AMAZING Results (Temperature)

Wasn't quite sure where to put this, so this'll have to do for now...

Software:
I8kfanGUI
RMClock

Notebook:
Dell E1505
Excluding replacement HDD, it's entirely stock (never disassembled)
Sitting on desk; no external cooling utilized

CPU-Z:
CPU-Z Validator 3.1

Results:



Haven't tested lower voltages yet, but as it stands, that's 35-36C maximum temps on a burn-in test (at top default frequency of 1.73GHz) using nothing more than free software programs...outstanding!

 

This doesn't really mean much without knowing the ambient room temperature... are you sure there also isn't any offset? My Intel T5550 reports temps 15 deg C lower than actual.

 

Ya any sort of temperature posting without ambient...isn't very helpful.

 

Funny; after logging off earlier, I thought to myself that I should have written listed room temp, as someone would make the accusation that I'm testing in an igloo...

Without further ado; ambient is 20.5C.

 

I guess ur temp sensors are faulty unless of course you've got it dissassembled an have a aircon blowing on it

 

The temp sensors obviously aren't faulty.

I mean, this is kind of amazing, but I assume your fan is on full blast? Also there is a certain amount of luck involved.

 

I'm guessing this is load? If it weren't then it's nothing too unusual.

 

I still doubt you niff... Trust me under no circumstances is any laptop other then a toughbook going to operate at such temps and even the ambient temp is higher so those sensors are faulty... 100% garrunteed.

 

Ambient was stated at 20.5C minimum on graphs is 22-23C.

What makes a toughbook so special that no other notebook can touch it, btw?

 

My laptop used to idle at 36 once I changed the thermal paste. If that's at load then I'm amazed.

 

No, the temperature sensors aren't faulty, but my guess would be that some offset value is set improperly (perhaps default?) in the benchmarking/temperature software the OP is using. At around a 20 degree C ambient, my T500 idles around 35-37 degrees C (used to be around 5 degrees lower when it was new), with the fans off and running on integrated graphics.

 

Yea, if these are load temps, lets see some idle temps!

Nothing idles below 30c.

 

The software could be using a different tjmax. That would scew your results.

 

Have a laser thermometer?

 

jlells01 says this is a "burn-in test", but doesn't define what that would be.

So, jlells01, is this merely turning the computer on, or are you running something like Prime or Hyper Pi? Can you give us a breakdown of what you are running?

 

Looks like hes running OCCT 3.1 CPU test medium data set not much ambiguity there.

 

Been meaning to do that for a while but never got around to it.

 

@niffcreature - I didn't really look at the pix. Prolly shows my naivete, but use prime and hyper-pi for most CPU testing. Thanks for setting me straight.

 

Your assumption would be correct, sir.

There's a drastic temperature difference realized utilizing a tool such as I8kfanGUI (locking fan speed to high) vs. sticking with the default fan modulation/programming.

 

It's wrong obviously. In world of physics that can't be achieved. You can't get 7º degree CPU temperature with a 20ºC room temperature.

 

Since when OP say they were running at 7 degrees?

 

My machine idles at 28C with 22-23C ambient. See my mod!

But there's no way any machine can run at load at 35C.

 

Sorry my mistake. It drops to 23º. was reading the wrong column.

 

A few things to consider:

Processor is undervolted from 1.263v to 0.988v
Fan is forced to run at max RPM's
The stock E1505 chassis pulls air from underneath, just like your DV6z cooling mod

 

Right, but my point is that I can idle 28-30C, but depending on the benchmark, it easily will exceed 60C, and if I push it 80C+.

That graph looks odd. Spikes to 35C and stays there?

edit: Just ran OCCT 3.1.0 just like you (for 15 mins only) and it leveled out at 66C. I know it's apples and oranges comparison, but honestly, no CPU can do that unless it's liquid cooled. But more likely, the CPU is limited. Was this done while on battery power or on AC? Show us CPU-Z results while running it. Record for 30 seconds using fraps while at load to show us the results.

I know when I had my NP8170, on battery it clocked the CPU way down and barely broke a sweat when being taxed, but it also was running only at something like 800MHz instead of 2.2GHz (or more with Turbo Boost). I have a feeling that's what's going on here.

Bottom line is: either the CPU speed is being limited, incorrect offset of temperatures, or the sensors are faulty.

 

0.275V less is huge: I can believe.

 

I can't. 15C over ambient at full load? You rarely see that with desktop CPU's with significant coolers, and Core Duo chips ran hot. I have about a 0.1625V difference from stock and at best it cools it 5C at peak, but still running at 70C, well with OCCT running at 66C vs 72C, and that's with my cooling mod and running a 120mm fan underneath it (at 5v).

My thoughts are if temps aren't faulty, delta to Tmax is set at 85C instead of 100C. 50C I can almost see with a 20C room.

In all my years of overclocking/undervolting desktops with a good air cooler at peak would run at best 20C over ambient.

I'd just like to see a quick FRAPS recording or even a screenshot, of him running OCCT with CoreTemp to see the TjMax value and reported speeds.

 

My T5550 runs at a whopping 0.3V lower undervolted (0.95V from 1.25V) and while temps dropped significantly, it still is no where near your temperature. My processor also required a 15 deg increase since Tjmax was set to 85 instead of 100. So as stated above, that would bring your temperature to a much more realistic number (50 deg C with 20 deg C ambient).

 

Ding ding ding! I think we have a winner!

 

As requested:
http://www.youtube.com/watch?v=5qqCGUaGTJs

(Note - OCCT CPU% gets pulled down a bit due to FRAPS recording.)

 

Thanks. I don't mean to question you, but something doesn't seem right. I'm sure it's running cool, and that was the point you were trying to make. I changed my mind from CPU-z to CoreTemp only because it shows the Tmax temp, as well as live temp readings. But thanks.

 

Copy this. My Acer idles down to 25C with my mod, but lowest idle every recorded is 24C, due to my fans shut down at 25C. However, I could have them running all the time, think my tempratures will reach some high 15Cs then.

 

Dwm 2011 10 05 02 00 02 85 - YouTube

I'm running I8kfanGUI, which locks my fan speed to high, regardless of temp.

 

Then it is a whole another story.

 

Honestly, if you want to know your real temperatures, get a laser thermometer. Just because your computer says one thing doesn't mean it's correct. I'd still bet that your Tjmax settings are off.

 

I can see a large flaw in that proposal. Trying to measure core temperature with a laser thermometer when there's a heatsink on top of the whole thing doesn't really work.

 

The back of the heatsink, while not as hot as the core beneath it, can still give you a fairly accurate reading of load over time since the metal will quickly absorb the heat and reach equilibrium between the heat produced and heat distributed. The point is not to get the most accurate reading of the actual temperature of the cores but to get a range of temperatures to see if they align with the software. Most reviews, including NBR, use surface temperatures to measure how good this dissipation is as opposed to pure software since it creates a level field in which to compare notebooks.

 

Yep funky, you're quite right. If you take off the heatsink while running the CPU in low power then the case temp should be about 1C or 2C less than the core. A range of temps can then be made to give a better idea of Tjmax and idle temps which can then be used to calibrate the chip, well at least with the C2D's. A lot of trouble to go to really and while the OP's temps may or may not be off a few deg's, IMO he's still done a great job with the cooling.

 

Just post an image of CoreTemp and it will show TjMax. I can bet it is at 85C not 100C like it should be.

 

Even with the proper Tjmax value, which is likely off a few degrees from the real Tjmax, the further away from DTS=0 you are the more inaccurate it can become. In other words in most cases CPU temps should be taken only as a rough guide.

 

No, it will show the "determined" TMax of the CPU which is a fixed value. Many times a program will see it as something different than it should be. If it sees TjMax as 85C and reporting 35C max temps, if you adjust TMax to 100C like it should be, that will show the max load temp as 50C instead.

 

Yes I understand what you are saying but perhaps I haven't made myself clear. I'm not disagreeing with your query to the use of correct Tjmax.

There was a time I use to take CPU temps seriously. For instance on my P8400 Tjmax is 105°C. Using an infra-red thermoscope (the laser is just a pointer and doesn't measure temp ) the real Tjmax was closer to 108°C when DTS hit 0. Even using 108°C as Tjmax still doesn't give good results at lower temps which can still be out by quite a bit when DTS is a fair way from 0. I even use to go to the trouble of calibrating these idle temps but long ago realized it's not worth the hassle. You have a relative reading and that tells you more than enough if your CPU is running hotter, colder, within spec, and that is enough IMHO. Doesn't really matter much if the temperature conversion is out by a few degrees.

Here's an example of the 2 cores with idle calibration but way off just for laughs. One core at 92°C the other at 12°C.