mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Hardware (https://www.mersenneforum.org/forumdisplay.php?f=9)
-   -   Error message during torture test (https://www.mersenneforum.org/showthread.php?t=13169)

esqrkim 2010-03-14 09:37

Error message during torture test
 
I built a system using Intel Quad 9550 on an Asus MB. I have 1GB X 4 of Crucial Ballistics (1066 MHz) in the four memory channels. When I perform the Blend torture test, I get some cores that are failing the test within couple of minutes. It is not always the same core that fails. The CPU temp went up to about 46C. Here's what happened:

1) First test: Almost immediately, cores 3 and 4 failed with an error, but no description of the reason was given.

2) Second test: Within a minute into the test, core 1 failed with the fatal error message "Rounding was 0.5, expected less than 0.4"

3) Third test: Core 1 failed with the same fatal error as above, and core 2 failed due to illegal sumout.

4) Fourth test: All cores have completed 4 tests without any failures, but then core 3 failed with the rounding error. The CPU temp is hovering around 53C-55C.

What's going on?

Thanks for your input.

MatWur-S530113 2010-03-14 10:12

Hi,

the torture test mainly tests CPU, MB and RAM. The temperature of the CPU is ok, I recommend to run an intensive test with memtest (memtest.org). [SIZE=2][URL="http://www.memtest.org/download/4.00/memtest86+-4.00.floppy.zip"][/URL][/SIZE]

S485122 2010-03-14 18:27

Did you gave enough voltage to the RAM in the BIOS setup ? Look at the specifications of the RAM. Another thing is that very often you cannot use the four slots for the RAM concurrently at full speed. Try to see what the QVL says about your RAM.

Jacob

esqrkim 2010-03-19 23:21

The Crucial Ballistic RAMs are in the QVL for the Asus motherboard that I have. The RAMs do spec out 2v, so I changed that voltage to 2v in the BIOS, but still have the problem.

I tested each RAM module in A1 slot without any problems, so the RAM modules seem to be good. Then I took one of the RAM modules and tried all four slots one at a time. It turns out that A1 and A2 slots are good, but Prime95 fails when a RAM module is used in B1 or B2. I am concluding that it's a bad motherboard. Any thoughts?

Prime95 2010-03-20 01:09

[QUOTE=esqrkim;208954]I am concluding that it's a bad motherboard. Any thoughts?[/QUOTE]

I think it could still be bad RAM or CPU. When using both memory channels the on-chip memory controller might not be able to handle the dual-channel timings.

Try populating B1 and B2 and reduce the memory timings (not that it will pinpoint the culprit). It may be a case of swapping parts until the problem goes away. Do you have a friend with 4 equivalent memory sticks you could try?

S485122 2010-03-20 10:13

[QUOTE=esqrkim;208954]Then I took one of the RAM modules and tried all four slots one at a time.[/QUOTE][QUOTE=Prime95;208963]When using both memory channels the on-chip memory controller might not be able to handle the dual-channel timings.[/QUOTE]Since esqrkim tested only one module at a time it is not related with dual channel issues. As esqrkim concluded, there is a problem with the B slots.

What is your motherboard ?

Jacob

esqrkim 2010-03-20 17:39

The board is Asus P5Q Pro Turbo. I did request for a replacement board from the retailer. It arrived yesterday. I replaced the MB and did the following:

1) Started with one RAM module in A1, passed couple of torture tests
2) Added second RAM module to B1, passed couple of torture tests
3) Filled the remaining A2 and B2 with modules, passed 10 hours of torture tests.

Bottom line, it was something on the MB that was malfunctioning. By the way, I did have 4 modules of 1Gb, 667 Mhz modules from my old computer that I first tried out on the Asus Pro. There was no problem when i used 667 Mhz modules. So, I am concluding that something on the MB was not able to handle 1066 Mhz data transfer rate.

esqrkim 2010-03-20 21:45

One more thing.

I am overclocking the Q9550 from 2.83 MHz to 3.20 MHz with no stability issues with Prime95. The temperature is running about 3 degree C higher (61C). Is this normal for overclocked condition? I could not find an operating temperature spec for the CPU, so I'm not sure what is acceptable.

Thanks for your input.

S485122 2010-03-21 08:25

You can find some data about Intel processors at the following address [url]http://processorfinder.intel.com/[/url]

When I look at your processor the thermal specification is 71,4 °C. The Q9550S seems more resilient : with a lower thermal design power (65 W instead of 95 W) it has a higher thermal specification : 76,3 °C.

You could use a better cooler if you are not comfortable with those temperatures.

Jacob

esqrkim 2010-03-21 15:28

Thanks for the info on processor specs. A more extensive test at 3.20 MHz caused Prime95 torture test to freeze. When I tried overclocking at higher core voltage, I got temps that exceeded 70C, so I'm looking for a cooler that performs better than the stock version. I've started another thread for this. I'd appreciate your input on that thread.

Thanks.


All times are UTC. The time now is 03:26.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.