Printable View
I found a bug: using italian language, if I set max temperature to 100 C°, when I close the program and launch it again, max temp reverted to default 80° and frequently the sensors that I chosed are not in the provious position.
P.S. I found the Power Supply test very effective. After less than 10 minutes, it blew up my Tagan 580 W power supply with a couple of "bangs" and a disgusting burnt silicon smell....:)
I found a bug with linpack testing. I have a Core i7 920 and Windows XP x64 SP2, and when I run the linpack test (and also the PSU test), the CPU only gets loaded with 4 threads, not 8. On the normal OCCT CPU test, 8 threads does work when you enable the hyperthreading option. LinX 0.5.6 properly loads all 8 threads on the i7.
Is GPUmemtest to be trusted? I'm getting errors all over the place with my gtx280 which seems to be stable in all the games I play
BureBista > I didn't know about that feature. I only works with a limited numbers of PSU, right ?
003 > For the 1 min period : use the Infinite test ;)
Brama > i'll look into it, as i've got several reports on such bugs. I'm pretty busy at the moment though, so i can't give you an ETA :/
003 > for i7 and Linpack > This is not a bug. Your i7 only has 4 physical cores. Linpack does not support HT at all.
A somehow limited MB models. All ASUS equipped with EPU and all Gigabyte equipped with DES and maybe MSI with DrMOS.Quote:
Originally Posted by Tetedeiench
Something like that.
GPU memtest is buggered, imho. I've run Video Memory Stress Test for several hours w/o a single error in DirectX mode, while OCCT throws out errors all over the place.
How come LinX is able to load all 8 threads 100%? Maybe check with dualist to see how he was able to do that. OCCT is great now, it just needs to be able to load 8 threads on the i7 with linpack and it would be perfect! :yepp:Quote:
Originally Posted by Tetedeiench
Same for me. My Gtx260 throws shows 40,000+ errors.Quote:
Originally Posted by anzial
He's just launching two linpack, i guess. Nothing proves that this is better than launching only one, at the moment ;) I guess i'll implement that feature later on, but really, i don't think that a really good idea ;)Quote:
Originally Posted by 003
Yes, GPU: Memtest isn't reliable at all on GTX260/280 and cards of the same generation. I couldn't test it at all on those cards.Quote:
Originally Posted by icecold316
I guess that's what comes of limited access to test hardware :shrug:
Any changes in beta1?
I'm centralizing betas in my own board - much less hassle for me :D
http://www.ocbase.com/forum/viewtopic.php?f=5&t=19
Thanx
For those with GTX260 and such cards and GPU:Memtest problems : are you using XP or Vista ?
I suspect a problem with Nvidia's implementation in GT200 Vista drivers, as i've got a few GT200 users under XP without problems...
For those who still wonder, this problem seems to be due to a bug in GT200 drivers under Vista, not to a bug in my test ;)Quote:
Originally Posted by Tetedeiench
I'm working on it with Nvidia ;)
GT200 users, if you want to give a shot to GPU:Memtest, it works fine with Windows XP !
OCCT PT 3.0.1.b02 fixes the GPU:Memtest problem encountered by GT200 users under Vista.
More info centralized over there :
http://www.ocbase.com/forum/viewtopic.php?f=5&t=23
As usual, the download link : http://www.ocbase.com/download.php?fileext=beta .
It has been reported as ok with GT200 and vista.
YOU NEED 181.22 OR HIGHER DRIVERS FOR IT TO WORK
Are there any final release versions in sight yet? :D
Tetedeiench, I think nvidia has gimped your GPU test in the 182.05 and 182.06 drivers! With 181.22, the GPU test would get my GTX280 up to 46C+, but now on 182.05 it won't break 34C, the idle temp. FurMark still hits 45C+.
EDIT:
I'm on water of course.
OCCT 3.0.1 final
http://www.ocbase.com/perestroika_en/index.php?Download
Can you tell me more ? (OS especially ? )Quote:
Originally Posted by 003
i'm using a GTX260, XP32, and 182.06 drivers, and GPU:OCCT (from version 3.0.1, which haven't changed a bit since 3.0.0) still runs perfectly fine and reaches about the same temp than furmark ...
Perhaps you left the Error check mode enabled, which load the GPU much less, which is, for now on, normal (unfortunatly).
I came here to announced it :DQuote:
Originally Posted by xaver
It especially fixes the GTX260/Vista problem, which was due to a CUDA limitation under vista which i can't yet explain and thus label as a bug.
Be sure to use 181.22 drivers, at least.
my 8800gts 640mb fail the gpu memtest in the pass 2 .... it always doesn't detect errors in pass 1, but can't complete pass 2, I tried to manually change it to only make 1 pass, and it passes, I make the test again and again but just manually, I did it 5 times and it doesn't show any errors.
games are working fine.
I'm using 182.06
I think you're right, now that I have error check disabled, it seems to be working normally again, with one small quirk. It is reporting a lower GPU temperature than EVGA Precision 1.4 by about 6 degrees. I don't remember this problem on the 181.22 drivers, but I may just have a bad memory. I'm using XP x64 SP2.Quote:
Originally Posted by Tetedeiench
I recently discovered that G80 aren't compatible with Cuda 2.0. You need G84 or better cards. Sigh.Quote:
Originally Posted by phsinc1
This is seriously starting to hinder the GPU:MEMTEST usefulness for me, and i'm looking into CUDA alternatives... i may have fallen onto something though. Stay tuned.
I'm using version 3.0 and I ran the PSU test and my computer powers off within 5 seconds of testing. Does this mean I have an insuffient powersupply or the PSU is crapping out on me. I should say that my computer powers off after 4 mins of playing crysis. I absolutely have no issues with temperatures on my motherboard, CPU and video cards.
Current specs
PC power and cooling 1000W
e8400 @ 4.2
4 gigs of ram OCZ
Striker II extreme
285 gtx in sli