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

[Phynaz]

Assuming we are talking about the same thing. I am saying that there is a fan at the HDD bay (bottom right) when the 6990 IR graph was done. No other IR graphs have that fan, including 590.

If 590 IR graph does have intake fan too, then something is wrong with the IR graph. The area where the fan is blowing is no cooler than the top where the HDD seats.

Have you asked about it or posted at the site that did the measurements?

 

[Seero]

Have you asked about it or posted at the site that did the measurements?

Non, je n'ai pas un compte, et je ne sais pas français. Si je le fais, alors nous ne serons pas tellement de plaisir ici.

 

[kevinsbane]

If 590 IR graph does have intake fan too, then something is wrong with the IR graph. The area where the fan is blowing is no cooler than the top where the HDD seats.

The thermal image for the gtx 590 would also be consistent with an *exhaust* fan present in the location in question; arguably a better arrangement than blowing fresh air into the case at that point.

Is it possible that the GTX 590 is the one with the extra fan? Or perhaps they switched the direction of the fan between the two IR images? Doesn't make sense to me, but it does provide another alternative explanation for the IR images being what they are.

 

[thilanliyan]

If 590 IR graph does have intake fan too, then something is wrong with the IR graph. The area where the fan is blowing is no cooler than the top where the HDD seats.

If the hot air is blowing into that area, why does it HAVE to be cooler?

Also in terms of airflow, xbitlabs is saying the air is pulled through one heatsink to the other, then exhausted in the 6990. And the GPU temps they got seem to agree with that:

http://www.xbitlabs.com/articles/video/display/geforce-gtx-590_3.html#sect0
"The temperatures of the Radeon HD 6990’s GPUs differ by 10-12°C whereas the difference between the GPUs of the GeForce GTX 590 is only 2°C."

 

[HyperMatrix]

I don't know why everyone is tripping. I'm running this card at 1.000v, with 808mhz core clock, 1616mhz shader clock, and 4202 mem clock. Stable as a rock. 34c on idle, and 82c on load. Running old stock drivers from Asus website as that is what my card is.

I think people are just forgetting to up the fan speed when overclocking...or are not replacing it with watercooling vga blocks. This card is an AMAZING overclocker though. at 808mhz, that's a 32% OC. It's not that it won't go higher than that. It's that it worked at this rating for me, and I have no need to push it too high beyond that. But it very well can go higher. Seriously amazing card. Asus's claim of +50% OC is valid. With proper cooling.

 

[Phynaz]

I don't know why everyone is tripping. I'm running this card at 1.000v, with 808mhz core clock, 1616mhz shader clock, and 4202 mem clock. Stable as a rock. 34c on idle, and 82c on load. Running old stock drivers from Asus website as that is what my card is.

I think people are just forgetting to up the fan speed when overclocking...or are not replacing it with watercooling vga blocks. This card is an AMAZING overclocker though. at 808mhz, that's a 32% OC. It's not that it won't go higher than that. It's that it worked at this rating for me, and I have no need to push it too high beyond that. But it very well can go higher. Seriously amazing card. Asus's claim of +50% OC is valid. With proper cooling.

I'm sure you would be happy to back up this claim of a GTX590 with a 30% overclock.

 

[GaiaHunter]

If the hot air is blowing into that area, why does it HAVE to be cooler?

Also in terms of airflow, xbitlabs is saying the air is pulled through one heatsink to the other, then exhausted in the 6990. And the GPU temps they got seem to agree with that:

http://www.xbitlabs.com/articles/video/display/geforce-gtx-590_3.html#sect0
"The temperatures of the Radeon HD 6990’s GPUs differ by 10-12°C whereas the difference between the GPUs of the GeForce GTX 590 is only 2°C."

I read that as well, that I was thinking the 6990 would suck air from the front.

AT on the other hand says both are exhausts.

 

[pcm81]

I read that as well, that I was thinking the 6990 would suck air from the front.

AT on the other hand says both are exhausts.

Temperatures of GPUs do differ, however I believe that is because the 2nd gpu is essentially a crossfire slave to 1st gpu, hence 2nd gpu is only crunching when 1st gpu gives it work.

 

[HyperMatrix]

I'm sure you would be happy to back up this claim of a GTX590 with a 30% overclock.

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

 

[GaiaHunter]

Temperatures of GPUs do differ, however I believe that is because the 2nd gpu is essentially a crossfire slave to 1st gpu, hence 2nd gpu is only crunching when 1st gpu gives it work.

They got those temps reads at load - that kind of temp difference is quite huge. Not sure if that alone is enough to explain it.

And the GTX590 should show a similar effect.

 

[HyperMatrix]

 

[Keysplayr]

So you're GTX590 is faster now than GTX580 in SLI. I'd say that's not bad.

 

[HyperMatrix]

So you're GTX590 is faster now than GTX580 in SLI. I'd say that's not bad.

Yeah and I'm running it with my i2600k OC'd to 5ghz. =D Was 5.2ghz but I like to overclock a bit below the max for extra stability.

And also 16gb of 1833mhz memory. And running a Revodrive with 540MBps read time. And despite all this...Crysis one with all graphic card options fully maxed, and all game video settings at very high, still only gives me 50-70fps at 1900x1200.

.... that game. Lol.


Profanity is not allowed in the technical forums.

Idontcare
Super Mod

 

[Keysplayr]

Yeah and I'm running it with my i2600k OC'd to 5ghz. =D Was 5.2ghz but I like to overclock a bit below the max for extra stability.

And also 16gb of 1833mhz memory. And running a Revodrive with 540MBps read time. And despite all this...Crysis one with all graphic card options fully maxed, and all game video settings at very high, still only gives me 50-70fps at 1900x1200.

xxxx that game. Lol.

Watch the F-bombs and such. Not allowed in the tech forums.
Welcome to Anandtech!

 

[pcm81]

Yeah and I'm running it with my i2600k OC'd to 5ghz. =D Was 5.2ghz but I like to overclock a bit below the max for extra stability.

And also 16gb of 1833mhz memory. And running a Revodrive with 540MBps read time. And despite all this...Crysis one with all graphic card options fully maxed, and all game video settings at very high, still only gives me 50-70fps at 1900x1200.

Fuck that game. Lol.

But can it run dragon age 2?

LOL had to ask...

 

[HyperMatrix]

I haven't tried. I bought it for the 360. Is it supposed to be a demanding game on the PC?

 

[Keysplayr]

Metro 2033 is probably the hardest on systems right now.

 

[OCGuy]

Nice card you got there Hyper!

 

[kevinsbane]

Hypermatrix, is your card throttling at all?

 

[HyperMatrix]

Hypermatrix, is your card throttling at all?

No it initially was when I pushed it to 750 without upping the voltage, using the latest beta 270 drivers from Nvidia. Then I realized in order to allow smart doctor to change the voltage, you have to use Asus's own version of the driver.

After that, I pushed it to 1.013v and 854mhz and upped the memory alongside it, but I could see artifacts so I set it to 1.000v and 814mhz and it's been working fine.

I installed evga precision and have it report on-the-fly temperatures, individual clock speeds, utilization, mem speed and usage, etc....so I can see what it's doing all the time. As we speak, it's brought down gpu1 to 51mhz and gpu2 to 405mhz, with mem1 at 135mhz and mem2 at 324mhz. Kind of like intel's speedstep feature. But once you launch a game it goes full out.

I recently bumped the fan from 85% to 91% and even under load it doesn't go above 79 degrees celsius. So it's a pretty decent setup.

I tested with on the fly OC'ing to make sure I'm seeing a somewhat linear increase in framerate and I did. And if it wanted to throttle, it would have done that instead of causing artifacts and the random "driver has recovered" errors when I ran the clocks a little too high.

 

[Seero]

If the hot air is blowing into that area, why does it HAVE to be cooler?

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

Also in terms of airflow, xbitlabs is saying the air is pulled through one heatsink to the other, then exhausted in the 6990. And the GPU temps they got seem to agree with that:

http://www.xbitlabs.com/articles/video/display/geforce-gtx-590_3.html#sect0
"The temperatures of the Radeon HD 6990’s GPUs differ by 10-12°C whereas the difference between the GPUs of the GeForce GTX 590 is only 2°C."

IR pics from hardware.fr
6970 CF:

Can you see the heat dumped into the HDD bay?

Observe the temp on the cable between 6970CF and 6990. With 6970CF, the cables are hot and heat moves through the cable into the HDD bay. It should be clear that the cable in HDD bay is hotter than its environment. 6990:

6990 is completely different, it is clear that 6990 dumps a lot of heat into the HDD bay.

What is interesting is the cable is cooler than its environment, indicating that somehow, heat were dumped away from the cables. In fact, it disappears in that cold spot. Those cables disappears because they get cooled off while they travel near the bottom front of the case.

Now look at the routing of the HDD cable it comes from inside the case (it gets difficult to describe things in 3D), out towards the camera, then back to the HDD. Under IR, the part where it is inside the case disappears because the temp is so low. This effect is very different from 6970CF and 590.

Now, 590

Since heat as no where to go, the entire HDD bay is heated up, and wires are actually cooler than its immediate environment, but not cold, just not as hot. Now compare it to the 6990, not only cables are cooler than 590, but even cooler than 6970CF where there are no front vent. In fact, although my claim is that there are no HDD fan on 590, there is airflow, which came off the vent. Immediate area around the vent when evenly heats up. Yes the 2 HDD cables appears to be cooler, but again, it is closer to the camera (3d). Back to 6990, there is also one known airflow, the front of the video card, but why didn't temp evens out? There appears to be a 30+c temp different between the top of the vent vs the bottom of the vent.

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

 

[Seero]

No it initially was when I pushed it to 750 without upping the voltage, using the latest beta 270 drivers from Nvidia. Then I realized in order to allow smart doctor to change the voltage, you have to use Asus's own version of the driver.

After that, I pushed it to 1.013v and 854mhz and upped the memory alongside it, but I could see artifacts so I set it to 1.000v and 814mhz and it's been working fine.

I installed evga precision and have it report on-the-fly temperatures, individual clock speeds, utilization, mem speed and usage, etc....so I can see what it's doing all the time. As we speak, it's brought down gpu1 to 51mhz and gpu2 to 405mhz, with mem1 at 135mhz and mem2 at 324mhz. Kind of like intel's speedstep feature. But once you launch a game it goes full out.

I recently bumped the fan from 85% to 91% and even under load it doesn't go above 79 degrees celsius. So it's a pretty decent setup.

I tested with on the fly OC'ing to make sure I'm seeing a somewhat linear increase in framerate and I did. And if it wanted to throttle, it would have done that instead of causing artifacts and the random "driver has recovered" errors when I ran the clocks a little too high.

Gratz on your monster setup. However, I must warn you, > 1v is very dangerous. If you haven't read through this thread, there are lots of posters who spent efforts on pics describing how much heat those baby pushes out, which isn't really a problem assuming your case is as hardcore as your setup. The problem is, those VRMs may give under high voltage, like > 1v.

 

[HyperMatrix]

Gratz on your monster setup. However, I must warn you, > 1v is very dangerous. If you haven't read through this thread, there are lots of posters who spent efforts on pics describing how much heat those baby pushes out, which isn't really a problem assuming your case is as hardcore as your setup. The problem is, those VRMs may give under high voltage, like > 1v.

I've got 6x 120mm fans and 1x 200mm fan. CPU is watercooled and doesn't heat up the board whatsoever. Motherboard is sitting at 24c.

3 fans in the front pushing air into the system, 1 fan on the side pushing air into the system, and 1 giant fan up top pushing air in, with 2 stacked fans acting as an exhaust to get everything out of the case. got some very nice flow going there.

And yeah I don't plan on going beyond 1v. I'm happy at that mark. The card OC'd a lot better than I expected, considering all the negative reviews and fear mongering on the internet.

 

[Tempered81]

No it initially was when I pushed it to 750 without upping the voltage, using the latest beta 270 drivers from Nvidia. Then I realized in order to allow smart doctor to change the voltage, you have to use Asus's own version of the driver.

After that, I pushed it to 1.013v and 854mhz and upped the memory alongside it, but I could see artifacts so I set it to 1.000v and 814mhz and it's been working fine.

I installed evga precision and have it report on-the-fly temperatures, individual clock speeds, utilization, mem speed and usage, etc....so I can see what it's doing all the time. As we speak, it's brought down gpu1 to 51mhz and gpu2 to 405mhz, with mem1 at 135mhz and mem2 at 324mhz. Kind of like intel's speedstep feature. But once you launch a game it goes full out.

I recently bumped the fan from 85% to 91% and even under load it doesn't go above 79 degrees celsius. So it's a pretty decent setup.

I tested with on the fly OC'ing to make sure I'm seeing a somewhat linear increase in framerate and I did. And if it wanted to throttle, it would have done that instead of causing artifacts and the random "driver has recovered" errors when I ran the clocks a little too high.

Good deal. Keep us posted on this thing. It's nice to hear a real success story amid all the horror stories in this thread.

Keep pushin it!

 

[AnandThenMan]

The card OC'd a lot better than I expected, considering all the negative reviews and fear mongering on the internet.

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