Just tried prime 2.37... same s**hit!
I know the solution for all this mess with the winchesters and AMD 64 bit cpu's generally... sell them and next system will be INTEL.
Printable View
Just tried prime 2.37... same s**hit!
I know the solution for all this mess with the winchesters and AMD 64 bit cpu's generally... sell them and next system will be INTEL.
P95 v2.37 blend is still running and I jhave to go out. Try P95 at 200fsb first. Get that going, then tweak.
doesn't matter man, i've tried all of the above... it's just the cpu.
Although i've found another program that stresses the CPU more than prime, it is called "Toast" :D
Runs fine here... system is rock stable even if prime doesn't want to compensate with my cpu.
Quote:
Originally Posted by andyOCZ
huh, i have: Version 23.8.1 :idea: :confused:
open P95, help, about Prime95...should list version #
My overclock has slowly deteriorated. I was able to prime for 12+ hours at 250 x 10, now that won't overclock for more than 2 hours regardless of the voltage at the same speed. Even 240 x 10, which it primed for 17 hours at stock voltage was ok, now that won't prime. I think this chip is slowly dying - it's a week old. I believe AMD has a real problem on their hands with these chips, I'm with Byron I may go back to Intel, this is just getting ridiculous.
I am in the same situtation as Byron.
Prime95 v2.37 and v2.38 failed within one minute mark.
I am not even overclocking my system. I can loop memtest 1.4 test #5 hours with error free. Ran 3DMark03 and 05 and looping with error free. Pi 1M, 8M and 32M without trouble. Day to day operation is perfectly fine. No blue screen error and I can game HL2 with 6x AA and 16x FA hours with crash free...
I guess RMA is my only option.
So you guys have systems that are in all respects stable... never crash/blue screen doing everything you do on your PC, yet you would still consider RMA'ing the CPU just because prime95 doesn't run error free?
Those of you, having the issues with the blend mode.. Have you, at any point, ran very high volts on the RAM?
dnottis
I think you're jumping the gun. We'll figure this problem out!
p0k
Yeah, if your CPU is a $hitty o/c it's a nice reason to RMA it heh :)
Anyone ever try this app? Hot CPU Tester Pro 4
http://www.7byte.com/index.php?page=hotcpu
Cause at 2.5 I can run it at 1.55v. Prime doesn't have a chance at those settings, yet this seems to be ok.
Might be something like good old pentium 90 episode. Intel goofed off, someone found it, there were waves of "RMA" for those who bothered. Majority of people didn't bother and OS was updated "promptly".... Of course, it is not as simplistic as back than to find out what might be causing this things.Quote:
Originally Posted by Torin
EDIT: UPDATE: Time to give up. Don't bother with P95 v246 shown below. NO love for the week 37 Winnie with that version either.Quote:
Originally Posted by p0k
Guys, we have only begun to fight! ;)
I went into the nerdy world of the Mersenne Forums and found a version of P95 written and optimized for the Athlon XP. It's new and they said it was beta, but in between the geek speak I figured out it is missing the P4 optimizations. It's version 2.46 (About Prime95 says 24.6) and it's available here: ftp://mersenne.org/gimps/p95v246.zip
Don't give up yet. Please give this a shot. It seems to be doing ok so far on my MSI Winnie combo.
And yes 2.37 died after 3+ hours at 200 HTT for me also.
Ok guys,i've just tried and the 2.46 prime version... same bolllocks.
No more AMD for me from now on.
Thanks,
Byron
why do people need prime95 to be stable? personally i dont run prime95 as an every day app, so i wouldnt care if its stable or not,
i guess i have different opinions than a lot of people. personally i dont run anyway near my max stable clocks everyday, because i feel i simply dont need 2.7ghz of cpu power that my cpu is capable of, when 2ghz gets the job done just fine
0441 3200+ here, no problems on blend mode for 48+ hours. Prime version 23.8.1. 10x250, 1.57 vcore, MSI Neo2 1.5B1, Kingmax 1:1 2.5,3,3,10. Enermax Noisetaker 470 PSU.
Same test fails at 260 x 10 after about 20 mins. Maybe these babies are just more sensitive to heat/high voltage than the 130 process. I guess the electrons are getting too excited and jumping where they are not supposed to jump. I would use this analogy:
Winnies are like married men: when with thier wife of x years they can go for any length of time, but get them near a new hot thing they blow after 2 mins.
Just a thought :)
Also, for anyone serious about this, I would start a correlation chart for hardware/system settings for systems that fail prime. I've noticed a few OCZ PSU's. I thought they and Antec's might have issues with MSI and possibly any 939 MBs? Personally I had an Antec 380 TP that came with my Sonata case that blew right up, no shutdown.
Time to get a divorce me thinks :D
Or bring on the Sledge Hammer!!!!
Standards and expectations, I guess, though I agree with you. If some games were released that could actually challenge or need my 3ghz of CH power, that would another matter entirely :DQuote:
Originally Posted by blinky
As long as I can play games for extended periods of time, its stable to me :)
Ive given up on prime with my winni, as almost every setting dies after 2hrs
I installed my 2x256MB BH5 modules again just to make sure. Prime would fail in less than 30 minutes on this system at stock speeds when using a 2x512MB VX kit (I have a few more 1GB DC kits to try later). Here it is, "blend" stable for 12hrs at 250x11 2-2-2-5 (I haven't turned Vdimm back up to 3.6V after the VX test, so 250-255 is max at 3.3Vddr)
How much volts you had the BH-5's at stock speeds and prime was failing?
I have no problems running Prime on this system with 256MB BH5 modules. It's when I install 512MB modules it gets flaky even at stock speed.
Ok, so it is deffo a cpu issue.
Well guys, after several days of careful testing at 200fsb Prime95 and Winchester A64's (before week 41?) can be called incompatible. We have tried several P95 versions and a beta optimized for the AMD XP CPU, without P4 optimizations.
Some investigation into the issue with AMD may turn up a solution. I will say that AMD has been very good to our overclocking community and we should be thankful for that. :toast:
To address those of you that keep asking why anyone needs P95, the answer is that they don't. The fact that we have a limitation as to what we can run on our PC is the real issue. If I wanted to play Barney the Dinsoaur games and I could not for any reason that would be upsetting.
Now for a reality check:
1) We do know that Prime95 does not run for long on most or all of the "early" 90nm A64's.
2) We don't know if a revision has been made to the 90nm A64, if the CPU's are being "binned" differently (higher scrap rate) or if there may have been a FAB issue causing some of the CPU's to exhibit this behavior.
3) There is the possibilty that something in P95 Blend is "the cause" of this. We may have found a bug in P95 with our extremely accurate and critical 90nm Winnies. :) This might not be such a strech if we do find that all other apps that do heavy mathematical calculations work without fail. I have seem some evidence of this, as many of you have.
Bottom line is we still don't know exactly what is up.
Help me out ladies and gentlemen, please post your 90nm CPU FAB date and if P95 Blend will run. If we can isolate this to a time period that may help.
We also need some suggestions for a replacement stability program for these CPU's. Many of us have doubts as to stabilty, since our main tool is useless. A tool that does heavy calculations under 100% load is needed. A simple "warming" proggie that runs in the background is not at all convincing. We need something that test CPU and memory that we can trust.
Thanks for the help!!!!
The shot I posted above is from using a week 37 3500+ winny, full stepping:
ADA3500DIK4BI
CBBFD 0437VPDW
1102234J40023
It runs fine (and overclocks fine) when using 2x256MB sticks. I have only tried a 2x512 OCZ VX kit (memtest stable up to 260 DC 2-2-2-8 in this very same system), but this fails in Prime even at stock speeds (I'm testing other 1GB DC kits as we speak).
MSI K8N Neo2 Platinum - Q47353.1 0440A2 S Taiwan
A64 3200+ 90nm "Winchester" - CBBFD 0441UPCW
2x256MB Kingston HyperX PC-3200 (Winbond BH-5)
Thermalright XP-90 w/ YTech 120mm
21hrs Blend stopped by user @ 260x10 6-2-2-2.0 @ 1.76vcore 3.71vdimm
i will test it with my 2x512MB Mushkin Lvl II PC-3500 later on....
so I think with my week 35 3000+ im going to try and run blend for 24-36 hours at stock speeds and volts and if it fails, RMA the chip. Anyone disagree with this, or think I should do something else? Thanks.