HD4870 Furmark black screen crash.

[dnottis]

I'm getting the black screen in Furmark after about 1 min. Happened with stock cooler and now water cooled. RTHDRIBL will drive the VRMS to about 90deg c, Furmark to 126c. I've got 2x120mm fans on the backside of the card, 1 x 120mm on the top side. Everything is sinked. Where the hell is the VRM sensor located that is listed in the newer builds of Everest?? RTHDRIBL will push the VRMs to about 90, Furmark will push them to 126c. Using Everest build 1454. This was taken right before I went to a black screen. (specs below)






Specs -
Q6600 G0 L737B @ 3.6Ghz
Asus P5K-E/WiFi (Bios 1012)
OCZ DDR2-8500 DDR2-1066 4096MB Memory
1/2" ID Masterkleer, Swiftech MCP-655, Thermochill PA120.2 Rad, D-Tek Fuzion, MP-01
Thermaltake ToughPower 1000watt PSU
HIS HD4870
X-Mystique 5.1 Digital Dolby Sound Card
Lian Li PC V-2000
Acer X222Wbd 22" LCD

Each of the 2 6pin Pci-e connectors is connected to a separate 12v line on my quad rail PSU. Each rail has 36a.

[Jansku07]

GPU VRM is way too hot @ 126 so I suggest that you use better cooling for it; you can reduce its lifespan radically by torturing it (if there isn't any cooling for VRM then you're crazy). HD48XX-cards don't like Furmark so I suggest that you use some other benchmark before ATI fixes it.

[tkpenalty]

I would add a low RPM 80mm fan onto that (try finding the half thickness fans), and purchasing the swiftec copper ramsinks and installing them on.

[oli_ramsay]

The heatsinks you are using are far too small. As a temporary measure you should have a fan blowing over it until you get a bigger one to replace it.

As far as I know nothing on the back of the card needs cooling but the VRMs get the hottest.

[dnottis]

Guys there are 120s blowing over the front and the back of the card - and Furmark is still pushing it to 126c. Its not an airflow issue.

RTHDRIBL pushes it to about 87c max, its just Furmark that does this. In games nothing breaks 40c, VRM about 60c.

[erocker]

Seems the ramsinks you are using aren't efficient enough. Plus, you haven't mentioned how high your clocks are.

[dnottis]

Originally Posted by erocker

Seems the ramsinks you are using aren't efficient enough. Plus, you haven't mentioned how high your clocks are.

stock speeds. I just stuck bigger ram sinks on there and the temps hit 126c just as fast with 120mm airflow.

[erocker]

Well, if you are positive everything is being cooled enough, the VRM's are getting too hot for normal conditions. Time to stick the stock cooler back on it and RMA it. *Are you positive everest is giving a correct reading?

*I also just noticed that 110FPS in furmark cannot be achieved at stock settings. You sure you aren't overclocked?

[dnottis]

Originally Posted by erocker

Well, if you are positive everything is being cooled enough, the VRM's are getting too hot for normal conditions. Time to stick the stock cooler back on it and RMA it. *Are you positive everest is giving a correct reading?

*I also just noticed that 110FPS in furmark cannot be achieved at stock settings. You sure you aren't overclocked?

yea, Im running Furmark 1024x768 in a window right now. And yes running it 1680x1050 full screen has the same effect just lower FPS.

[oli_ramsay]

I think that everst may be giving you an innacurate reading, download GPU-Z and check the 2nd temp under the sensors tab:



Everest is reporting 57c and GPU-Z is reporting 47c

[dnottis]

Originally Posted by oli_ramsay

I think that everst may be giving you an innacurate reading, download GPU-Z and check the 2nd temp under the sensors tab:



Everest is reporting 57c and GPU-Z is reporting 47c

In GPUZ all my numbers there are always under 40. W1zzard said thats not VRM, those are 3 on die temps at different locations. (source: http://forums.techpowerup.com/showthread.php?t=66067)

[oli_ramsay]

See what I mean, everest is reporting a stupidly high temp and GPU-Z is reporting ~60c. I hope to God that GPU-Z is correct.

EDIT: Took too long to post lol, I still think that Everest is reporting it wrong though. If there was a temp diode in the VRMs the why wouldn't GPU-Z show it?

[dnottis]

Here is the latest version of Everest with GPU diode and VRM temp support for HD4xxx - be sure you have it, just came out a day ago.

http://www.9down.com/EVEREST-Ultimat...55-Beta-29037/

[erocker]

Originally Posted by oli_ramsay

See what I mean, everest is reporting a stupidly high temp and GPU-Z is reporting ~60c. I hope to God that GPU-Z is correct.

EDIT: Took too long to post lol, I still think that Everest is reporting it wrong though. If there was a temp diode in the VRMs the why wouldn't GPU-Z show it?

Actually the three temps taken with GPU-Z all come off of the GPU. There is no VRM monitoring with GPU-Z.

* I took these temps using an infrared thermometer:

Using a HD 4870 at stock speeds with the fan set to 55% (stock cooler) and running 3 instances of furmark in a row and using my Raytek Raynger ST Infrared Thermometer the backside VRM area of the card tops out at 64.2C and the back of the gpu area hits 62.17C. Furmark is quite a torture test!

[dnottis]

Originally Posted by oli_ramsay

See what I mean, everest is reporting a stupidly high temp and GPU-Z is reporting ~60c. I hope to God that GPU-Z is correct.

EDIT: Took too long to post lol, I still think that Everest is reporting it wrong though. If there was a temp diode in the VRMs the why wouldn't GPU-Z show it?

Here is mine -

[oli_ramsay]

I used a cheaper method AKA the back of my finger. The backside of the VRM area got pretty hot but certainly nowhere near 100c.

dnottis, try just touching the back of the card with your finger and if it's too hot to touch then you'll need some bigger heatsinks! (make sure you have no sweat/moisture on you fingers 1st though !)

[dnottis]

Originally Posted by oli_ramsay

I used a cheaper method AKA the back of my finger. The backside of the VRM area got pretty hot but certainly not anywhere near 100c.

dnottis, try just touching the back of the card with your finger and if it's too hot to touch then you'll need some bigger heatsinks! (make sure you have no sweat/moisture on you fingers 1st though !)

It gets hot but I can still touch it. Its certainly not 250F like VRM temp in Everest says. My handheld thermal reader shows 60's c on the backside of the card. I can touch the sinks on the VRMs too - again hot but they arent burning me I can leave my finger on them.

[oli_ramsay]

If it's still crashing at stock speeds and at reasonable temps sounds like a bad card. If I were you I'd put the stock cooler back on it and RMA it.

[erocker]

Originally Posted by dnottis

It gets hot but I can still touch it. Its certainly not 250F like VRM temp in Everest says. My handheld thermal reader shows 60's c on the backside of the card. I can touch the sinks on the VRMs too - again hot but they arent burning me I can leave my finger on them.

That sounds normal to me! My thermal reader shows the same, and I'd trust that over any program. I think you are good. The furmark blackscreens on the other hand are actually pretty common and is the fault of the program not your card. You may want to get some nice forged copper heatsinks for the VRM's but really 60c is good and you'll be able to get some better clocks out of it. Good luck!

[dnottis]

Here you can see the sinks and the fans for airflow.

[erocker]

Your setup looks very nice! It may help to have a fan (somehow) directly over the rear of the card, blowing directly down on it. Something like an Antec Spot-cool comes to mind...

[dnottis]

Originally Posted by erocker

Your setup looks very nice! It may help to have a fan (somehow) directly over the rear of the card, blowing directly down on it. Something like an Antec Spot-cool comes to mind...

Look at the third pic, there are 2 x yate loons blowing over the card from the front end of it.

[erocker]

I saw that.

[dnottis]

Originally Posted by erocker

I saw that.

ok I do have a bracket that holds a 120 fan right over the top. I had that in last night and but same temps and same issue with Furmark. So it didnt help at all.

[mat9v]

Furmark is unstable on it's own. Try Ati tool artifact scanner for a few hours. Do you have any problems in games otherwise? I can't find any stable overclock or underclock with Furmark but Ati tool shows artifacts when core is set over 835Mhz. And 830 is perfectly stable for long hours of Crysis goodness