Sign in with
Sign up | Sign in
Your question
Solved

Is Prime95 running too hot?

Last response: in Systems
Share
October 10, 2009 12:52:32 AM

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.

More about : prime95 running hot

October 10, 2009 1:55:47 AM

Wow that's a lot of information. Going to have to read it a couple more times. 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.

I believe the HSF is on tight. The first time I put it on I did not get the push pins all the way through. On the second try all four looked ok. I do not remember there being a plastic cover on the bottom of the HSF. I remember there being two or three grey stripes / 'blobs'. I assumed it was some kind of compound that would spread when the CPU heated up. Was that a bad assumption?

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
m
0
l
Related resources
a b B Homebuilt system
October 10, 2009 2:01:07 AM

Never used speedfan, and not sure its reliable. That's the one that comes with the mobo software?

I use HWmonitor. Might want to load that up and see if 2 of the 3 match.
http://www.cpuid.com/hwmonitor.php

As for the goo, yes, its suppose to be there. But there would have been a plastic shell on there to protect the goo from smearing during shipping. That for sure would have had to come off.
m
0
l
October 10, 2009 2:42:13 AM

No, I had to download SpeedFan. It's one I'd read about in other places. It's also recommended in the previous link you gave me.

Regarding the cover - I would have noticed if it was on right? Since I saw the goo, that means I took the cover off? I know I checked for a cover. But now I'm second guessing myself.

I downloaded HWMonitor and am running it now too. It's temps are more inline with Real Temps. It's reporting out 74-77 C. Two hours at temps just under or at 80C if HWMonitor and Real Temp have it right. It's holding steady. But should I stop it? Can that do damage to my machine?

Interestingly the voltages are different than Speed Fan's. The measures, not just the results. There's no +12V with HWMonitor, which was the one concerning me the most with SpeedFan. ???
HWMonitor reports:
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


Thanks.
m
0
l
a b B Homebuilt system
October 10, 2009 2:49:55 AM

Speedfan is a very good program. It's reliable.

What kind of case do you have?

Try making sure all your cables are tied up nicely for good airflow.
m
0
l
October 10, 2009 3:04:25 AM

I have speedfan saying ~64C, hwmonitor ~77C, and real temp at 80C.

My case is a CM Elite 310. 120mm fan out take in the back, I added another for intake upfront (the HDD's reporting 32C). Vent in the side where I could add another 80/90mm fan, I'd seen intake recommended there but I feel air moving out of it. Or maybe that's why I need an intake fan there? To keep cooler air from leaking out the side before it passed over the processor? Hm, if that's the case I'm tempted to block that vent and see what happens.

The air coming out that back fan isn't too warm - comparatively speaking the air coming out of the PSU which is above the back fan at the top of the case is hotter, and definitely warm.
m
0
l
a b B Homebuilt system
October 10, 2009 3:10:33 AM

I think the airflow is your problem. With only the single 120mm fan in the back, there isn't much pushing the air out the case. That's a relatively small case.

Try removing the side and see what happens to your temps.
m
0
l
a b B Homebuilt system
October 10, 2009 3:28:59 AM

aford10 said:
Speedfan is a very good program. It's reliable.

What kind of case do you have?

Try making sure all your cables are tied up nicely for good airflow.

It was PC Probe II and fan xpert from Asus I was confusing and thinking of that isn't reliable, or so I've heard. I'm a hwmonitor or realtemp user.

Next time you reboot, see what the bios lists your voltages at.
m
0
l
October 10, 2009 4:42:15 AM

I've let it run with the side of the case off for 45minutes. The temperatures dropped a little in the first 5-10 mins, but have been holding fairly steady since. The new core temps are:
SpeedFan ~62C; HWMonitor ~75C; RealTemp ~77C.

Previously they were:
SpeedFan ~64C; HWMonitor ~77C; and RealTemp at 80C


Should I be concerned about the other temperatures listed in SpeedFan? One is high. I haven't been able to find information online about what these temp's represent; these are with the case off:
Temp1: 33C
Temp2: 25C
Temp3: 66C

HWMonitor reports 2 additional temperatures: GPU Core 31C; HDD 32C
I'm also running GPU-Z and it reports GPU Temperature ~30C


I'll look at the BIOS for voltages next time I reboot. If it's relevant, I've using the Gigabyte GA-P55M-UD2 motherboard and a Corsair CMPSU-450VX PSU. I'm also only running an HD 4670 GPU.

I started going through CompuTronix's Core i7 and Core 2 Temperature Guide again, I have a bunch of notes but not much progress. Of course it hasn't been updated yet for i5/i7 8xx's so I'd have to pick one of the other chip's thresholds to follow.
m
0
l
a b B Homebuilt system
October 10, 2009 4:55:55 AM

Speedfan picks up those temps from your cha sensors on your board. You can also use speedfan to change the fan speeds. If any aren't at 100%, you may want to increase them.

Have you OC'd at all?
Does your HSF feel loose at all?
m
0
l
a b B Homebuilt system
October 10, 2009 5:02:20 AM

Use a different thermal paste? When I got my new CPU I removed the stock compound and used Arctic Silver 5 instead. I don't know if that's the best out there, but I do know it's pretty good.
m
0
l
a b B Homebuilt system
October 10, 2009 5:05:08 AM

Artic silver is good stuff. And applying new thermal paste is a good suggestion. But since those push pin HSF's are such a pain in the pooper to put back on once you remove them, I would try the other ideas first.
m
0
l
October 10, 2009 5:07:19 AM

The fans are running at 100%. The ones being monitored anyway. The CPU fan is running at ~2180RPM. The front case fan is running at 1125RPM. That's all the fan headers the motherboard has so the back fan is connected directly to the PSU.

I haven't OC'd. I did make a change for the memory. I turned XMP on so that my RAM would be picked up at their 1600 CAS9 settings instead of the 1333 settings the motherboard defaulted them to.

HSF feels secure. Granted I can do a better check when I shut down later.
m
0
l
October 10, 2009 5:22:50 AM

Looks like I replied too slow, missed a couple responses. Thanks for all the feedback and suggestions.

I'd been considering replacing the stock heat sink but was waiting for more 1156 models / reviews. If I can avoid it I'd really rather not take off the stock fan to apply some paste only to replace it again later. I'm not completely opposed to it. I just think I'd like to try other options first, if I have any.

Sounds like suggestions so far are:
Add a case fan
Use some thermal paste
Replace the stock cooler

Any others?

Starting with the easiest, looks like I'll be adding a case fan. Do you think the side fan should be an intake, or exhaust? The vent is low enough I think as an exhaust it may push all the cool air coming in from the front out before it can flow over the CPU. Also, my GPU has a fan already pushing air down away from the CPU which means some of it likely flows out the side vent anyway. So would an intake be better because it would force more air over the CPU?
Cooler Master 310 'specs' from NewEgg
Side profile picture
m
0
l
a b B Homebuilt system
October 10, 2009 5:23:08 AM

If it feels secure, then it probably is on fine. You can be sure by unmounting the board and looking at the back. If you can see all 4 prongs of the HSF, it's on.

In that case, you should apply new thermal paste.
m
0
l
October 10, 2009 5:41:04 AM

Boy when I stopped Prime95, the temperature dropped 20C in seconds. Within a minute or two it reached a steady 28C (SpeedFan); 43C HWMonitor; and 44C RealTemp. I don't know if this is quite idle yet but I'm not going to watch it any longer tonight.

That was a good 5 hour run. Glad to see the temps didn't peak past 80. But I've got some work to do now to get 'em lower.

Thanks again for all the help so far. Any other thoughts, please let me know.

EDIT: To answer the last question, the 4 prongs are sticking out the back of the motherboard. I mentioned earlier that I had trouble putting the HSF on the first time. I knew it wasn't on right b/c the prongs weren't sticking out the back. On my second try they felt better going in (b/c they actually clicked in b/c I pushed hard enough) and I checked and the prongs were sticking out the back of the board.
m
0
l

Best solution

October 10, 2009 4:09:03 PM

ekoostik,

Sorry you're having trouble getting tweaked up. From Intel's Processor Spec Finder - http://processorfinder.intel.com/details.aspx?sSpec=SLB...

All Core i7 8xx variants:

Vcore Max 1.400
Tcase Max (CPU temperature) 73c
Tjunction (Core temperature) 78c

Also, SpeedFan's Core Temperatures could be low by 15c. I assume you're running 4.39? Whenever SpeedFan reads a new processor for which it's not yet coded, it defaults to Tjunction Max 85c, which makes it read 15c too low. This should be corrected in 4.40. I'll bet if you add 15c to SpeedFan's Core temperatures, they'll be equal with Real Temp. It's just a simple offset calibration.

From the Core i7 and Core 2 Temperature Guide - http://www.tomshardware.com/forum/221745-29-sticky-core...


Section 11: Offsets

SpeedFan can be configured to correct for inaccurate Tcase (CPU or Temp x) and Tjunction (Core x).

(A) From the Readings tab, click on the Configure button, then click on the Advanced tab, and click on the Chip field, directly under the tabs.

(B) Next, go to SpeedFan's installation Program Group, and click on the Help and HOW-TO Icon. This help file can also be found by searching for the filename speedfan.chm.

(C) Under Contents, click on How to configure, then click on How to set Advanced Options. Read this section, including Other interesting options, with emphasis on Temperature x offset.

(D) If additional help is needed, click on the following link to SpeedFan's homepage, then click on the Support, Articles, Screenshots and F.A.Q. tabs: http://www.almico.com/speedfan.php

When CPU and Core Offsets have been completed, SpeedFan will be accurate. SpeedFan is also extremely useful for observing temperatures and Vcore using the Charts tab, while thermal benchmarking with Prime95 Small FFT's.

Tips:

(A) Tcase may be labeled as CPU, Temp 1, Temp 2 or Temp 3, but is most frequently labeled as Temp 2. Follow Section 8, Note 2 to correctly identify which label corresponds to Tcase.

(B) Tjunction is labeled Core 0, Core 1, etc.

(C) Graphics Processors are labeled Core.

(D) Graphics Cards which display a sensor labeled Ambient, must not be used for measuring room temperature.

(E) SpeedFan flame Icons are alarm limits which can be adjusted to Warm Scale using the Configure button.

(F) SpeedFan Aux 127 is an unassigned input which can be disabled using the Configure button.

(G) Core 0 typically carries heavier loads and higher temps during single threaded gaming and applications, so SpeedFan should be configured to "Show in Tray" Core 0.


Comp :sol: 
Share
October 11, 2009 5:09:38 AM

CompuTronix - thanks for the information. While it took me a while to absorb - and I know I don't grasp it all yet - it has been helpful. I like learning and knowing what I'm doing so I appreciate you pointing me in the right direction. I bumped the offsets for SpeedFan's Core readings up 15. They are now in line with Real Temp's. I also was able to identify the Tcase - it was being reported as Temp3.

Two questions I have not been able to figure out yet: First, what is Temp1 and Temp2? They seem to be fairly consistently around 28C and 25C, respectively.

Second, and this relates to Tcase and Tjunction - With the Tjunctions at the 'correct' setting (offset +15), the Tcase is consistently ~10C (+/-1) less than the Tjunction (Core0-7) readings. Based on the information in your guide I expected this difference to only be ~5C. Does the 10C difference mean there's a problem?

The idle temperatures seem to be: 21.5C ambient, 32C Tcase, and 42C Tjunction. (Remember the Tcase has no offset applied, the Tjunctions have +15). CPU-Z is reporting a Core Voltage of 1.152V.

And one final note, I just ran an Intel Burn Test. 5x High, 8 threads. And it failed. While I was concerned before, now I'm starting to get worried. Here's a screenshot:



EDIT: Took a couple tries to get the picture loaded.
m
0
l
October 11, 2009 6:10:08 AM

Thought I should add a few more details. While posting my previous response my monitor went black, then flickered slowly, never coming fully back just moving slowly between all black and showing me some things on the screen. The mouse cursor was about 5x the normal size and it didn't really move where I told it to/wasn't very responsive either. I eventually had to do a hard reboot.

However, after the IBT failed while I was preparing my post and before the monitor freaked out on me I noticed something odd in CPU-Z. On the memory tab the timings were being reported as 4.0 - 9 - 9 - 24. Normally they are 9.0 - 9 - 9 - 24 (which is what they are now after the reboot). The CAS Latency was way off!?

Since I had to reboot I went into the BIOS to check some other items. Skora had previously asked me to check my voltages. What I found in there was:
Under PC Health Status
Vcore: 1.172V *On the main page Vcore was 1.152. I bounced back and forth between the 2 and there was always this minor discrepancy
DDR15V: 1.536V
+5V: 4.999V
+12V: 12.302V
Current System Temperature: 29C
Current CPU Temperature: 37C ------ Are these the Tjunction and Tcase measurements? They are off by 8C here, not the 10C I saw w/ SpeedFan but more than the 5C I was expecting.

Under Mother Board Voltage Control
Load Line Calibration DISABLED
Everything else on this page was set to AUTO. This includes: "CPU VCore 1.18750V", "QPI/Vtt Voltage 1.100V", "PCH Core 1.050V", "CPU PLL 1.800V", "DRAM Voltage 1.5000", etc.
m
0
l
October 11, 2009 6:58:24 AM

ekoostik said:
... what is Temp1 and Temp2? They seem to be fairly consistently around 28C and 25C, respectively.
Could be southbridge chipset and motherboard. Please consult SpeedFan's Help and HOW-TO icon, and SpeedFan's homepage. Click on Support, Articles, Screenshots and F.A.Q. tabs: http://www.almico.com/speedfan.php
ekoostik said:
... Tcase is consistently ~10C (+/-1) less than the Tjunction (Core0-7) readings. Based on the information in your guide I expected this difference to only be ~5C. Does the 10C difference mean there's a problem?
Never expect default temperatures to be accurate. It simply means that BIOS is mis-coded, since programmers sometimes don't receive accurate Tcase thermal data. Calibrate Temp3 offset.
ekoostik said:
... Intel Burn Test... failed...
Increase Vcore and try again.
m
0
l
October 11, 2009 7:35:30 AM

ekoostik said:
While posting my previous response my monitor went black, then flickered slowly, never coming fully back just moving slowly between all black and showing me some things on the screen. The mouse cursor was about 5x the normal size and it didn't really move where I told it to/wasn't very responsive either. I eventually had to do a hard reboot.

However, after the IBT failed while I was preparing my post and before the monitor freaked out on me I noticed something odd in CPU-Z. On the memory tab the timings were being reported as 4.0 - 9 - 9 - 24.
Nothing unusual here. Anything can happen during an instability crash.
ekoostik said:
Vcore: 1.172V *On the main page Vcore was 1.152. I bounced back and forth between the 2 and there was always this minor discrepancy...

Load Line Calibration DISABLED
If you read the Overclocking Guides, then you know that Load Line Calibration should be enabled, since the objective is to minimize Vcore flutuatations, especially between idle and load, which is called Vdroop.
ekoostik said:
Everything else on this page was set to AUTO. This includes: "CPU VCore 1.18750V", "QPI/Vtt Voltage 1.100V", "PCH Core 1.050V", "CPU PLL 1.800V", "DRAM Voltage 1.5000", etc...
If you read the Overclocking Guides, then you know that auto settings are highly discouraged, especially Vcore, since auto will apply excessive voltage, which in turn increases temperatures.
ekoostik said:
Current System Temperature: 29C
Current CPU Temperature: 37C ------ Are these the Tjunction and Tcase measurements? They are off by 8C here, not the 10C I saw w/ SpeedFan but more than the 5C I was expecting.
Not quite. BIOS doesn't read Tjunction, otherwise you'd see 4 Core temperatures. While "Current CPU Temperature" is Tcase, "Current System Temperature" typically refers to the motherboard.

Comp :sol: 
m
0
l
October 13, 2009 8:49:45 PM

aford10 said:
If it feels secure, then it probably is on fine. You can be sure by unmounting the board and looking at the back. If you can see all 4 prongs of the HSF, it's on.


I reseated a couple of the pins that looked questionable - though they all looked like they were in. I took a couple out completely and pushed them back through and pushed the others a bit just to be sure. I also moved the front 120mm intake over to the side. Temperatures are idling a degree or two lower, around 41C. Last night I ran Prime95 for 2.25 hours. After the first 15 mins it ran at a fairly steady 74 +/-1 C, though the cores eached peaked at around 77C. That's about 5C cooler than I was getting before. Kind of wish I'd only made one change at a time (push pins, then moved fan) so I could see what was responsible for the change. But I'd still like to do better so I'll be working on it some more. I'm going to pick up an aftermarket heatsink. Just need to find some time to read reviews and get that done.

I've also passed the machine through a few more Intel Burn Tests. At 5x, Max memory, and Auto thread it has run fine. I've also done 10x, Standard, Auto. My previous fails have been when I set the threads to 8 and I admit I've been afraid to re-run that test just yet, since the last fail.

CompuTronix said:
Increase Vcore and try again.


I was hoping to get this stable at stock before messing with these kind of settings in the BIOS. But if that's what I need to do then it looks like I have a date with some BIOS manuals and overclocking guides. Although it will still be minor changes I make until I feel comfortable with what I'm doing. In the meantime I've printed out your Sticky: Core i7 and Core 2 Temperature Guide (20 pages, even with a lot of white space deleted!) as I want to make sure I have accurate temperature readings and understand their impact before I move on to the next level.
m
0
l
January 27, 2010 10:40:10 AM

Best answer selected by ekoostik.
m
0
l
!