OCing the e6400, higher temps but stable

C

clayton006

[H]ard|Gawd
Joined
Jan 4, 2005
Messages
1,089
Hello all. This is my first attempt at overclocking and I think I'm doing well so far.

I'm using a SilverStone Nitrogon NT02 and some antec silver compound. (Yes I know it's not artic silver and I plan on picking some up very soon)

Well I have it installed, but I think I need to re-apply the thermal compound because I probably used too much. It seems that Artic silver's instructions differ from what antec says.

At 3.0GHz, my CPU idles at 52-55 using TAT and it's been running orthos right now for about 6+ hours and stays right at 70c. I know it should be cooler than that, but I think it's my compound that's not making a good connection. The fan is blowing fast, and in the right direction for the fans in my P180.

Any other suggestions? Oh and my voltage I think is at 1.31 or 1.32 on the CPU.

I guess my question is, I think I want to max out at 3.2GHz, is it ok to be running these temps like this... or should I be shooting to get this a bit cooler?
 
I think I screwed this up when I followed quack instructions to apply the thermal compound to both my HSF and my Core. I think I should follow the artic silver instructions on their site and do 1 line down the center and set her on top of that.
 
The paste shouldn't have that much to do with it. Sure overdoing it will hurt temps, but it shouldn't be as bad as poor contact. The heatsink you're using is pretty decent. I looked at some pictures. Is it a bolt down method instead of the regular push clips?

Maybe you don't have it bolted down tight enough. When you remove the heatsink, look at the distribution of the thermal paste to see where it made good contact and where it didn't.
 
I'll check it when I got home. Actually after the first round of temps, I took the motherboard out again and I tightened down the bolts even more... I'm just not sure that it's even.. I think that may be more of the problem...

It's a bolt down one.. so I may have to tighten this down like I do cylinder heads on my TransAm, in a sequential pattern (35% first time around, 65% next time around in a star pattern) :D

Should I be concerned at 70c at full load if I can't get these temps down further?
 
Are there any suggestions on a good way to make sure the bolt on setup for the HSF is put on correctly? Should I hold the top of it down and twist the bolts as I go? Any good method for this?
 
Well I just got home to take a look at orthos (I'm at the doc's office on my poclet pc now. And hard forum looks good on the pocket IE.)

anyway

The temps settled in at around 66 - 68 under load after 16 hours. Do you think it is worth me taking out the board and heatsink to see if I really did mount it wrong?

I probably should, but if it is running well then I don't want to mess around with it unless I need to.
 
Well I backed off the bolts on the HSF and I held it down the top with my hand as I tightened them all with a moderate bit of pressure. Now on Orthos my temps at full load range from 60c - 63c and idle at about 46c - 50c

I think I am happy with this for right now. When I graduate college next week I might play around with getting the thermal paste on there just right and securing the HSF.

Do you think these temps are reasonable under load?3

EDIT: Well it looks like the temps climbed back up into the 66c - 70c range. I may just be stuck here.
 
yep that high but you can live with it for now, for a little while,

It's a bolt down one.. so I may have to tighten this down like I do cylinder heads on my TransAm, in a sequential pattern (35% first time around, 65% next time around in a star pattern)

Should I be concerned at 70c at full load if I can't get these temps down further?
Click to expand...

yes even pressure is the name of the game. so HS sits as flat as possible, dont overdo it tho and break something.

It wont kill it but its high,

Check the flatness of the hs, stand a single edge razor blade on its sharp edge on the bottom of the heatsink, it should not have any gaps where you can see light thru where the edge of the blade meets the heatsink.


consider doing a Lap job on the heatsink and maybe the cpu. search forum for Lapping or google heatsink Lapping.
 
Have you tried running CoreTemp and what motherboard are you running? TAT does not seem to be reporting the proper core temperature on some boards. If you are using the Intel 965 chipset then either program will do the job.

I've found great stability up to about 70C to 72C but stability starts to decrease when you go over a CoreTemp reading of 75C and it's not a very good idea to be up at 80C like I was.

Surprisingly, my E6400 kept running at that extreme level.

I'm finding that I need approximately 1.376 v to 1.400 volts as reported by CPU-z or Speedfan while running Orthos at 3360 MHz.



Beyond this level the volts and heat go way up real fast with the Intel heatsink and fan.
 
My volts right now are at 1.33 for the vCore. I think I might use another program and this to see. I know the nvidia ntune isn't probably very reliable, but it was showing my CPU in the 58c range instead of 68c - 70c

I tried pressing in on my heatsink for a while and that didn't effect Idle or Load. I think I'm going to re-apply the paste though cause I did a crappy job on it the first time.

I'll try using Core Temp and compare the two. Any link to this? I'm sure I could find one, I'm just at work and didn't know if I could here.
 



This is with the processor being slightly underclocked (forgot the FSB) but these are the temps under full load with orthos.... which one should I trust in this picture? If Core Temp and Nvidia are right, then I'm running right where I should be....
 


Here is my temps with the processor at 3.0GHz.
The same voltages and clock speed ran fine for 20 hours on orthos a day ago. This difference between the programs is odd...?

EDIT: It's been running a little longer and the temps are still holding steady....
 
With the Intel 965 chipset CoreTemp and TAT are usually very close in the temperatures they report. On some boards though, I've heard users that are finding that TAT is reporting some numbers that seem too high. I did some testing and I'm quite sure that CoreTemp is the number you should believe. I'll post a link to a previous forum if I can still find it.

TAT has not been updated for quite a while and was never designed specifically for the Core 2 Duo anyhow.

You're also using an aftermarket cooler which should be keeping your temps lower than if you were using the Intel heatsink and fan. What size fan and speed are you running? What's your room temperature there and how's your case for ventilation?

At stock CPU volts, even up to 3000 MHz and a little beyond, the C2D really doesn't put out that much heat. You usually don't start to see big temp number until the volts start to go up.
 
I set my E6400 to 3000 MHz ( 375 X 8 ) for some comparable numbers.



On my P5B Deluxe, CoreTemp is usually a couple of degrees higher than TAT.

For voltage CPU-z was reporting 1.328 volts at no load and 1.296 volts while running Orthos. I'm using the Intel heatsink and fan.

Room temperature here is 17C so you'll probably need to add about 4C to my numbers to get to normal room temperature of 21C ( 70F ). Idle temp is showing 26C.

Speedfan v.4.32 was just released and it reports the exact same core temperatures as CoreTemp does. I find it hard to believe that two different programmers following the Intel guide lines wrote two totally different programs and that both programs are wrong. It's much more likely that they are both right and TAT is wrong.

On a well clocked P4 your Silverstone Nitrogen NT02 cooler dropped temps by 6.5C at full load so keep that in mind when comparing.
http://www.pcmoddingmy.com/e107_plugins/content/content.php?content.221.4

Do all the math and your CPU running at 50C under load makes more sense than it running at 63C to 65C as TAT is reporting.
 
Well that is reassuring.
Thank you for your quick reply. My ambient temp is about 73f.
My fan is on full blast (and kinda loud.... but I could care less) :D

I keep it at 3GHz for a few days and see how it pans out. I forgot to mention that I'm running the eVga i680 SLI board.

Thanks again
 
The E6400 is very happy running at 3200 MHz ( 400 X 8 ) with default voltage. With DDR2-800 memory it makes for a real simple overclock.

Turn your fan down a notch and save your ears. The C2D can run 24/7 at 70C so you have lots of headroom to play with.
 
I'll see how much I can deal with and go from there. I still wonder a little bit why TAT is so far gone.. but.. I'll trust Core Temp for now since it makes the most sense.

When I looked at the HSF and the Core, I put a blade up to it and it looked pretty even, if anything it was VERY minor, and nothing like other posts I have seen.
 
Well funny thing...

My Windows Vista will lock up after about 4 hours when I'm not at the computer.
When I was running only the CPU stress test (and not the RAM and CPU) in orthos no problems, but when I ran the mixed test, after about 4 hours the computer would reset itself and I would be prompted at the windows login screen.

I then ran CS:Source last night with a 16 man bot server (locally) and I checked the event viewer and I saw that at around 3AM the computer restarted. (I started it at about 11).

My ram voltage is at 1.85 and I think it's supposed to be 1.9 according to Corsair's site. The ram is unlinked from the FSB and its at stock timinigs. My FSB is set to 1.5v and my proc at 1.33v. I know it's not overheating in either TAT or CoreTemp (I have it logged, no problems there). Do I need to up some volts on this thing?
 


These are some shots of my case.... I really don't have a good intake fan in my p180 so maybe I should get something put in there to cool down the ram? The front fan isn't really pulling much air even with the front of the case open. I think I'm going to get a higher CFM fan today after work.
 
The #1 problem people have is fear of giving their DDR2 the voltage that it needs.

All DDR2 memory will run comfortably at 2.15 volts to 2.25 volts without turning cherry red or reducing its lifespan one bit.

Check out this review of how memory scales with voltage.
http://eclipseoc.com/index.php?id=1,72,0,0,1,0

It continues to go faster with tighter timings well beyond the conservative voltages I recommend.

Here's a good review of some top of the line Corsair memory:
http://www.anandtech.com/memory/showdoc.aspx?i=2900

The Micron memory chips that Corsair uses for these were originally rated at DDR2-667, CL5 with 1.8 volts.
It's amazing how they perform at 2.4 volts and Corsair even warranties them to 2.5 volts.
 
Well mine are stock 800 chips (maybe they OC'ed them originally).
So I'll go home and check it out and raise the voltage.
 
The Micron chips used on my DDR2-800 Team Xtreem ram are part number D9GMH but I've run them at DDR2-1066 speed at CL5 with no problems.

Micron lists these chips as DDR2-667, CL5, 1.8 volts but most high performance memory manufacturers are happy to use these chips on their DDR2-800 or DDR2-1000 rated memory. With some extra voltage the best of these chips can run at whatever speed you like.

The promos chips that many other manufacturers use also like some additional voltage. I don't believe that they can take the high voltages that the D9 chips can take but they work fine at 2.15 volts to 2.25 volts.
 
Well I've had my RAM up to 2.0 any higher and Vista won't load.

I've ran Orthos twice now... and about right to the hour the workstation locks up.

My vCore is at 1.36 now and FSB at 1.5 and I've also upped the MCP and SPP an extra .1v and I'll see how this goes. If this doesn't work I'm going to install XP on another drive and see if it does it there too......

I wonder what can be causing this hard lock.... my temps in Windows seem to be ok (both in TAT and Core Temp). But one thing I did notice, in my BIOS the MCP temp is like 57c? Should I be worried about that?
 
Just wanted to add a bit about temperature checking programs. unclewebb is right and TAT could be reporting wrong temps with certain boards. Mine does and it's consistenly 12-13C higher than CoreTemp. SpeedFan 4.32 agree with CoreTemp and physically, the heatsink is not very hot.

I even got further as pushing voltage to 1.50 and trying to run at 3.5 GHz for my poor E4300 and it hovered around 82-83C in TAT, which is in the area where throttling is supposed to occur. I didn't notice any throttling and the 2 other is reporting 68C consistenly, well within the throttling threshold. This prove partially the fact that TAT is wrong. I also observed the reports around and I often see 2 things : Either TAT is 2-3 degrees close to others or always 12-13 degrees so it may be adjusting the offset as well, sometime wrongfully on certain boards. To hit the nail in the TAT coffin, it's also a old program not updated in a long time and the official purpose was to be used to test laptops with Yonah chips to see if a cooling design works fine or not during prototyping. It's even considered confidential if you look in the TAT docs.
 
Now more or less I just need to get this running stable... do you think I need to up the vcore a bit?

UPDATE: Well now it's past the first hour... we'll have to see how it works out...
 
Well this morning I turned the computer on and it appears that it got to about 7 hours and then locked up. I upped the vCore again to reflect 1.4v in the BIOS. Is this what I should be doing? The memory is already at 2.0v (any higher and I can't boot). FSB and other voltages are posted in my upper posts.

The heat seems to check out well within spec.
 
On my P5B Deluxe setting it to 1.45 volts for the CPU gives me 1.392 volts while running Orthos. Anything less and I lose stability.

Every board is a little different but CPU volts while fully loaded is the important number to go by. For higher voltages and core temps rely on SpeedFan v.4.32.

 
Update:

Well I got my CPU running stable at 3.2GHz at 1.4v (1.386 in CPU-Z). I had to also up the RAM to 2.0 and the MCP and SPP to a value I can't remember.
The Rig has been running strong for a few days now.
 
You must log in or register to reply here.
Back
Top