5870CF: Temps in vantage

only second run of vantage ever, max temp: 76C on GPU2, 69C on GPU 1
fans never went above 30%, which means almost completely silent.

can't see where people are getting 95C, ambient is ~75F I'll run any other scenarios that are requested. Have Empire TW, Napoleon TW, WoW, ME2 and Metro 2033 that I might play around with.
Measured and logged with latest version of GPU-Z


here's the comparison link


ORB - Compare
^ with 1333mhz ram, CCC settings set to high performance.

will update tomorrow with TIM replaced.

 

what's your AlienFusion set to and what what settings did you run in vantage?

 

Those temps are nice and cool!
Fan speed is reported incorrectly by most (if not all) programs, AFAIK, so you can disregard it.
Try running Furmark Hot as Hell edition for 10 min to see the max temps you can ever get on GPU.

 

It's on high performance. I have the basic version of vantage, so the settings are the default.

will do later today. installing wow atm otherwise I would run it now.

yup! I *almost* thought something was wrong the first time I eyeballed the temps, but they're both working @ near 100% and @ 3d settings.

 

after a 1hour 26minuite run

GPU1 81C
GPU2 96C

theres a GPU3(w/e that is) here as well 88C


The forum has a 5xxkb limit for txt files(and the notepad is 1.01mb), otherwise I'd post it up.

I want to say it worked like a charm, but needless to say, there's nothing charming about a GPU that runs at near the boiling point. Tomorrow will be interesting.

 

GPU2 must be the memory sensor.
Yeah it's hot, but the test is beyond realistic, right?
You can never get to that point during gaming.

 

I can with an overclock of 850/1150 when playing GTA IV. I got a max memory temp of 93C so far. But who cares, these cards can handle those temps and I've got 3 yrs complete care so I'm not worried.

 

Technically, 93 is not 96 Plus, you OC'ed the cards.
BTW, quite a few 5870's have died from overheating already.
But I agree, with the 3yrs warranty you have nothing to worry about.

 

The only ones that have died are the crappy Asus ones. No Alienware or Clevo ones have died as far as I know.

 

hehe, me too. 3 years

I wouldn't know, it would make sense, but why wouldn't there be a GPU4?


as for the TIM tomorrow, I'm going to try straight TIM replacement, then lapping of the HS, then memory pad replacement and test the results individually.

 

3 for Core, Shader and Memory and one logical showing some sort of average/max between them, not sure though.

Looking forward to your tests! Make some photos please, if you can.

 

Run furmark multigpu in extreme burn mode. enable postfx and displacement mapping.

My 5870 would reach 100 degrees Celsius after about 15 minutes.

 

Going to head up to the store and see if I can find a cheap camera mount to make a video instead, otherwise I'll be sure to take some snapshots for sure. I believe you were correct however. There's another GPU temp sensor here in GPU-z, instead of being GPU4 it's just called GPU temperature, in the txt log, it's right after the core/memory clock and before the fan speed.

for the sake of consistency, I'll run both with the orginal settings (Xburn mode, postfx on DPM off) and those settings as well.

 

Vantage:
11950
ORB - Compare


GPU temp: Max 87C
GPU 1: 73.5C
GPU 2: 87.5
GPU 3: 80C

For comparison, Dell TIM values are:
GPU temp: 87
GPU1 73.5
Gpu2 87.5
GPU3 82C


hot like hell up next.

I decided to stop after a 5minuite run.
GPU1 85
GPU2 107
GPU3 98
Ambient room temp was 10 degrees higher this time, so I will re-run when it cools off a bit.

still waiting for my video to process, WMM is a horrible program and my 6 core BE 1090 desktop @ 3.8ghz is taking forever to process the 1GB movie. However I suspect during the installation when one of the gap pads came off it may have damaged the thermal properties of the pad, I'll have to order up some new pads in order to rule this out. Also, I'll verify that the TIM is making full die contact. Otherwise the results are somewhat clear, it should be noted, that the results are absolute maximums, which in some cases may only be for 1 log file line. An averaging of both txt files may be a better way to determine if there was an improvement, however I don't feel there's enough consistency to do this.

also, for anyone that wants to know, the Gap pads for the RAM look to be approx 1.0MM. I believe in the video I say it's about 1.5-2mm, however After reconfirming the size of 1 mm (lol) I'd say they're 1mm for sure.