the HDD bay because there are obstacles blocking its path. Radiation can't bend and will only travel at a straight line. The thermal reading of the HDD bay can't be a result of thermal radiation. Look at the bottom of the case in the 6970CF's IR graph, look at the heat distribution and compare it with the heat distribution at the HDD bay of the 6990 IR graph.

That is not what I am saying. I am saying since the components above the 6990 are warmer in the "450w" mode, it is possible that more of the heat generated by the card is being radiated upwards, leaving less heat going through the heatsink into the HDD bay. This can result in cooler temperatures in the HDD bay. For the 590, if less heat is being radiated upwards for example, more of the heat could be pushed through the heatsink into the HDD bay area.

Regardless, their testing method for the IR images is slightly flawed. Even a small difference in time when the case is opened and images are taken can have a large effect, especially in regards to how hot the AIR in certain places is.

 

[Seero]

That is not what I am saying. I am saying since the components above the 6990 are warmer in the "450w" mode, it is possible that more of the heat generated by the card is being radiated upwards, leaving less heat going through the heatsink into the HDD bay. This can result in cooler temperatures in the HDD bay. For the 590, if less heat is being radiated upwards for example, more of the heat could be pushed through the heatsink into the HDD bay area.

I see what you are saying. I am not going to pretend I know, because I'm not an expert on thermal radiation, heat transfer rate etc. I believe you too see my argument and somewhere we need to come to some...agreements.

Regardless, their testing method for the IR images is slightly flawed. Even a small difference in time when the case is opened and images are taken can have a large effect, especially in regards to how hot the AIR in certain places is.

I can agree with the bolded. I believe the purpose of the IR graph was originally made to demonstrate how much heat can be generated by video card into the case, not to be used as comparison. After all this pages of debates, I say, airflow is very important, and cables in front of vent/intake can have very negative effects. Besides, those are some really poor ventilations. Again, that maybe its purpose.

 

[thilanliyan]

I see what you are saying. I am not going to pretend I know, because I'm not an expert on thermal radiation, heat transfer rate etc. I believe you too see my argument and somewhere we need to come to some...agreements.

Yep I see what you are saying as well. I wish there was a clear explanation of those images but sadly only the people at hardware.fr probably know.