Sign in with
Sign up | Sign in
Your question
Solved

Prime95 Error - Suggestions?

Last response: in Systems
Share
April 21, 2012 11:54:20 PM

Hello guys and girls

I built a computer two weeks ago with the following parts:

Motherboard - ASRock Z68 Extreme3 Gen3
Processor: Intel Core i5-2400 (no overclock, stock settings)
HDD - Samsung by Seagate Spinpoint F3 500 GB 7200 RPM Drive
SSD - None installed
PSU - Seasonic 620 W M12II 80 PLUS Bronze
RAM - Crucial Ballistix Sport 1600MHz ddr3 2x4GB (2 sticks of 4 GB each) (no overclock, stock settings)
DVD Burner - Lite On iHAS 124-04 Black OEM
Blu-Ray Burner - LG WH12LS39 12X Super Multi Blue

It has been rebooting/BSOD-ing intermittently since the first few days I built it. Ran Prime95, gives up on at least 1 core after 5-10 mins, says "Hardware failure detected". Running test "In-place large FFTs"

- Ran memtest86+ for 9 hrs 45 mins, says RAM is good. I therefore see no reason to suspect RAM.

- I removed stock heat sink, installed CoolerMaster Hyper212 EVO today, made no difference. System intermittently restarts. Prime95 still fails.

Also installed additional fan in front and tweaked cable routing.

- I can provide minidumps if some one is willing to take a look.

Should I RMA the board and see if that fixes the problem?
April 22, 2012 12:09:05 AM

Check your temperatures with RealTemp or HWMonitor. See what you get in Prime95. Do you think you installed the heatsink and its thermal compound correctly?
m
0
l
April 22, 2012 12:16:33 AM

kajabla said:
Check your temperatures with RealTemp or HWMonitor. See what you get in Prime95. Do you think you installed the heatsink and its thermal compound correctly?


It is Core 1 or Core 4 which give up. Temps are between 40 - 43 deg C (104 - 110 deg F). Temperatures used to exceed 50 deg C (122 deg F) before installing the extra cooling.

I think I installed the heat-sink + thermal compound correctly. I had this problem before installing the new heat sink too. A wee bit of thermal compound fell on the CPU, tried wiping that with a coffee filter.
m
0
l
Related resources
April 22, 2012 12:20:57 AM

Does the small fft's test fail also?
m
0
l
April 22, 2012 12:27:59 AM

Is it always the same 2 cores? If it is, you might have a bad CPU. If there are errors on all cores, it might not be.
m
0
l
April 22, 2012 12:41:28 AM

jsc said:
Does the small fft's test fail also?


Been running it for almost 15 mins now. All cores running - temps between 41 and 44 deg C (105-11 deg F)
m
0
l
April 22, 2012 12:49:55 AM

math1337 said:
Is it always the same 2 cores? If it is, you might have a bad CPU. If there are errors on all cores, it might not be.


I will run large FFT's test again to see if cores 2 or 3 fail too. Small FFT's test still running without a hitch on all cores.
m
0
l
April 22, 2012 1:11:24 AM

math1337 said:
Is it always the same 2 cores? If it is, you might have a bad CPU. If there are errors on all cores, it might not be.


Core 3 failed once in one of the torture tests.
m
0
l
April 22, 2012 2:06:35 AM

those temps idle temps or Prime95 blend temps?
m
0
l
April 22, 2012 2:13:37 AM

logang said:
those temps idle temps or Prime95 blend temps?


Those are temps displayed by HWMonitor while running Prime95.
m
0
l
April 22, 2012 2:23:39 AM

2 weeks, and stop wasting your time and take the cpu back asap, you may fall within the rma period, which would just a be simple swap with a new cpu!
m
0
l
April 22, 2012 2:37:50 AM

teh_gerbil said:
2 weeks, and stop wasting your time and take the cpu back asap, you may fall within the rma period, which would just a be simple swap with a new cpu!


Let's hope that's the only part that needs replacement.
m
0
l
April 22, 2012 2:46:23 AM

Indeed! Either way, 2weeks old and so long as their's no physical damage it shouldn't be erroring like this.
m
0
l
April 22, 2012 3:25:12 PM

Is there any way replacing 1600 MHz RAM with 1333 MHz RAM will solve the problem? I did some reading and one person from Intel (Chris something) says that running memory higher than 1333 MHz might cause warranty claim to be turned down.

Should I try this first?
m
0
l
April 22, 2012 3:30:26 PM

No. It can indeed cause warranty problems, but it cannot cause physical problems.
m
0
l

Best solution

April 22, 2012 3:50:53 PM

Memory tests are nice, but they do not rule out memory fails, they can only confirm them. Suggest you run with only 1 dimm then repeat with only the other and see if it fails consistently.

Of MB, CPU and power supply (all of which can cause your symptoms) PSU fails most, MB second, CPU distant third. But do rule out memory.

Also, I've seen similar problem caused by PCI network adapter. Reseating it fixed the problem. I've seen similar with a failing HDD -- only detected that when a linux based backup hung, no smart data reported fail but swapping hdd instantly stabilized system.

Std debugging is to pull part until it stops failing.
Share
April 22, 2012 8:13:08 PM

Ok. Here is what I did:

1) Ran 1 stick of RAM, Prime95 failed on Core 4.
2) Removed that, ran the other stick, Prime95 failed on Core 2.

So:
- Prime 95 fails on at least one of the 4 cores,
- Any of the cores can fail while running Prime 95.
- In some instances, system reboots almost immediately after starting Prime95 for testing "In - place Large FFTs".
m
0
l
April 22, 2012 8:37:01 PM

sanirudh said:
Ok. Here is what I did:

1) Ran 1 stick of RAM, Prime95 failed on Core 4.
2) Removed that, ran the other stick, Prime95 failed on Core 2.

So:
- Prime 95 fails on at least one of the 4 cores,
- Any of the cores can fail while running Prime 95.
- In some instances, system reboots almost immediately after starting Prime95 for testing "In - place Large FFTs".



Bummer. Memory is cheap.

Can you double check in BIOS that the voltage / CAS latency is set correctly for your memory. Google tells me it should be CL-9 latency (9-9-9-24), 1.5V, you shoudl cehck against your memory package. Rule out the BIOS is doing some funky auto OC based on JDEC, etc. Maybe double check what voltage the CPU etc are set to. There was a bios update for that board to run later i7's. If you haven't loaded bios it's a tough call -- if something is bad and it fails during bios load the MB may not be usable.

Suggest you download and run a diagnostic from a memory stick in your USB drive. Before running this test unplug your hard drive and your optical drives. Make it fail again.

Do you have a known good power supply you can swap in from another computer ? IF the power cables are long enough you don't even need to un-mount the old one and mount the new one.
m
0
l
April 22, 2012 9:48:31 PM

tsnor said:
Bummer. Memory is cheap.

Can you double check in BIOS that the voltage / CAS latency is set correctly for your memory. Google tells me it should be CL-9 latency (9-9-9-24), 1.5V, you shoudl cehck against your memory package. Rule out the BIOS is doing some funky auto OC based on JDEC, etc. Maybe double check what voltage the CPU etc are set to. There was a bios update for that board to run later i7's. If you haven't loaded bios it's a tough call -- if something is bad and it fails during bios load the MB may not be usable.

Suggest you download and run a diagnostic from a memory stick in your USB drive. Before running this test unplug your hard drive and your optical drives. Make it fail again.

Do you have a known good power supply you can swap in from another computer ? IF the power cables are long enough you don't even need to un-mount the old one and mount the new one.


Thanks for the quick reply.

-Before running memtest again I can tell you for certain that the RAM is running at 1.5V with CAS at 9-9-9-24. CPU Runs at 1.16 V.

-The bios has been updated Dec 2011, no updates have been released for that board by ASRock since then. I am running latest BIOS, I checked that today.

-I don't have a spare PSU.

I will run memtest again as per your suggestion, but RAM did not fail memtest before. I was still getting random reboots then.

I set RAM to run at 1333 instead of 1600. It now runs at CAS 8-8-8-20. Also Running Prime95. Temps at 104-109 deg F (40 - 43 deg C).
m
0
l
April 22, 2012 10:53:15 PM

Dude, replace your CPU, the further away from the purchase date, the less likely they are to take it back.
m
0
l
April 22, 2012 11:09:33 PM

teh_gerbil said:
Dude, replace your CPU, the further away from the purchase date, the less likely they are to take it back.


I will let it run till tomorrow and see what happens. Prime95 has run for 1 hour now without a hitch. Temps at 44 - 48 deg C (111 - 118 deg F).
m
0
l
April 23, 2012 12:09:13 AM

Those temps are 30 lower than mine!
m
0
l
April 23, 2012 12:53:40 AM

azeem40 said:
Those temps are 30 lower than mine!


I don't overclock, and I have plans to install a GPU at a later date, such as 6790 or 6850. I would like to see what temps I get then.
m
0
l
April 23, 2012 1:57:13 AM

Ok. So I stopped Prime95 after 3 hrs 55 mins of testing. No errors or warnings were reported. I will add another core and test tomorrow for 8+ hrs. I think I will keep the ram underclocked to avoid further problems. At least with this one RAM stick, it mostly means none of my components are defective.
m
0
l
April 24, 2012 10:44:56 AM

Installed another stick RAM. Prime95 ran 9 hrs 15 mins without a hitch. Temps not more than 50 deg C (122 deg F). Closing this thread. Thank you all for helping me out.
m
0
l
April 24, 2012 10:46:29 AM

Best answer selected by sanirudh.
m
0
l
!