Cant get Prime 95 Blend Stable

[ric9887]

I have a QX9770, Blackops (G28), Cellshock 1866 2x1GB and am only trying to run everything at stock speed of 3.2Ghz, 400FSB 8x & 1600 8-8-8-16.
The CPU block is Fuzion V2, NB is part of the loop.

I can test with Everest, OCCT which are stable for >4h however running Prime95 25.6 I get a failure <1-2 mins.
The failures are always on the cores which have a lower temp.
@idle temps Core Temp .99
c0:43
c1:43
c2:50
c4:50
@load
c0:53
c1:53
c2:58
c3:58
EIST=Native
C1E=Disable
Freq Unlimited=Disable
CPU=1.2875
CPU Vdroop=Disable
CPU PLL=1.508861V
CPU VTT=1.200V
NB=1.250
SB=1.500
DRAM=1.648
Everything else is set to Auto

I dont quite understand the discrepencies between the the core temps. Is it usual to have these differences on Quads. The waterblock has sprung screws, and I have tried re-seating the processor, removed/reapplied AS5.

Can anyone suggest if these readings look normalish, and suggest what I can do to make Prime95 stable ? Any help would be appreciated.

I am beginning to think I have a duff processor. Whatever I do, the test always fails on the cores which show the lower temps - never on the higher temp cores.

Also Core Voltage in CPU-Z and Speedfan are persistantly showing 0.864v, I am not sure if this a bug though. The Aegis Panel shows 1.27. BIOS is 1.875 - EIST Native or Disabled, C1E Disabled.

Regards
Rich

[ric9887]

I have managed to find a setting which allows Prime to run and not fail on a couple of cores which it was previously.

I have changed the FSB to 395, and used a 10 multiplier. The memory is running 1:4 CPU333 @ 1560.

Core Temps @ load
0:60
1:60
2:64
3:64

I still can manage to run a 400 FSB :-( Not sure what to do next

[saaya]

hey rich, did you test the cpu on another board after all?
a 9770 that cant run 400fsb stable is very weird and i went through a lot of settings with you on msn that didnt seem to help a lot.
sure, 395x10 is nice, but having to tweak that much and still 8x400 is not stable is just not right...

[ric9887]

Hi Saaya, I have just had the box of kit back from being tested. It was found that the same thing happened in another BlackOps board unfortunatly.

I have had the BlackOps board changed for a Rampage Extreme in the hope that it is less sensitive, but initial testing looks like its exhibiting exactly the same problems. Need to get to know the BIOS a little better and more testing to know for sure.

Looks like im just unlucky with the processor :-( Will keep you posted.

Rich

[jolliffee]

Rich it strongly looks like yourQX9770 is at fault, not the board. If the CPU is the same in another BlackOps it is the CPU.

Try using IntelBurnTest, it is quicker to find a problem & RealTemp to find your correct core temp. CPUZ does not show correct voltage currently but the new version of AGIS Panal is very good.

I have two QX9770 on BlackOps running great, and no problem at default settings

[ric9887]

Hi jolliffee, well I bit the bullet and got another 9770 and it certainly looks to be a faulty chip. This new one is fine at stock and runs Prime95 perfectly.

Whats really odd about the other chip (the faulty one) is it ran every other test fine at stock, including Inteliburn - 20x and all the numbers matched. It would stop on 1 or 2 of the processors after only a minute or 2 running the Prime95 Blend test.Unfortunatly I have lapped it, so perhaps I damaged it doing that, or perhaps it was just a lower end of the threshold chip.

[JohnLoph]

how long for test on prime95 that refer to stable?

(b'cause i got fatal error on 11th hrs)