XPS 15 R2 (L502) Temps Only thread please.....

I know, the problem is that the board and other components tend to fail due to the heat. :/ Very rarely is it the CPU that dies, but the heat it generates can kill other components.

 

But it is possible that the heat is carried away enough from the surrounding components that it isn't a problem. I would think this might be indicated by the case being so cool, but I'm not sure how else we could check.

 

here are temps at normal use with just the web browser open...

 

well i may hope dell have engineerd the mainbord so it can handle it.
else it would be a faulty design.

 

They've apparently done a better job than Apple, but that's not really saying much. Over the last two days, I've literally worked the cr@p out of my laptop and it hasn't even hiccuped. I've encoded a half dozen dvds, ran prime95, 3dmark06 and even finally downloaded furmark. Yes it gets hotter than most people are used to, but I have no doubt it is doing so by design.

What I'm itching to do now is get a hold of one of the K versions, cool the sh!t out of it and run it up in my desktop.

 

Is there any way you could post temps and the processor speed at that time?

As stated previously it would be very useful if we can get some hard numbers upon which to make some assessments.

Thanks in advance.

 

Maybe this means that even if you proceed to repaste then temps not change but CPU run faster.

 

That's the thinking, faster for longer.

 

and maybe idle temp lower.

 

Why is your 2720QM only showing 2 cores?

 

Maybe others core are off, because web browser is an easy task.

 

Not sure why? Before the clean install it used to show 8 cores.

 

It should be showing 4 cores, regardless of what is running or not. All four cores are active at all times and will show in HWmonitor. If you can't get it to show all 4, try SIW. I like it better because it shows the gpu even when it's not running.

 

I've had HWmonitor only show 2 cores before. Also had Prime95 only let me run 4 threads, not sure why, but when I rebooted and ran prime95 it was back to 8 threads and 4 cores. I'm guessing it just needs a kick to ramp up the other two cores and HWMonitor and prime95 config aren't enough to do it.. I also had closed the Intel Turbo Boost monitor, so I wasn't sure if that had anything to do with it or not. Guess I could always test it.

 

Perhaps the other cores aren't detectable by some of these programs when they are power-gated.

 

So what should I do? After I did the Clean install it only shows 2 cores before clean install it showed 4 or 8 cores.

Plus I am having some problems with YouTube, When I start up a video i hear this fuzzy sound and it goes away, Could This be a problem due to missing a Driver after the Clean Install of Windows 7????

 

Possibly due to missing driver. Always check the Device Manager for any yellow icons, though sometimes a generic driver will be installed if you don't install the specific driver.

As for the core number, take a look at the task manager. It should show the correct number of cores. I'm betting its the software that doesn't quite know how to deal with Sandy Bridge yet. But you could go to Intel's website and download the drivers for the chipset, just in case.

 

The machine has quite a few zones for temerature measurement. Would the processor core temps being high but the rest of the chassis measurements being low help - ie show that the heat is carried away? None of the measurements here seem to show all the zones.

 

Some temps, with furmark and prime95 running. Not sure whether they're useful...

Memory Channel x Rank Max temps are a bit scary, though I dont know what they are. Peak of 123C. Other parts of the machine stay a fair bit cooler. CPU dropped it's frequency from 2.6Ghz to 2.2 and then to 2 when furmark was running.

 

frequency 2.0GHz and not throttling?

 

What would show me whether it's throttling? I was thinking that the CPU going below it's base speed under load was throttling? But I dont know.

The red line graph at the bottom is power use, to see whether it tries to draw more than 130w. Sadly I left my coffee machine on though so it's not useful. Can do it again if it'd be useful.

 

The other two burnt out?









I'll get my coat.........................

 

I think it might only indicate when TM2-triggered throttling is engaged. Maybe TurboBoost is superseding that mechanism. Although this is a bad situation--it shouldn't be going below the base clock speed if the cooling system is adequate. Zygotic, did you keep it going for a while to see if it climbed back up to 2.2 GHz? This is a control system, so it is quite possible that it overcompensates slightly before settling at the base clock speed.

For fun, I used Prime95 and HWiNFO32 on my XPS m1530 (see signature) to see what it did. It ran the cores up for about 15 minutes until the main CPU sensor reaches 100 deg C, then it drops back to 800 MHz for a minute to cool down to ~80 deg C, then brings them back to full speed. The cycle repeats ad infinitum. (EDIT: I believe this is TM2 throttling, but it didn't show up in HWiNFO32 as throttling, so I don't know what would show up as throttling there.)

Now this is without Furmark running too, but it seems notable that the DIMM and chipset sensors in zygotic's XPS 15(~60 deg C) are a full 15 degrees lower than my XPS m1530 (~75 deg C), seeming to demonstrate that it is much better at removing heat from the case in general.

 

I don't understand why clock is under base frequency and not there are Throttling warnings instead in Macbook pro 15 test there is Throttling warning but frequency is still 2GHZ (base frequency for 2635QM).



Can you use Tmonitor to show history frequency while you doing test?
http://www.cpuid.com/softwares/tmonitor.html

 

To ensure you are not having any core parking issues you can disable the feature.

Google: How to turn off core parking in windows 7 or go here:

UltimateComputers.net Forum - View topic - Disable Core Parking Windows 7 Tweak may improve performance

I have used this on my laptop and it really helps to load balance and keep the CPU from spiking as it’s described.

 

I found this to be interesting. I've been running Folding@Home SMP Core on my L502X with the smp tag set to use 4 cpu cores. Running on just 4 cores and using %50 of cpu resources, it's able to hold 3.0Ghz indefinitely. The picture was taken at 30minutes and it's now been 45minutes with no change, so I'm guessing it's just going to stay at 3.0Ghz unless temps in my house change or something frys.

Also, when running on all 8 cores, the temps were the same, but it was only sustaining 2.6Ghz. I know this is how intel set it up, but I find the temperature info compelling. Temps were nearly identical whether using 4 or 8 cores, the only thing different was the speed the cores were running. This kind of reinforces the idea that things are operating as Intel intended. Even if it's a little hotter than we'd like. With cpu intensive operations only, my temps seem to hover around 80C +- a couple degrees. It's not until I fire up the GPU that temps move to the 90c range.

 

Tmonitor (and Tmonitor64) says that the processor isn't supported, so I can't do that.

I'm about to run a longer test. It did sit at 2Ghz for at least a few minutes last time but i'll pay extra attention this time.

Quite like the new core smell from it, as the fan gets going

 

Ok, reran a test. 10 min with just prime95 running, then a further 20 with furmark going. With just prime95 it will sit with all four cores fluctuating between 2.2Ghz and 2.7Ghz. When I add in furmark they clock down to 2.0Ghz within a few minutes and stay there.

No throttle warning in hwinfo (see what you meant now about the warning). But then hwinfo incorrectly reads my machine as having Cougar Point stepping B2 where other tools report B3 - so maybe hwinfo doesn't know what would constitute clocking on the SB processors.

Agree with acruxksa that it's adding in the GPU that pushes the temperatures into the 90s. Interesting to see that core0 and core3 are consistently ~5c cooler than 1 and 2. Also seems to be addition of the GPU that kicks the CPU down to below it's base speed.

My power monitoring showed a peak draw of ~120w. But thats on the power consumption of the whole house so it could be that the fridge or something else cycled on or off, which would swing the figure up or down without me being able to see.

I've noticed that the GPU load reported by furmark moves between 50% and 99% between updates of the stats at top left, and that hwinfo32 makes the whole machine stutter for a fraction of a second when it polls the sensors. Are those things normal?

BTW, nothing uncomfortable about the temperatures here, from outside the case - keyboard, palm rest, even the base. Wouldn't want to touch the grille over the exhaust vent for long though.

 

Retested and dimmed the LCD to minimum when it dropped to 2Ghz. Didnt clock back up.

Seems to drop CPU to 2Ghz either when the GPU hits 93C or when one processor core hits 98. Until then it's unwavering at 2.2Ghz, then switches and is unwavering at 2.0. Pops back up when CPU is ~88 or CPU is ~56. Repeated it a few times - both seem to happen at similar times, on the way up and the way down so not sure which it is

 

Hmm. It throttled. CPU hit 100C. This was after running 4 or 5 prime95 + furmark tests with hwinfo32 open.

 

It throttled, but it's still at 2600MHz

 

Yeah, not what I thought it'd do. Trying to find whether there are other ways of dealing with high temps than dropping the multiplier.

 

Are those tests stressing both gpu's at the same time? Just curious.

 

Dang dude, even your GPU hit 94 C, thats a toasty system.

 

No, you can only have one active at a time, unless you are using that software solution whose name is currently escaping me.

Though this raises a good point--if you were using integrated graphics instead, you might see even lower TurboBoost and more throttling.

Man, I wish I had my system already so I could test it out too. Theoretically it should be shipping as we speak...

 

The outside didnt feel like the inside was hot enough to boil water though!

Just noticed the CPU was drawing 51.5W with a TDP of 45W. Hmm. Might not do that again!

 

Wow looking at all these temps are really making me not get this laptop, I really don't wanna hit 90c, the most 85c, but gettings temps in the 90's is just too hot.

 

hmm almost all sandy bridge laptops are hitting the 90's and higher..

dont forget your not using your system @ 100% load in normal conditions.

 

Yeah I know, but let's say playing a game like BioShock or COD, are my temps gonna go up to 90's??

 

The CPU might be. The laptop as a whole apparently doesn't get that hot. What reason do you have to like 85 but not 95 otherwise? The CPU actively manages the temperature to be below a certain safe point (100).

 

Repasting will drop the load temps, if you are careful and do it right. At this time I wouldn't recommend anything other than AS5 though for the repaste.

 

Yeah I'm more concerned about the laptop as a whole, not the CPU, I just don't want the laptop to be hot from the outside you know, sorry for the confusion.

 

Here is proof that i can get 8 hours on the dell xps 15 with a 9-cell battery...

 

Dkwhite, why only AS5? I'll try ICD7, when I get my unit, unless you convince me otherwise.

Did you guys see Noslow5o's pics of the CPU, GPU, and the heatsinks in the L502 thread?

It almost looks like the procs and HSs aren't mated very well. Look at the stock paste patterns.

 

It's really not hot on the outside, other than by the fan exhaust vent. If you look at the screenshots the temperature is far from even across the chassis - the cooling system seems to do a good job of dragging air through and pushing it out. I'm not worried at the moment - I can get the CPU and GPU to get really hot if I try to make them get hot, as can MBP people and people with other brands (google sandy bridge mobile temperature, for example)

I haven't installed much yet as it looks like it needs attention from Dell for a broken trackpad button. But while watching videos, surfing, listening to music etc the fan comes on at a low level occasionally but I feel no heat from it on my lap and the bottom doesn't feel warm to my hands.

 

Copied from the main L502x thread:

A quick thermal paste review running only 3dMark06. Temps on all cores are between 3-4C lower than before, with the gpu being 6C lower on average. Typical temps from default free 3dMark06:

Core 0 - 85C
Core 1 - 84C
Core 2 - 84C
Core 3 - 80C
GPU - 68C

So, there is a little improvement, but not much. Running Furmark and Prime95 resulted in closer temps to stock paste, but core temps didn't go past 90C. GPU did hit 92C also. This was after 10 minutes of 1080p Burn in and Large FFT's for max heat. This is ~3C cooler on cores and ~2C cooler on gpu.

As you can tell from my pictures earlier, the thermal transfer surface area of the i7 and GT540M are very small. My i7 920 has about 300% the thermal transfer surface area for a not much faster chip. I must admit I really don't think the paste will yield much better results. I think the cooling efficiency of the laptop is too poor to keep temps under control at 100% load for any sustained period of time. It's main purpose is to slow down the duration of time in which the cpu reaches TDP, and allow the cpu to throttle itself afterwards to keep the frequency as high as possible without going over it's limit. This goes along with the theories in the high temp thread, and this repaste only confirms those theories.

 

Have you noticed if the clock is increased when it is under load after repaste or some other improvement related to the turbo boost?

I think that the only thermal compound that make the difference is Coollaboratory Liquid Metal Ultra, but it has many cons.
In this thread you can the difference between a repaste with AS5 and another one with Coollabs Liquid Metal Ultra.

 

They aren't, the spread pattern on mine was horrible, which is probably why Dell uses tons of extra cooling "paste" (theirs is more like cement).
I had to use more than the normal recommended amount for properly made components when pasting to ensure contact throughout, and the copper on the heatsink was not even and had rough valley like grooves in it. Something I've never seen even on "cheap" $20 desktop CPU coolers.

Liquid Metal Ultra is awesome, but be careful with it. It conducts electricity so your application has to be perfect. Also, it corrodes many types of metal, except copper and your CPU, so again, make sure your spread is perfect.
Oh, and its expensive, $20 for a tiny tube.

I agree with NoSlow, the cooling system design Dell is using is too small for the heat output of these performance components. They're using a single fan to cool both the GPU and CPU with a very constricted air inlet on the bottom.
Compared to Macbook Pros, they have two fans per laptop. Not sure of other laptop types, I've only worked on these two in the past couple years.
With the huge chassis and all that space inside the XPS, I feel like Dell had enough room to design an awesome cooling system, they had the room for it. They just didn't do it.

 

I wouldn't use it just because it corrodes aluminum. Half of the heatsink is aluminum, and the Liquid Metal Ultra is not much better, if any, to warrant possible damage to the heatsink in my opinion. From my recent research, I decided to use the Shin-Etsu based on controlled test results as well as the support of many overclockers on various forums. ICD 7 is also a great paste. Which paste is the best is a never ending debate, so I go with what I'm familiar with or what proven results show. AS5 is a decent paste, and I would use it if I had to, but there are better pastes out there. When it comes down to testing ALL of these pastes, the differences can vary by as little as 2-3C, so there's no clear winner, although the AS5 has continually run behind the newer pastes like Shin-Etsu and ICD 7.

The bad part for us is, the thermal paste is not the bottleneck, and therefore is not going to solve all our temp problems. The problem lies in the heatsink itself, as well as the small thermal surface area of the chips that barely touch the heatsinks. When you top that off with a medium flowing fan blowing on an extremely small air cooled radiator the size of my thumb, you have the necessary equipment to cool down my iphone cpu. I'm not blaming dell here because I know they did what they had to, and intel does what it needs to do by squeezing out every bit of power until TDP, then letting off a little bit to keep the frequencies up.

 

Maybe Macbook Pro have better cooler system, but it throttle too
and Macbook 13 with i7 2620M is locked to 2.7GHz it can't use turbo boost.
Intel Turbo Boost is MIA on new 13-inch MacBook Pro? (update: negatory) -- Engadget

XPS15 & Macbook pro 15 can touch very high temps, but their case not are so hot how procs temps would suggest.
This is a little strange...