Refurbished E6410: processor differences regarding heat development

I consider buying a refurbished Latitude E6410. I consider three machines, with the i5 520M, i5 560M and the i7 M 620 processors. I wonder if the 520M will develop less heat than the 560M, especially on the palm rest. If not, is there a difference between these and the i7 processor?

 

Thay are all nominally 35W CPUs with maximum speeds as shown here.

My experience with the more recent Intel CPUs is that the fastest versions do tend to run hotter and push the limits of the design power envelope more than the slower members of the family.

You can, however, limit the maximum CPU speed using the maximum CPU power state setting in Windows power management and this will reduce the heat if you want the computer to run a little cooler.

John

 

John Ratsey: Thank you for your reply. I settled on the machine with the i5 560M and received it yesterday.

The good news: Instead of 4 GB RAM as specified, there were 6 GB; there was a Samsung 840 SSD although the specifications said "SATA" (but there was an icon that said "SSD"); and the graphics card was Nvidia NVS 3100M instead of integrated Intel HD, as specified (although I dont't know whether this last thing is good).

The bad news: the level of heat is higher than the impression I got from the reviews on Notebookcheck (search for "emissions") and Notebookreview. While testing the machine, I have used a thermometer on the speakers on the left side of the keyboard, where the temperature is slightly lower than on the right and the left palmrest, where the level of heat is uncomfortable.

With no settings changed, the level was around 28.5 degrees Celsius (the number viewed in isolation is not very interesting, as I am most interested in the palmrest, and when I started testing, I assumed that a change in temperature above the speakers would be about the same as on the palmrest). With CPU and GPU turbo deactivated in BIOS, max. level has been around 27.2 degrees. I have not seen a significant change when in addition reducing the max. CPU condition in the power settings from 100 percent to 67 percent. Actually, the level of 27.2 was reached when I was writing this post, using the 67 percent setting.

The machine is supposed to have been washed inside and outside. I don't know if this was insufficient in relation to the fan and its functional ability to reduce heat.

At the moment, with the 67 percent setting and turbo deactivated in BIOS, I am at around 32.8 degrees Celcius on the right palmrest. I will edit the post further when I get more results.

It could be that I am a bit too sensitive, but isn't there anything I can do? To deactivate the dedicated graphics card, is that possible? (I only changed a setting in the Nvidia control panel from "let the 3D program decide [whether to give priority to quality or performance]" to "[most] performance [rather than quality]". I have read that the Latitude fan control program is difficult or impossible to use to change the fan settings on the machine. Is there something that could be done with the fan? Has the problem been reported before?

 

It is possible that the model with the dGPU has a better cooling system although the Dell service manual does not show any difference. The dGPU should only run when there is an appropriate graphics workload. There are other reviews here and here which discuss temperatures and fan noise which should help you decide whether what you are observing is typical.

Personally, I would not consider the external temperatures you are observing as unreasonable (some passive cooling reduces the fan activity). These observations are also very sensitive to room temperature. I would be more interested in the internal temperatures. If you run HWiNFO and select temperatures you can see what the temperatures (CPU and some other components) are doing. What is the maximum CPU temperature when running wPrime 1024?

While the computer has been cleaned as part of the refurbishment process it is less likely that the heatsink was repasted. Thermal paste tends to deteriorate with age and often is not optimally applied in the factory to start with.

John

 

I have decided that I will sell this machine to someone else who is less sensitive with temperatures, and I expect to get back most of the money I invested, as the specs are fairly good. I never got around to using HWiNFO, as it required a third-party program in order to by-pass some setting in W7 Pro. preventing me from using a program with an unsigned driver.

The reason I start this thread again is that I am curious about your statement that "Thermal paste tends to deteriorate with age". Does this mean that if I find someone selling a new (not refurbished), say, T520 on Ebay, the thermal paste will have deteriorated in the nearly five years the machine has been in the box, without the CPU or other components having been used?

 

I've never had any problems running HWiNFO. I use the stand-alone zip file version which doesn't get installed.

Aging of thermal paste will be worse on a machine that has been used. Usually, a good user can do a better job of paste application, using better quality paste, than gets done in the factory so it would be prudent to do this if you are sensitive to heat and fan noise.

John

 

The GPU always runs in the Nvidia E6410 - no graphics switching. It does get hotter than the integrated Intel CPU and as you have found out, when you push it, the palmrest gets quite warm. I applied coollaboratory liquid ultra and the temperatures were a lot better. The early BIOS had even more issues with heat - the most recent one helped with my system. You can turn down the CPU speed using Throttlestop or disable one of the cores - both have a detrimental effect on performance.

 

I thought the opposite is true.

Basically, the i5-520M, i5-540M and i5-560M are manufactured identically. These get then binned. Clocks and such are set according to binning.

It's even possible (altough I do not know in this case) that the i7-620M is the same part (if true, there would be 1 MB of Cache disabled on the lower-binned i5 models).

In any case, the higher-binned products tend to be more power-efficient. If you would clock them down to precisely match the frequency characteristics of a lower-binned CPU, they will require less power for a given amount of work.

Often times, there is a separate, second effect going on: a higher-clocked model (say, i5-560M) can finish a given task more quickly than a lower-clocked model (say, an i5-520M). Therefore, the i5-560M can return to a sleep state earlier. This can make it more power-efficient as well, as the i5-560M will not consume significantly more power under 100% load than an i5-520M consumes under 100% load.

 

Some years back, in the days of CPUs such as the P8600, I would have agreed with your statement. However, having had two almost identical Samsung notebooks except for one having the i5-3317u (up to 2.6GHz with TurboBoost) and the other the i7-3517u (up to 3GHz with TurboBoost) I know that the latter generated more heat when running at maximum load although both have the same nominal power rating. It appeared to me to be a case of pay more money and have some extra speed unlocked. Unfortunately, the extra fan noise was somewhat noticeable and I reverted to the slower computer. It's possible, however, that if the faster CPU was limited to the speed of the slower one then it might have been slightly more efficient than the slower CPU. I don't recall testing that scenario.

John

 

I've heard the rationale about being able to complete a task faster and enter a lower power state earlier but it's just never worked that way in my experience. So far as I've seen, a faster CPU generates more heat and the small amount of extra time in a C or P states does not offset this. I also took a 2.6GHz i5 and downclocked it to the speed of the 2.2GHZ of the same family, since I had that CPU to directly compare. The result was either identical or seperated by a negligeable difference that would require scrutiny to discern.