Need some help trying to understand a temp change, please.

M

Midian_Falls

n00b
Joined
May 14, 2007
Messages
38
Gentlemen (and ladies if present),

I seem to be having an issue with my core temps as of late, and I'm not entirely sure just what the heck is going on. On a prior OS build, under 100% Orthos load, with the same Bios settings I never peaked above 37 degress C. No, I'm not joking, I watched my temps like a hawk during the initial OC stage and full burn over 8 hours (heh at work :p ) and I never saw above 37C.

I made some additions to the box, 3 more HDDs in a 5-pack mini-rack at the top of the case (w/ 60 or 80mm fan, can't remember which offhand) and I ran the 5 Sata cables under the motherboard. The cables run the full length of the board to the bottom of the box where the Sata connections are. I realize the additional HDDs up at the top of the case would cause *some* increased temps, not sure about the cabling to be honest.

Last night I did another Orthos run for giggles and was shocked to see my temps in the 45-46C range. WTH? I gave a couple degrees for ambient temp, even downloaded Speedfan/Core Temp to verify what I was seeing in NVMonitor. I checked the HSF, double checked, even made a make-shift air-duct from the back of the HDD rack with a CD case angling the air up to the 200mm fan at the top of the case, no changes. NVMonitor reported about normal System & GPU temps.

I left Orthos running over night, just shy of 8 hours when I got up this morning and it was a steady 43C under load. I realize that isn't a bad temp, but dammit I had a lower one previously and I'm just trying to make some sense out of it.

Reverse the HDD Rack fan? Remove the cabling under the board? Suggestions, please...

--MF
 
You're running a Scythe Mine heatsink on a 3.4GHz overclock and claiming 37C on Orthos load?

...not bloody likely, unless you're sitting inside a freezer. I think something was off on your old measurements - even the new (higher) ones are suspect.
 
You could be very well correct in the old measurements being off. I wasn't using anything other than NVMonitor to watch my temps which appears to have a 1-3C difference from Speedfan and Core Temp. Core Temp 95 isn't stable on my machine though, it crashes a couple minutes after starting.

I'm not trying to brag or win an E-pen contest, that's the temps I saw, that's why I'm upset that it's running hotter than it was. How it could be so far off I've no idea, other than when I did the installs I was at stock settings on the first run and worked my way up to 3.4GHz. This go around I had the machine already clocked before the installs, I dunno.

I've torn the box apart, removed the Sata cables from under the board and reversed the fan, I'm not really seeing any improvement. I'm showing 40-45C (hasn't stabilized yet) between NVmonitor and Speedfan currently, 100% Orthos load, but only 45 minutes into the session. I could be chasing a false reading, but it made me feel good about the build.

Spanki, what would be considered good to acceptable temps for an e6600 at 3.4GHz?

--MF
 
With the Scythe Mine as a HSF, it wouldn't suprise me to see load temps in the upper 50C-60C range. Not that the Mine is a bad HSF, but it's not quite on par with the top air coolers available and if you told me you had a Tuniq Tower or Ultra 120X running 50C or even 55C on load, I wouldn't have surprised me enough to comment (depending on your setup).

Having said that, it also depends a lot on:

- whether you're talking about the CORE temperatures or the CPU temperatures (the 'CPU' temperature is always several-to-many degrees cooler than 'Core' temps)
- your ambient air temp (ARE you sitting in a freezer? :) ) There's another thread in this forum that reports each additional +1C degree of ambient temperature can mean +2C of cpu temp... your mileage may vary.
- your system airflow (the 900 is a great case for airflow - even better if you have the fans howling loud)
- the amount you had to bump the cpu voltage to get that overclock (the higher the voltage, the higher the temps - by a non-linear scale.. a .1 or .2 bump can result in a LOT of extra heat).

...in other words, if you're talking about cpu temps, haven't bumped your voltage up much, you're ambient air temps are low and your case airflow is good, then I'd be very happy with the 43-46C numbers you're seeing with that HSF and that speed under Orthos load.

On the other hand, if you're sitting out on the porch in Texas and the ambient air temp is 30C+ and/or you've got your cpu voltage bumped up to 1.45 or so and you're reporting Core temperatures, I'd be a little concerned that the numbers you're seeing are just wrong.

37C 'core' temperatures for 3.4GHz under Orthos load, with that HSF, is very difficult to believe - 37C 'cpu' temperature, measured in a cool (winter time) environment with good case air-flow might verge on believable :).
 
Just as some point of reference, Core Temp reports the same temperatures for my cores as SpeedFan reports for my cores (not 1 - 3C difference.. the same numbers). But Speedfan also reports several other temperatures... currently, in my system at idle (as I type this), they are:

34C - System temp (not sure where this diode is)
42C - CPU
47C - Core 0
48C - Core 1

...note that it's pretty hot in here right now and I'm running a stock intel HSF, so my temps are prety high for just idling, but note that the cores are 5-6C hotter than the 'cpu' temperature (usually about 6C on my system).
 


I hope that works, sorry about the size, ImageShack just plain sucks. I tried several different formats, but all at the native resolution were crapped out.

I appreciate your information, believe me, I'm using NVMonitor and Speedfan for the temp monitoring, now what they are actually monitoring is anyones guess, lol. Core Temp just refuses to run on my system for longer than a minute or so, but even then it's equal to Speedfan on temps.

I live in Louisiana, AC is what we live on. It's "cool" in here (my office) but not frigid by any means.

Yep, I got 5x120 (2 front, 1 mid, 1 rear opps forgot the 5th, 1 side) and 1 200 howling currently, it's not really all that bad. Certainly less annoying than the HP Netserver LC 2000r I got sitting behind me.

Voltage-wise, if you can make it out on the image is 1.3875v. I busted my ass to keep voltage down and the system stable. I've got 8 hour Orthos runs down no problem.

Is there any other program out there for core temp monitoring you can think of? You do have me worried that something is jacked up on this system and I'm burning it up lol.

I appreciate the input greatly, thanks.

--MF
 
I'd say you're good to go... very nice temps, in fact and if it every gets too hot, the cpu will throttle back anyway.

Another program you could try is "PC Wizard 2007" by the same folks that make CPU-Z (I don;t have that link handy).
 
Thanks Spanki,

I'll look into PC Wizard 2007, I appreciate the suggestion.

I'm going to rescind my first post on the temperature being 37C, I honestly don't know what the final "3.4 OC Temp" was now that I think about it. The 37C could've been a 3.0 or 3.2 run at a lower voltage and just stuck in my head. I found my first 8 hour Orthos run SS which showed a 40C temp in NVMonitor, probably more like 42-43C in reality (Yes, I'm a geek, I SS'ed my first 8 hour Orthos run :p ).

I'm going to blame retardation or senility for this entire waste of time post. Sorry to take up your time, but I appreciate the advise and information.

--MF
 
If CoreTemp doesn't work for you, try downloading Intel's TAT program. It's also very accurate, much more so than Speedfan.
 
You can still download coretemp 94. Coretemp 95 doesn't run right on my machine either but 94 works everytime with no lockups. :)
 
You must log in or register to reply here.
Back
Top