Sign in with
Sign up | Sign in
Your question

[SOLVED] Failing Prime 95 test on new build?

Last response: in Systems
Share
January 11, 2010 12:17:37 PM

Hey guys, I'm having some trouble with a new build.

Mobo (P55M-GD45):
http://www.newegg.ca/Product/Product.aspx?Item=N82E1681...

CPU (Core i5 750):
http://www.newegg.ca/Product/Product.aspx?Item=N82E1681...

Ram (OCZ3P1333LV4GK):
http://www.newegg.ca/Product/Product.aspx?Item=N82E1682...

When it was first assembled I was getting a lot of BSODs so I checked the bios and found that memory was set to 7-7-7-18/1066/1.5V as opposed to the listed 7-7-7-20/1333/1.651V, so I adjusted the settings to the latter. At this point, I stopped getting bluescreens but the system is still failing on Prime 95 (two of the cores would fail in the mixed test within an hour, the other two would run fine for 6+ hours). I've tried using a different CPU (same model) but same problem. I've also repeated this on a different motherboard.

Anyone have any idea what may be wrong? Thanks :) 
January 11, 2010 12:22:35 PM

Adjust CPU and system settings in BIOS. If you have overclocked at all make sure power restrictions are off and that you have selected All Cores for overclock. It is probably just something you have overlooked in BIOS. Check the entire BIOS a few times over and if the problem persists troubleshoot your hardware.
m
0
l
January 11, 2010 8:28:49 PM

Ah, that's the thing, I'm not overclocking. This is the first time I had to manually adjust the timings and voltage for my RAM after assembling a machine (I've never overclocked). So far, I haven't touched any of the CPU settings :( . I just disabled spread spectrum to see if that'd help with the test. I want to avoid having to up the voltage (loving my temps so far).

There's one thing I'm not sure I understand, as far as I can tell, the CPU voltage seems to change depending on the CPU load. Then, when you adjust the CPU voltage in the bios, are you adjusting the maximum voltage or am I making no sense at all?
m
0
l
Related resources
January 11, 2010 9:41:42 PM

You're making sense. Some motherboards have AUTO settings for voltage control when hitting higher temps to avoid CPU/board damage. Look for voltage control and switch to manual.
m
0
l
January 11, 2010 10:30:17 PM

The i5 has turbo boost, where it ups the frequency (and voltage) on a single core when running single threaded apps. It also has a power saver and runs at reduced frequency and voltage when idle.
m
0
l
January 13, 2010 4:20:26 AM

First, thanks for the replies :) 

I just ran P95 all day. When I got home, one core failed at 9 hours and a second at 11 hours. The other two were still going after 12 hours. I also tried increasing the CPU voltage by 0.06V which made the test fail after a minute :p . I just upped the DRAM voltage (and brought the vcore back down) a bit ~1.7 1.658 and planning on running P95 again tomorrow. How high can I set the DRAM voltage to for it to be still 'safe'? Also, would I be correct to assume that, the odds of me running into problems are pretty low on regular <100% CPU use if I can run P95 for 9 hours without fail at 100%?
m
0
l
a b à CPUs
January 13, 2010 11:08:14 AM

Have you updated your BIOS yet? These boards went through a lot of changes in their first few months, including updates for stability and power regulation among others. Running on the latest BIOS will help rule out other causes.

Also, if you have not done so yet you should test your RAM with memtest86+. Let it run overnight. Longer if you can. Version 4 is the latest. You can get it here: http://www.memtest.org/
m
0
l
a b à CPUs
January 13, 2010 11:11:21 AM

cowsled said:
I just upped the DRAM voltage (and brought the vcore back down) a bit ~1.7 and planning on running P95 again tomorrow. How high can I set the DRAM voltage to for it to be still 'safe'?

You shouldn't run the RAM at more than 1.65V. That's the max for the 1156 socket CPUs. Is the same for the 1366 CPUs, too.
m
0
l
January 13, 2010 11:43:48 AM

Yes, I've updated the bios. I ran memtest before and after I adjusted the memory timings and voltage. Both ran for min 8 hours with no problems. Current voltage is set to 1.658 (not 1.70 doh), still too high? :( 
m
0
l
a b à CPUs
January 13, 2010 12:25:03 PM

I couldn't tell you if .008 is going to make that much of a difference. Maybe not. But let's see if someone else know.

Maybe you have a heat/air circulation problem. Download CoreTemp or RealTemp to run alongside P95. What do your temps run at?
m
0
l
a b à CPUs
January 13, 2010 1:28:23 PM

I agree with ekoostik, it certainly sounds like an overheating problem. The fact that it failed faster when you raised the CPU voltage seems to support this. We need to know what your temps are while running Prime95. A lot of people seem to have problems installing the HSF on Intel builds. I wouldn't be surprised if you have a push-pin that isn't fully seated.
m
0
l
January 13, 2010 8:12:41 PM

Actually, I have a mugen 2 installed over the CPU, load temps under 100% peak at about 48-50C (at 2.8Ghz). GPU Temp hovers around 47C. The ram feels reasonably cool to touch after 8 hours.

I contacted OCZ today to see if they'd have any suggestions for me. I was told to set my QPI to 1.3V, I assume that QPI is VTT? Running yet another torture test! XD.

[UPDATE]: I set the ram to 2T and everything's fine now! Thanks guys!
m
0
l
!