Prime95 running too hot?

[ekoostik]

I'm running Prime95 on a new build - an i7 860 that is running at stock speeds with the stock cooler. SpeedFan is reporting the cores are running at ~62-65 C. Real Temp 3.0 is reporting the cores running at ~77-80 C. Which is correct, and more importantly is it running too hot??

The room the PC is in is about 22 C / 72 F.

 

[ekoostik]

Well the "good" new is that the temperatures have not gotten worse over the last hour. Holding at about 78-80C Real Temp and 62-65 SpeedFan.

In looking at the speedfan temps, I just noticed some of the power numbers are not what I expected. Granted I don't know what these numbers represent.

The following is being reported, I bolded the ones that look suspect (12V - is that a problem??):
Vcore1: 1.10V
Vcore2: 1.54V
+3.3V: 3.30V
+5V: 4.97V
+12V: 16.32V
-12V: -0.92V
-5V: -7.37V
+5V: 3.63V
Vbat: 3.34V

 

[Blazer]

those psu numbers look horible, list sys specs.

 

[ekoostik]

My system, no OC:
i7 860
Gigabyte GA-P55M-UD2
Corsair CMPSU-450VX
Gigabyte Radeon HD 4670 1GB (GV-R467ZL-1GI)
G.Skill Ripjaws 4GB(2x2) 9-9-9-24 1600 (F3-12800CL9D-4GBRL)
WD Caviar Black 640GB
Sony Optiarc Black 24x
Cooler Master Elite 310 (RC-310-BWN1-GP) - 1 rear 120mm exhaust, I added 1 front 120mm intake, has one side vent where an 80/90/120mm fan could be added (intake or exhaust?)
Dell UltraSharp 2208WFP (1860x1050)
Windows 7 RC x64

I also checked the PSU numbers with cpuid's hardware monitor. It didn't report a +12V, which was one of the oddest of SpeedFan. HWMonitor reported:
Vcore: 1.10V
Vin1: 1.54V
+3.3V: 3.30V
+5V: 4.97V
-12V: -12.48V
-5V: -1.79V
+5V VCCH: 3.63V
Vbat: 3.34V

EDIT: Updated sig with my specs

 

[Blazer]

the specs on NE for that psu look good, and its a single rail which is good, having used speedfan many times myself i would tend to believe speedfan for the temps, the voltage that SF reported could be incorrect, maybe fron the reporting capabilites of the i7, did you use the stock heatsink thermal pad, if so it may take a while to burn-in and lower temps, try cpu burn and max out, several runs may be needed to see a temp drop, although artic silver would be better it would still need burn-in time.

 

[ekoostik]

I did use the stock heatsink fan. Interesting, I hadn't heard of burn in time to let the thermal paste set, but it makes sense. I did run the Intel Burn Test, but only for about 3 passes. I think 1 pass of 10 or 20 on High was the most strenuous with the other two runs probably 10 Standard. That's what you meant by cpu burn right? The Intel Burn Test? I'll run it a few more times later today.

 

[Blazer]

get cpu burn here http://majorgeeks.com/CPU_Burn-in_d3921.html this site can be trusted.

in this order

1st start task manager/performance/to watch cpu is 100%
2nd start speed fan to watch temps, shut down if over 70C
3rd start cpu burn and run for 30 minutes at a time and let completely cool before next run

get cpu hot and let cool, again, again, again, again, this can be done at random whenever you want but the sooner you do the better the set of the thermal compound.

5 runs should be good enough, if temps go over 70C recheck that cpu is firmly seated and that push pins for the heatsink latch are truely locked down.

 

[ekoostik]

blazer- the site says it's for x86. (and it's from 2003.) Is it safe to run on Windows 7 x64?

 

[Blazer]

nope that one wont work, sorry !

here is one i just tried, it works, with speedfan also, tried it on my Q6700, leave at default for 100 % load.

http://majorgeeks.com/CPU_Stability_Test_d215.html

 

[Old Hippie]

SpeedFan is reporting the cores are running at ~62-65 C. Real Temp 3.0 is reporting the cores running at ~77-80 C

Are those idle temps?

If so, that's really warm!

Speed Fan 4.39 reads my 920 cores @ 10C cooler than what they are.

The CPU temp seems close but the core temps are way off.

Try CoreTemp and see what it says.

 

[ekoostik]

Originally posted by: Old Hippie

...
Are those idle temps?

If so, that's really warm!

Speed Fan 4.39 reads my 920 cores @ 10C cooler than what they are.

The CPU temp seems close but the core temps are way off.

Try CoreTemp and see what it says.

No those were under load. But I also discovered that SpeedFan was reporting about ~15C LESS than the real temp. I tried CoreTemp as you suggested. At idle CoreTemp, RealTemp and SpeedFan (with offset +15C) all report around 41-43C.

When running Prime95 they get awful close to 80C.

Ambient temperature is between 20-22C. Typically pretty close to 21C.

 

[ekoostik]

Originally posted by: blazer
nope that one wont work, sorry !

here is one i just tried, it works, with speedfan also, tried it on my Q6700, leave at default for 100 % load.

http://majorgeeks.com/CPU_Stability_Test_d215.html

After I posted questioning whether or not the CPU Burn-in would work I ran the Intel Burn Test. I had run it a couple times before without any problems. So I didn't expect anything out of the ordinary. I ran it at 5x, 8 Threads, on High. Unfortunately, on the last test it failed. Here's a link to a screen shot of the failed run:

http://s867.photobucket.com/al...IBTFailure20091011.png


I'll try your CPU stability test now. But I'm starting to get worried. Since I'm not OCing or doing anything extravagant what started as a conern about temperatures seems to be revealing other issues. I'm not sure what to do about the failed Intel Burn Test, or really what that even tells me.

 

[Blazer]

i would look very carefully at the latch pin's where they insert into the mobo [ not the push pin's ] but what the push pin's insert into, they should be a clear color and the push pin should be black, the clear latch'es need to be fully in the mobo holes.

if incorrectly latched in place the heatsink will not be in full contact with the cpu and cause your exact problem, its one of the reason i perfer an aftermarket heatsink used with a support retainer on the otherside of the cpu socket.

 

[ekoostik]

Well I've run CPU Stability for almost 90 minutes now and no problems yet. It doesn't seem to be taxing the system quite as hard as Prime95 though as temperatures are staying between 57-62 for the most part. (In my first post I was concerned when Prime95 was pushing the temps to 62-65 ala SpeedFan but those were 15C too low, so Prime95 had been pushing temps to ~80C.

I forgot to metion in my last post that when IBT failed, I also discovered that in CPU-Z the CAS# Latency was incorrectly being reported as 4.0 clocks instead of 9.0 clocks. It was a long night last night.

Well I recently found an entry in the System log with Source = HAL stating:
The platform firmware has corrupted memory across the previous system power transition. Please check for updated firmware for your system.

Last night I thought maybe the bad CPU-Z CAS# being reported was a result of the IBT fail. I'm not positive about the timing, but I'm now thinking the IBT fail may have been because of the bad CAS#. Some quick googling has turned up an explanation of the problem being the pc not waking up correctly from a sleep state. If this really is the cause that would be great because it means I can go back to focusing solely on my temperature problem.

blazer - I will pull the motherboard out in a little bit and double check the latch pins.

Thanks again to all for the help.

 

[ekoostik]

And wouldn't you know it. Just after my last post at right about the 90 minute mark CPU Stability crashed. Shoot.

It left me 2 messages in the Application Log. I'm not sure either are very helpful:

1.) Source: Application Error
Faulting application name: CPUStabTest.exe, version: 6.0.0.154, time stamp: 0x2a425e19
Faulting module name: unknown, version: 0.0.0.0, time stamp: 0x00000000
Exception code: 0x00000000
Fault offset: 0x00000000
Faulting process id: 0xf78
Faulting application start time: 0x01ca4aadfbf0b5c8
Faulting application path: C:\Users\ekoostik\Downloads\CPU Stability Test\archive contents\CPUStabTest.exe
Faulting module path: unknown
Report Id: 9f3b4270-b6ad-11de-8cf6-00241dd3b260

2.) Source: Windows Error Reporting
Fault bucket 1290129360, type 1
Event Name: APPCRASH
Response: Not available
Cab Id: 0

Problem signature:
P1: CPUStabTest.exe
P2: 6.0.0.154
P3: 2a425e19
P4: StackHash_9e13
P5: 0.0.0.0
P6: 00000000
P7: 00000000
P8: 00000000
P9:
P10:

Attached files:
C:\Users\ekoostik\AppData\Local\Temp\WER6B5.tmp.WERInternalMetadata.xml

These files may be available here:
C:\Users\ekoostik\AppData\Local\Microsoft\Windows\WER\ReportArchive\AppCrash_CPUStabTest.exe_e34aaa232fd73baf9f6335e05037a87980b62d_ed0ed605

Analysis symbol:
Rechecking for solution: 0
Report Id: 9f3b4270-b6ad-11de-8cf6-00241dd3b260
Report Status: 0

 

[Old Hippie]

Just after my last post at right about the 90 minute mark CPU Stability crashed

Some guys get all bent outta shape about any testing failures but you will probably never stress the CPU anywhere near what these programs do.

In other words, I wouldn't worry about it.

Now that you've got the temps readings correct, I'd get rid of the programs that aren't needed. Pick the one that suits your fancy and let the rest go.

Although your temps are too high for me, they are within the acceptable range according to Intel.

It's pretty well known that the stock cooling systems for these new chips are barely enough to keep within the acceptable range.

You can check all your installation work like blazer mentioned but even if it's perfect it probably won't bring it to what I consider acceptable cooling.

I like my CPU to idle under 40c and stressed under 70c.

If you want temps like mine ideals you're gonna hafta go the extra mile and invest some money into after market cooling.

I went with Corsair H50 and a 38mm Delta PMW fan for my 920 @ 3.2 overclock and get @ 35c at idle and 60c when stressed.

Just about right.

 

[Blazer]

crash after 90 minutes, i am with old hippie, you should invest in a aftermarket heatsink and fan, this one is a fantastic cooler and comes with a backplate as all heatsinks should and also has a very quite fan regulated by the cpu/mobo temp sensor.

http://www.newegg.com/Product/...x?Item=N82E16835103046

 

[ekoostik]

Ok, I'm sold. I'll get some thermal paste and an aftermarket HSF. Do you think the temperature is what is causing my failure in the stress tests? Will an HSF make the problems go away?

Edit: BTW, the CM HSF you linked to doesn't list compatibility with 1156. Any recommendations for the new socket?

 

[Blazer]

not alot of choices, i like quite so try and keep the dba on the low side, also i would assemble and fit before appling artic silver, will require mobo removal to assemble with the backing plate.

http://www.newegg.com/Product/...156+heatsink&x=15&y=44