Printable View
Had anyone tried Prime95 with the Small FFT to max the FPU? I could run loops at 2.5G at blend with 1.65v but Small FFT wont run for over an hour at anything over 2.4 Ghz. Geez, these AMD64 CPUs are killing me - they overclock easily but just aren't stable in doing so. I can post at 2.6 GHz but true stability diminishes over 2.4 Ghz. :rolleyes:
You may be right man. There are quite a few of us with the same issue and many Winnies are the same week. I have yet to really run this thing hard due to work, so this weekend I will check for stabilty and ignore P95.Quote:
Originally Posted by dnottis
I have a week 37 Winni 3200+ (bad overclocker :( ). Managed to pass ~16hours of Prime95 blend at 2313 MHz (9.0 x 257) @ 1.575v.
Could you possibly try 2x512Mb modules? I have no problems with Prime either when running 2x256 BH5 (can run 265x10 or 253x11). But as soon as I installed my new 2x512 VX kit, it's not prime stable even at stock (memtest stable at 260x10).Quote:
Originally Posted by bachus_anonym
Hi
Well I did a new run and made a screenshot of it
http://www.upit.dk/u/P95-2600mhz.JPG
I have played several hours of Doom3 and running 3dmark too, absolutely no problems. I doubt P95 has problems with Winchesters, but as there are more ppl having trouble I cant say for sure, I have run both small Fft test for 24 hours and large Fft test also 24 hours, just didnt take a screen of it, but I have now :). Ohh yes I use 2x512 Pc 3200 Ballistix
Menty
What week is your 3200+?Quote:
Originally Posted by Mentaly
Hi
It says in my sig :) 0433
Menty
You guys should STOP using BLEND. If you want to test memory, test manually by -
1) Select FFT size that doesn't fit into caches
2) Select a LARGER time interval between changing FFT size
Blend tests CPU core as well, while the winches fail on memory. Hence if you want to test the memory controller MOST fully, use the mode that tests the memory only, as outlined above.
Let's say Blend runs for 10 hours beginning to end.. then 1/4 of the time it's testing the ram and the rest of the time it tests the caches, fpu. If you test the memory 100% of the time, it will fail faster.
Hi
Well I wrote I tried the others too, and none of those complaining about prime has talked about any settings special settings in prime. If P95 doesnt fail within 24 hours I seriously doubt its gonna fail ever.
Menty
great, not only do we have 3 "stock" tests, endless combos of CPU's, RAM, and mobo's NOW were going to start CUSTOM settings on P95?!!!
we need to start lowering the amount of variables of this problem NOT adding!
I dont know why so many ppl are so concerned with p95, just dont use it, its that simple.
There are other programs to test stability of your cpu and etc...
Quote:
Originally Posted by dnottis
I can't even get 2.2 Ghz to run more than an hour with SMALL FFT. Can blend for hours, these AMD64 chips just suck.
Err. this winnie has been priming more than 4 hours... Of course, this is also first time I was able to boot winnie on mach2 without holding breath... Puny 3.0+ with 9 multi... so I am stuck @ 280FSB... lol. Guess 2.53ghz is still decent for chip rated at 1.8ghz...
Correct me if I am wrong but priming at 1.8v only gets me to -51c on evap temp vs newcastle @ 2.95ghz 1.8v which went down to -43c. I am guessing it also runs few degrees cooler than newcastle. Now I am pondering if I should grab 3500+ winnie to replace my dead newcastle since I got it running for one of winnie....
Well maybe here is why... :rolleyes:
http://www.3dxtreme.net/images/oc/chipset.jpg
I am happy it's a Korean, but damn guys....
http://www.3dxtreme.net/images/oc/nf3ultra.jpg
Anyways, I just cleaned and applied Arctic Silver 5. Maybe chipset errors is the reason why I can prime 12 hours, then fail after 15mins. BTW this board was purchased from ZipZoomFly.com about mid november.
So prime failing at stock is valid RMA reason I take it, sweet. I am worried if my week 35 will pass at stock, and wont be that bummed out if it doesnt, cause it might be nice to have a later week for better OC. :fart:Quote:
Originally Posted by dnottis
If you are paying for a product and not getting what you paid for at stock, then yes I would say that would fall under "defective product" and would be worthy of an RMA. Overclocking something far out of spec and not getting what is expected, say an AMD3000+ @ 2.6GHz, is not worthy of an RMA.Quote:
Originally Posted by stelleg151
Ive given up on prime personally w/ this chip. Im glad im not the only one. Ive done so many tests, each having different results that I cant even trust prime. What I now do is just loop UT2004 with 32 bots... UT crashes much quicker then other games, especially w/ all those calculations with the bots. Most of the time it crashes within 10-20 min if system is unstable
I can't even run Prime95 at stock speed!!! :mad:
My setup:
MSI Neo2 Platinum w/ BIOS 1.41
AMD64 3500+ Winchester
OCZ 3200 Platinum Rev2 2-2-2-10-1T
OCZ 520W PS
I just spoke with AMD tech support and they claim in their lab they were able to run Prime95 with any 939 motherboards. They do not aware of any issue about Prime95 with Winchester... :stick:
AMD tech support suggested to run this utility first http://users.bigpond.net.au/cpuburn/. Oh well, I guess I will give it a try once I get home from work.
Interesting comment you have there...Quote:
Originally Posted by [ r2 ]
So I have a new car with 4 doors. And if one of the doors does not open and since I am single and no wife and kids. So I guess I should not care? :stick:
That's a poor comparison... it's more like you have a new car with 4 doors, and women and children cannot open them but you can, and since you are single with no wife or kids, you shouldn't care.Quote:
Originally Posted by p0k
Heh.. analogies. They are all false. Forget it :p
ALL
Are many people with MSI K8N Neo2's experiencing these problems?
p0k
Prime fails on what? BLEND. Blend tests memory controller. That CPU burn-in thing tests the CPU core and caches only.
Hi IvanAndreevich,Quote:
Originally Posted by IvanAndreevich
Prime95 test failed on Blend test...
Looks like I will be calling AMD soon to RMA the processor...
Hi all !
will try S&M 1.3.0 utility:
"S&M is a pretty efficient CPU stress-test, which also includes memory tests. The interface is available in Russian or English"
http://www.xbitlabs.com/articles/cpu...64-fx55_7.html
I think is very good. It can burn mine winnie to 59 C, when Prime95 to 55 C...
Helllo guys,
This is worth a try. Download P95 Ver 2.37 from here: http://www.midnightmachine.com/modul...wcat.php?cid=4
This is the last version and not the one currently available at Mersenne which is 2.38. Are we on to something?
Give that a try and let us know how you do. I am running the blend test ATM and it seems to be chugging along happily on my Winnie. :banana: :banana4: :banana: :banana4:
andyOCZ, the version 2.38 doesn't run with your winnie then?
I've tried the 2.36 with same results, going to try 2.37 now.
And about the winnie problem yes i can't run prime95 as well but system is rock stable.