Nope. The issues were still there for me when my Winchesters were plugged in.
Printable View
Nope. The issues were still there for me when my Winchesters were plugged in.
What does DRAM drive strength do?
If I attempt to run my ram at 250MHz (it used to run 2.5-3-3-10 1T with my 130nm) gives me thousands of memtest errors. IMHO the memory controller on this chip is flawed. Every single error is becuase the first byte is low by 8 bits (example 5611fbfd returns 5611fbf5) every error, every time. If there was an actual problem with the memory, I would imagine the errors would be less predictable than 3378 out of 3378 being of by 00000008 bits. Something is very fishy if you ask me. I'm RMAing this week 41 CPU before my 30 days expire.
FD versus F5:
1111 1101 (FD)
1111 0101 (F5)
That means one data bit stuck at 0 for quite a while (3378 errors).
Well, Lets add another post to this thread... I hate prime :) hah
Well lets begin, I have only had prime run once for over 8 hours, and that was when I was running 250HTT with ram at 2.5-3-3-7 timings...
However, after multiple failed attempts (somewhere between 1 - 5 hours) I finally tried stock settings, and well prime failed again... I had my ram timings set to 2.0-2-2-5 in the DFI bios (2/17) everything else auto ...HTT was set to 200 and cpu multiplier was 9x... started prime last night before i went to bed, and woke up to a rebooted computer...
Half the time the computer resets in the middle of the night, the other half of the time it just errors out.. All my settings pass memtest..
Currently I am trying 260HTT 2.5-3-3-7 (everything else auto) bank interleave disbaled, and well the system failed prime after 1.5 hours.. I am able to run 3DMark03, superpi at any meg, and memtest error free...
Is there any alternative to prime95?? Or is there something wrong in the code? or is it just an imcompatibility between the integrated mem controller?
Here is my CPU info:
ADA3000DIK4BI
CBBHD0444XPDW
1148299k40751
i have a very similar system and have the same prime problem. my system fails prime at stocking setting. i too was able to do anything @ 270 x 9 w/ memory @ 2.5-4-4-7 1:! except prime. i was thinking it must be my 20 pin fortron 530 psu. i was going to get a 24pin one. now i see you already have a ocz modstream 520. i really don't know what to do now :(Quote:
Originally Posted by aver
edit: i have tested the same cpu on both nf3 gigabyte and msi neo2... and it's prime stable up to 2.4g. now i think it could be the motherboard.
I have noticed with my Winnie that it doesn't like at all high memclk with prime. Have you tried lowering the memclk ?Quote:
Originally Posted by aver
As I am writing this, my 3000+ is at 270*9, memclk 150 (use A64Tweaker) for an actual mem frequency of 202Mhz. Perfectly Prime stable (custom 2K-4k) for over 21h (and counting :) ) at ~1.51v (VID: +5%) - But it would not run same prime test more than 2h at same Vcore and memclk 166 (actual mem frequency 221Mhz). And it takes ~1.59v (VID: 1.425 + 8.3%) to run at memclk 183 (actual mem speed 247Mhz).
Hope this helps
dom,
Thanks for the advice, I will try that tonight.. I don't know if i trust prime on this winny though... I know there is something wrong if prime fails, but I dont think it means that the system is not stable at the level I am going to use it.. my biggest concern is if I am doing any damage to the cpu.. Its temps dont go past 40C so I feel safe, but we will see...
But I will get back to you on the divider issue, I just read in another thread about how a guy saying that using a divider on the Ultra-D seems to work alot better for system stability, I will have to give it a go..
Ya, I had a Ultra X-finity 500W and I thought it was subpar so i replaced it with the Modstream, but now Im thinking it might not be that bad.. I ll have to test it tomorrow.. But the modstream is rockstable, all the rails are over, and i have 6 fans blowing in this case..Quote:
Originally Posted by snowice
Have you tried upping the chipset and ldt voltages? Since my systems was failing at stock speeds something is wrong.. I am running 2/18 bios right now and I am going to test with 260HTT loose timings though..
I am following this guide:
http://www.dfi-street.com/forum/showthread.php?t=5844
Ill let you know how my results are, most peoples have been positive so if you have already tried everything might be worth a shot..
Anyone have any insight as to how high the LDT and Chipset voltages should be used for safe operation? My chipset hasnt been over 40C so i think the temps are alright..
The DFI NF4 board does not work well if you just "plug an play". Each CPU type, Clawhammer, Newcastle and Winchester require special timings and settings to be prime stable. DFI Sreet is a good place for info and some inluding myself have posted our settiings in the DFI NF4 Fixes thread here on XS. Take a look.
switching from Neo2 to DFI Ultra-D i could prime for 100MHz more (2500 to 2600) with the same CPU/Mems ...
Interesting
So has anyone figured out a cure for the winnie and prime problems yet. My winnie will pass prime all day at 2.6ghz but the mem has to be on a 166 divider. if I run at stock setting with the 200 mem divider it craps out instantly but if I drop it to a 166 it flys. been toying with all day, so far im at 240x11 @ 1.55v (166 divider) prime stable. I would really like to run my ram higher but I guess that aint happening.
random question, what is the cure for hte prime symptom: expeced .4 but got .5? as a fail? what does that usually suggest the problem to be?
This could be many things. You may be overclocked too high. You may have your memory overclocked too high. You may need more CPU or memory voltage. You may have too much memory voltage. Your memory may be getting too hot.......Quote:
Originally Posted by el rolio
If you are at 200mhz with you memory and CPU, and you have an early Winchester, from week 37 to 40 or so, you may have a bad CPU. Later CPU's are generally not affected. Read the first few posts in this thread.