Printable View
48 hours is usually sufficient (all my rigs pass at least 24 hours*31 days of Prime), my 3200+ winchester refuses to go above 2.4GHz (at any voltage or timings)without failing Prime95 within 24 hours, the RAM is using the 166 divider so the memory is @200MHz.
never made a cpuz screenshot when i've run prime but got those oneQuote:
Originally Posted by dnottis
http://www.lesfurieux.net/hfr/for_3200mhz.jpg
http://www.lesfurieux.net/hfr/for_5295mbench.jpg
http://www.lesfurieux.net/hfr/for_.jpg
http://www.lesfurieux.net/hfr/for_370cpumark.jpg
1.36Quote:
Originally Posted by Mrki
In my experience If you can run Prime95 for 2 or more consecutive passes of 1-5 torture test "Blend" run 3dmark 2001se looped 24hrs, I think the system is pretty fvcking stable!
265x10=2650Mhz, 1.73v; DDR530 6-2-2-2.0, 3.7v on the OCZ DDR Booster; Bios 1.4B1 :)
i've been running Prime95 Large FFT for about 2hrs now... just passed 448K FFT :)
i had to raise vcore full 0.1v above what SuperPi32M and 3dmark01 are stable with...
what's with the 270x10 in your sig then?Quote:
Originally Posted by bachus_anonym
http://www.xtremesystems.org/forums/...757#post622757Quote:
Originally Posted by trans am
If that's stable then why are you messing around with 2650? :fart:Quote:
Originally Posted by bachus_anonym
because despite 2700 being stable for all SuperPi runs and 3dmark01 - it is not stable enough for D2OL Distributing Project i participate in for XS D2OL Team :) 2650 is just fine but unfortunately not 2700 :(Quote:
Originally Posted by trans am
i need a rig that is capable of running 24/7 100% LOAD for daaaays without babysitting it
how long does it take you to get a fatal error in the P95 torture (small FFT) test?Quote:
Originally Posted by bachus_anonym
@ 270x10? 5s :) and raising vcore to max 1.81v REAL changes nothing in this department :(Quote:
Originally Posted by trans am
i tested vdimm up to 4.0v.... no luck.
Well when i first got my a64 system ( 3500 gig of geil msi k8n neo2 ) i ran prime for 12 hours on blend.. then 34 hours on max heat/power.
both with no errors and i consider that "prime" stable pc never crashes or errors ever.
when my new ram arrives gunna overclock and do the same.
Quote:
Originally Posted by OmegaSys
U didn't say about cooling... If Cooling make CPU Temp = constant u can pass every hour, everyday.. :D. PSU also is important 2 stable ...
Why never made a CPUZ screenshot... ?Quote:
Originally Posted by Boogotop
n What is ur cooling ? Stepping code , week of CPU ?
I haven't tried the K8NS Ultra, but I'm on my third nForce3 board now, and this NEO2 is the first that doesn't have a "Thermal Throtteling" setting in BIOS. I'm wondering if it is hidden and enabled by default. And if so, I'm wondering if that could mess things up a bit when coupled with the buggy winchester temp readings.
I had some problems earlier with my setup scoring 500-1000 points less than it should (got 12000 ALU at 2.8GHz and 1.74Vcore) in the Sandra 2005 CPU bench. Don't know if it was caused by sandra 2005 or something else. I have since then installed Sandra 2004 and gone through several BIOS versions.
My 3500+ winny is primestable at 254x11 and 1.63V (VapoPE), and the required Vcore curve isn't steap up until that point. But I can't for the life of me make 255x11 stable. I'm getting fairly good stability at 1.74Vcore (up to 260x11), but its a bit flaky (will prime for anything from 1 to 12hrs).
I'm using a Zalman 7000A-Cu on the cpu and an FB123 with a Panaflo 120mm fan blowing down on the RAM. My cpu temps seem stable at 54 celsius with the case closed (about 50 under full load with the case open). I'm using an OCZ 520w PSU. My voltages seem a little erratic (vcore seems to vary from 1.58 to 1.63 when set to +13% according to CPU-Z, Everest, and the BIOS ) but I've gotten the same results with an Antec TruBlu 480w and a Seasonic 400w. This is my second Neo2 as well.Quote:
Originally Posted by Zanr Zij
-OmegaSys
Ok guys I am doing some control testing, I am running 2 x 512mb of OCZ PC3200 Platinum Rev 2 TCCD at the 2-2-2-10 settings shown below on a GB K8NS Ultra-939. On my 754/DFI UT rig, these same sticks running together will run P95 for days at 200fsb. Looky see what happens here:
(Of course this is my wonderful Winnie 90nm 3500+)
Dont suppose you've got a newcastle you can try in the exact same system?Quote:
Originally Posted by andyOCZ
Cheers
Oli
I am waiting on an FX53 CPU that will be 130nm for testing in the same rig. I'll will ket you know what happens.Quote:
Originally Posted by Ol!ver
If anyone feels like submitting a request for a Prime95 revision, please send it here: woltman@alum.mit.edu The more we bug them, the better our chances on getting this issue worked out.
I sent an email already regarding this issue:
RE:attn prime95 authors serious problems running prime 95 and AMD64 Winchester
To Whom it may concern, I am a member of a computer enthusiast forum called xtremesystems.org. We at xs strongly believe there is a major compatiblity flaw/ or hardware flaw on the aforementioned. When running Prime95 torture tests on the A64 Winchester (90nm) We are experiencing errors at stock cpu speeds. Is there a way to revise the software for the new processor, or is this definitely a hardware flaw on AMD's part. Please help us.
Warm regards,
Alvin Hayek
:toast:
I also made a thread on the AMD support forums. IF you want to add, please go here: http://forums.amd.com/index.php?act=...=0#entry280162
so here's first part of my testing with "Prime95 & A64 S939 90nm Winchester" combo...
265x10=2655MHz @ 1.74v REAL, DDR530 6-2-2-2.0 @ 3.75v OCZ DDR Booster
MSI K8N Neo2 Platinum - Q47353.1 0440A2 S Taiwan
A64 3200+ 90nm "Winchester" - CBBFD 0441UPCW
Thermalright XP-90 w/ YTech 120mm
CLICK for Full Size Screenshot - 2048x1536 270KB
http://www.xtremesystems.org/forums/...chmentid=20613
that was with In-place Large FFTs but customized to use up to 415MB
it's only 3h36m test but it makes one full loop. most of the time, if i get errors, then they appear during Large FFTs...
ATM though, i started Full Blend Test @ 267x10.... we'll see how long it will survive ;)
BTW... to make my setup stable as shown in the screenshot i needed to raise vcore by 0.1v above what i'm stable with SuperPi 32M and other benchmarks including 3d benches...
as for vdimm this needed fine-tunning as even 0.05v too much or too little would give me error during 1024K FFT...
Looks good so far. How long did it take to fail when you were overclocked at 2.7?
No, don't stop it at 3hours+!!! ;)Quote:
Originally Posted by bachus_anonym
We will wait for your Blend Test results. :)
2700Mhz is absolutely maximum STABLE for this CPU. Prime95 fails at this speed almost instantly, within maybe 1 minute... although u think i might be possible to make it stable at 2670Mhz... so far it passed 8K FFT :)
I noticed that failure is more random on the Winchester as well. I can run Torture Test Priority 10 all day long without errors. The next day, I can run it again and error out within minutes.
For the sake of testing stability with Winchesters, I'm just going to run Super PI 32M and loop 3Dmark2001 for 12 hours until a solution is found.