02-13-2012 08:01 PM
Hello all,
I'm new to this forum. I've had my Portege R835-P81 for 2 days. During those two days, I've already used it for processor-intensive activities (compiling large C++ and Java code). During this time, my CPU temperature reaches 100 degrees celsius, which is the upper limit for the i5-2435M. At 101, the CPU will force a thermal overload condition and power off the laptop. It avoids this by throttling its clock speed to 800MHz as it waits for the temperature to return to 99C. Then, the cycle repeats until the load on the system is reduced.
I read another post on this, and it seems this same problem exists with the R835-P56x series, the predecessor to mine. Does anyone know if there is a proper fix for this problem, or must I underpower my CPU in order to use it without burning it to a crisp?
I have to say I'm very disappointed with this. Toshiba, please expect your users to use the CPU at high load for more than 30 seconds. Consistent load temps of 100C will ruin my CPU over time as the die expands and contracts. I find it unreasonable to advertise a system that cannot perform at its specifications.
Any ideas are appreciated. I will say that I am already aware that I might be able to save the temperatures by reducing my CPU clock frequency to max out around 2GHz or so. I'd like to hear that there is another solution, although I don't expect one. After the warranty period expires, maybe I should lap the heatsink and put silver thermal paste on it?
02-14-2012 08:31 AM
quite strange that Java and C++
is pushing the CPU close to extinction.
are you compiling and running Turing machines
at the same time?
I usually get the temp very high when recoding
videos. but I am using mainly freeware without
sophisticated memory caching algorithms like
VirtualDub, WinFF, etc using XviD codecs.
they usually give me an hours long re-write-temp-cache-
led-lightshow of my harddisk indicator. that may
contribute pushing the cpu almost over the edge.
even so, like with the battery life or petrol use -
companies do always disguise under what circumstances and
for how long they achieved fully trumpeted maximum
performance.
ever noticed that the maximum number of the speed indicator
in the dashboard is, whatever car you may buy, always far
beyond reach, whereas it is very easy to push the rev counter
needle into the red?
regards
Damain
02-14-2012 01:29 PM
Damain,
Thanks for your response. Actually what is going on is a 5-threaded C++ compile of Libre Office using G++. I haven't tried video transcoding yet, but I am sure that will be a doozie as well. If I were to try it, I would probably use ffmpeg or similar, though. There was some hard drive activity during the compile, but perhaps not as much as you might think. It was using plenty of RAM during this process, of course.
One more thing, it idles around 65C. Under minor load, it reaches around 75C.
You are right about advertising, of course; I am just pointing this out so other people are aware of the problem. I do consider it a problem and a design flaw, and although it is as common as grass in the plain, I don't think it's okay to disguise specifications.
I would like to know exactly what kind of fan is in this PC, how fast it runs, and the size and material of the heatsink/heat pipes. I wonder if anyone on this forum has pictures, white papers, or anything of the sort? It may be that it is top notch and cannot be improved, but I have difficulty believing that off hand.
Speaking of the fan, is there any way to control the fan on this machine, or to read information about the heatsink temperature and fan speed? That would be very nice! If anyone knows, please enlighten us. =)
By the way, are you also using an R835-P81? I assumed so, but you didn't mention it specifically.
-David
02-15-2012 10:30 AM
02-15-2012 02:24 PM
It's really easy to open it up and see what's in there yourself if you are inclined enough. You can also search for R700 heatsinks as they are very similar, but the plate that touches the cpu is made with aluminum and is bigger on the R700. Or watch this guy's review on youtube. He basically complains about the heat sink fin's being alumimum and not copper. I don't know if it would have made much difference though. I think it would have worked much better if the fins and vents were twice the current length and move the vga port to some other place.
When I replaced the thermal paste, the original goop looked okay, so that didn't really make any difference (maybe 1-3C). So I have to say it's the heatsink itself that can't handle all the heat if it's loaded at 2.8GHz for more than 30 seconds. It does a great job on light to moderate load, but not so well when it's severely loaded. Mine stays idle around 47C right now.
I did add a small copper block heatsink on the north bridge chip though. It doesn't really heat up, but since I had a small vga ram heat sink and thermal tape, I put it on. But this doesn't help with the cpu heat at all.
02-16-2012 03:37 PM - edited 02-16-2012 03:38 PM
65 C idle is not a normal idling temperature. I would consider exchange/refund. [normal is usually high 30s to very low 50s]
02-17-2012 08:58 AM - edited 02-17-2012 08:59 AM
Hi alexmuw,
"65 C idle is not a normal idling temperature.
I would consider exchange/refund..."
that depends what idle is considered to be and
where. would you mind referencing the Toshiba
source of your quote (not a software program)
prior me or others starting claiming refund for
no reason?
regards
Damian
02-22-2012 01:36 PM - edited 02-22-2012 01:42 PM
Hello all,
Just an update on this. I've enabled some changes in my BIOS that have improved cooling to some degree. Using "Optimized Cooling" under the fan control section of the BIOS brings my idle temperatures down to the 49-54C range. This is MUCH better and what I would consider acceptable. However, I still have not solved the inherent problem that still exists under heavy load (~100C temperatures).
It is true that Toshiba has *lied* to customers regarding the copper-core heatsink. Only my heat spreader and heat pipe are copper. The heatsink itself is aluminum. I can see this without even opening the laptop, as the heatsink is clearly visible through the exhaust grate. While this is really awful, I cannot say I had expected the copper heatsink, as I did not even see that advertisement before purchasing - I suppose they don't advertise it anymore after it was discovered to be a bait and switch.
I am considering the option of thickening the heat spreader on the CPU by inserting a properly-sized copper block between the CPU and heat spreader, in the hopes that an increase in heat capacity may quell high temperatures and pipe the heat into the heatsink faster. I am not sure if this would bring a marked improvement, however, because the act of putting a separate piece of metal between the heat spreader and CPU may cause a severe degradation in heat transfer, even if it is the same metal (Copper). I am also considering searching out a copper version of the heatsink itself, if one exists anywhere for purchase. I am hardly expecting there to be a public source of these, but I may get lucky.
In any case, the idea of sending it back is not a pleasant one. I need this laptop for school and have grown very partial to it. Aside from the thermal issues, the laptop is absolutely perfect for my needs. The only thing I'd ask for is a couple of extra hardware sensors within the chassis, especially for fan speed, and a FAN CONTROLLER. I would like to ruin the fan keeping my (much more expensive) CPU cool, but so far it seems that only the BIOS can control that.
Thanks to the person who posted that Youtube video about the cooling design. That proved very helpful and illuminating. Interesting to hear your results regarding the thermal paste. I'm curious about which is worse, the heat spreader or the heat sink. One of them is bottlenecking the heat dissipation. If I could see the heatsink temperature, I'd be able to tell which it is. I have a feeling it's the heat capacity of the spreader itself that is the problem, though, judging by how quickly the chip temperature increases. A heat spreader with more mass should be able to slow the temperature changes (unfortunately that works both ways, cold to hot and hot to cold).
Incidentally, I've also been able to tweak the power management settings enough where I can get 6.5 hours of battery life with ease while at school. The laptop draws less than 10W while idle, reaching as low as 8.5W. Very impressive!
-David
02-23-2012 07:57 PM
Hi David,
Glad to hear some one else is having similar problems with overheating. I haven't had such high temps, but rather rapidly fluctuating temperature and fan speed. I've been fooling around with power setting, but with not much luck. Interestingly, the problem is very different (and better) when I'm running on battery and not connected to the AC adapter. I'm interested to know how you changed the fan settings in the BIOS. Any other settings you can change in there? Also, did you change any other settings that you think made a difference? Thanks!
10-22-2012 09:42 AM
I had the same problems with my Portege R705-P35 which resembles in design with the R835. I my case I did not change any settings and I had to actually disassemble the laptop to figure out the problem. First of all, the thermal paste was dried and I bought a IC Diamond one which seemed to lower my CPU temp from 63C idle (85-90C load) to an 55C (75-85C load). I was still bothered by it and I took the laptop apart a second time to find out the real problem. The fan blows air on top and underneath the motherboard. The air exists through the heat sink from both top/bottom. However, on the bottom, there are a few plastic thin sheets that probably protect from getting dust/water from the keyboard. The sheet next to the fan was not well glued to some metal part and was flopping around and blocking the air from going underneath the motherboard (where the CPU is). So glued that, reapplied my thermal paste (since the first time I placed a little bit to much) and now ... it is awesome: 38-44C idle (65-75C load).
All my tests for CPU on load were done using Firefox with a 1080p quality YouTube video, Media player and VLC player playing two movies.
Copyright © 2013 Toshiba America Information Systems, Inc. All rights reserved.