M-6864FX heat (And like Models of the M-Series with HD2600)

I got an M-6862 which appears to be the same as the entire gate M-Series lineup with the Radeon HD 2600 cards. Now I find that under stress my GPU and CPU get very hot. I do not think the internal cooler is able to handle it. I am wondering if it's just my unit or other units also. If anyone has one of these M-series with HD 2600 it would be very kind if you could post your temperatures under similar conditions.

My stress test appears to get the CPU to about 90C and the GPU over 100C. 105C even.

I run the FUR benchmark while monitoring the GPU temperature with the AMD GPU Clock tool. And the GPU temps hits over 100 with this benchmark which is far hotter then 3DMark06 can manage.

URL to the fur benchmark
[ http://www.ozone3d.net/benchmarks/fur/ ]

URL to the AMD GPU CLock Tool
[ http://www.techpowerup.com/downloads/1128/AMD_GPU_Clock_Tool_v0.9.8.html ]

Would anyone with a similar machine be kind enough to run FUR in stress test mode for 5-10m and see how hot their GPU gets? I suggest running FUR in windowed mode so you can have the AMD GPU Clock Tool window by side to see if your gpu gets to hot so that you can abort FUR.
Also to really test it run prime95 on the other core.

My machine gets VERY HOT. The GPU hits 105/106C and CPU hits 90C if Prime95 running on the other core and about 84C with FUR alone.

 

well i ran a quick stress test for 5 mins and got the following results:
(room temp about 75F, no laptop cooling pad, stock gpu settings, running on plugged in power/high performance, undervolted cpu)

CPU - 67F
GPU - 85F

 

I think your machine might have temp problems.

Can you do this for me, so we can compare temps accurately?
Download Orthos CPU Stress test, or w/e you call it.
Download HWMonitor.
Run Orthos for exactly 5 mins.
Use snipping tool to copy your HWMonitor results and post them here.


These are my temps running Orthos for 5 mins with no cooling pad. NOTE: Its undervolted.

I will run your FUR benchmark when I have time, but I can tell you right now, my GPU will not go anywhere near that 100C mark.

 

Do you mean 85C and 67C? 85F is impossibly low.

 

Your CPU seems to suffer from the same problem the M-152XL's does - the temperature reading is not right. The CPU CANNOT run below room temperature just sitting on your desk.

I have noticed that if you put your computer to sleep then pull it out, you will get accurate readings of the temperature via HWMonitor (or at least more believable readings)

And I agree with sig, 85F does not seem right for the GPU.

 

Anyone doing this test run FUR at 640x480 windowed mode in stress test mode and keep the temperature utility side by side. Before starting fur start the 'Orthos CPU Stress test' to really load up for your computer's cpu. Because fur alone will only use up one core and your GPU. So orthos will heat up the other core. And this about the worst possible thermal scenario you can get. Both cores at 100% and GPU at 100% and give it 5-8min.

Make sure to run it for at least 500sec as the last 10-15C climb after 300 seconds. Fur shows you the seconds it's running for so pay attention to it. You really need those 5-8min to reach the final temperatures.

 

Thanks for the orthos tip. Very nice load testing utility. About that 100C we'll see just give fur a good chance... 5-8min at least. (hopefully 10) and run orthos at the same time to make sure the cpu gives it all the heat it's got also. And that way you will test your computer at it's thermal worse.

I am honestly hoping you don't hit 100C because then I know it's my laptop and I can take for service. But if you do hit near 100C that would suck because that would indicate poor thermal design of the M-Series.

 

Mine CPU gets to 80C on orthos alone but I replaced the T5750 with a T7500 from a previous broken laptop. So it's a 2.2GHZ 800Mhz FSB CPU vs the 5750 which is a 2.0Ghz 667mhz FSB part.

 

Ok, I am not going to run orthos and another benchmark at the same time, sorry, i dont want to permanently dmg my machine.

Do the test i said and post your temperatures. That way we can compare without going to the extreme.

 

Damage your machine? That's not possible. More over if you follow the process I described the temperature rise is fairly gradual it takes about 8min to climb up over 100C so you can simply stop FUR by clicking the close button at any time when you feel the temp isn't comfortable for you. There is certainly no risk at damaging the machine as I run ran the test for 10M after the GPU reached 106C (well I am less concerned because I am still within my 15 days replacement period). But you can close fur anytime you see the temperature start getting close to 90C or 100C or whatever you do not find comfortable. The point is to tell if your machine will reach a terminal temperature of say 85C or will it keep climing past 100C like mine. For example if it gets to 90C you know there is more to go.

Point is there is nothing to be afraid off since you can stop the process at any time. As soon as fur is closed even with orthos running the temp begins to drop quickly down to 80C for me for CPU and 75C for the GPU.

 

Also I've said 10min run of orthos got my CPU temp to 81C.

Core #0 was at 80C and Core #1 was at 81C. Keep in mind this is the T7500 cpu not the T5750.

 

If you have not undervolted yet, you are fine, no worries about heat.

Rereading your posts, you say your CPU/GPU hit these temps when you have them running 100% max for 8 mins......................................

When I said running FUR would not make my GPU hit 100C, I didnt not think you were talking about making it run 100% for 8-10 mins....

 

No thanks.

 

You really aren't interesting in seeing how safe your laptop is? This isn't a run away nuclear reactor. I mean you're claiming your laptop will never get anywhere near 100C but are refusing to try to take it to even 90C? (assuming it gets there). I am not sure what scares you from attempting this benchmark. As I explained the temperature goes up very gradually over 10m period so whenever it's past your comfort zone of say 85C you can stop the process. But you will know to be careful with your machine because when playing something intensive it can over heat. This way you're seeing what happens to it in a real full load stress test in a controlled environment. As opposed to same thing happening when you're playing some game without you knowing. It's as useful for you as it is for me.

Also about under volting I did try RMClock and under volted but I have a suspicion the actual under volting isn't working. The reason for that is that while RMClock is reporting the designated voltage "Core Temp" utility is reporting standard voltages. I consulted the RMClock manual for that and here is what it had to say about that

"Q: CPU voltage indicated by RMClock greatly differs from the voltage reported by my vendor-supplied monitoring software. What's wrong?

A: Nothing. RMClock reports the CPU VID, i.e., just a requested CPU voltage level. It is up to the motherboard logic to decide whether this request should be handled or ignored. Most motherboards designed to properly support processor's power-saving features (e.g., AMD Cool`n'Quiet(tm) technology) should adjust the real CPU voltage so that it's close to the requested VID level. Other models may apply some extra over-voltage (e.g., +0.1 or +0.2V), which is useful for overclocking. Finally, there are many motherboard models that can be set up (via BIOS settings) to supply constant voltage to the CPU, which is far from being good for achieving significant power saving."

So this means that in fact chances are we're not undervolting the CPU as much as we think (or if it at all). I will do more testing on that later on. I suggest you verify your voltages with another utility.

 

I'm getting ~100C on the GPU as well, but neither of the CPU's cores goes over 72C.

 

Thanks. Chances are then my machine is working within spec. My CPU gets to 80-90C but I have replaced the T5750 with the T7500 so chances are that because the average thermal profile for the T7500 is hotter then the T5750 is why I am getting higher CPU temps.

Though I am getting the feeling that under volting via RMClock isn't actually doing anything because the requested voltage by RMClock appears to be ignored by the motherboard. Even though RMClock is reporting the requested voltage other utilities show original voltages. When I get home from work I will do more investigation on that. But if anyone has any ideas about that please contribute .

 

yup my bad...obviously meant

CPU - 67C
GPU - 85C

 

I wouldn't worry about it too much. If you get the chance to look at the heatsink, you'll see why the GPU is so hot. The contact between the two is very poor, in addition to the contact surface being low grade aluminum instead of the copper surface like the CPU.

 

Your right. I am not interested in seeing if my laptop GPU can hit 100C. If it ever does that, then we will talk. Until then, no thx

I cannot speak if RMClock is actually undervolting or not. All I can say is that it does SOMETHING because my temperature did fall.

 

I did look at the heatsink as I had to take it off to replace the T5750 with the T7500. There are a few issues with it. Gateway uses some 'putty' like substance on the GPU and the heatsink. I scraped the putty off and put in a bag and applied AS5 to the GPU and the CPU and the Northbridge and reinstalled the heatsink.

Then I took it off (I always do this to see how AS5 gets spread to make sure there is contact). To my surprise the AS5 on the GPU and the Northbridge didn't get spread! Which means that is a HUGE amount of clearance between the GPU and the Northbridge.

I left the AS5 on the CPU and cleaned AS5 from the GPU and Northbridge then I reapplied the huge think layer of the putty like thing to the GPU and the thick cooler strip to the Northbridge and reassembled. So I would say the cooler design in the M-Series is simply horrible.

I wonder if it would be a good idea to get a thin copper strip and apply AS5 on both sides and put it between the GPU the cooler. Because this 'putty' approach isn't to my liking at all.

Very disappointed with gateways' cooler design on the M-Series. The fact that other people who aren't senselessly afraid getting 100C on the GPU means that the thermal design of the M-Series is most likely faulty. Need to research a DYI method of correcting it.

 

Hm, now you make me wonder if mine isn't making the proper contact, since I also took it off and applied AS5. Guess we'll find out tonight.

 

I would love for someone who hasn't taken off the heatsink to run this test but they seem to be afraid of something. Ok let me know of your results once you inspect it

 

i had a temp of 83C after a stress test for 10 minutes and then i undervolted it which dropped down to 62C

 

Ok few questions were you running Orthos in the background while running FUR? And were you on AC power or battery? I am asking because running FUR alone gets my GPU to about 85C with orthos is how it hits over 100C. On battery the machine scales the card down to about 60% speed. Also you can't undervolt the GPU only the CPU so the 83C is that the CPU or the GPU temperature?

 

I just checked, and sure enough, there wasn't good contact between the heatsink and GPU die. Luckily I saved the thermal putty in a container, so I put it back on, applied AS5 between the GPU die and putty, and the heatsink. My GPU temperature are now in the 80C's; should drop a little more once the thermal compound settles in.

 

Hmm, so you ran FUR + Orthos for 10m and now the GPU stays around 80C?

 

Right. It hang around 83-85C for about 6 minutes, and crept up to high 80's afterwards. The thing is, it didn't jump up to 100C in two or three minutes like it did before. This tells me there is some gap between the die and heatsink, and you need some sort of medium between the two to effectively transfer heat...or use a lot of thermal compound.

 

Chances are copper is far better then their thermal putty thing and AS5 is most likely better also. But AS5 is a grease not putty so it will stick to which ever surface and not really form a 'layer' like the putty. I wondering if getting a thin copper strip and putting AS5 on both sides and make HS -> AS5 -> Copper Strip -> AS5 -> GPU sandwich would be more efficient then gateways' thermal putty thing.

 

I dropped by a Bestbuy and installed AMD GPU Clock Tool and Orthos and FUR and run it on their store machine for 10M or so. GPU got to 88C and didn't go any higher. So I guess the my T7500 is the cause because it seems to produce 10C or more heat vs T5750 under load.

I picked up a copper strip/foil thing so if I have time tonight or later on I will try the copper strip sandwich idea vs the thermal putty maybe the results will be more impressive. HS->AS5->Copper Strip->AS5->GPU and see how that works out.

 

What I'm wondering is why my 6682 will kick the timings up to 600/700 when all i've been doing is browsing the web.

 

I replaced the Putty with HS->AS5->Copper Foil->AS5->Copper Foil->AS5->GPU and it appears to work a tad better then putty after 12min of fur and orthos the temp never went above 98C. Where with the putty it would get to 106C in 8m or so. I think there might sense in doing the 3d layer of copper and as5 to get better contact. the best would be get a piece of copper with proper thickness it would beat the my solution and the default thermal putty.

 

Interesting note to ppl doing this test Orthos has a few modes of operation it is defaulted to "Stress CPU and RAM" you should change it to Stress CPU Only and not stress ram because stressing ram wastes time and lets cpu cool down.

 

Replaced the copper film strips with a copper sheet that's 1mm thick that's 1.4cm x 1.4cm for both the chipset and the GPU. Now the temp never goes above 86C with orthos in cpu stress test mode and fur at the same time for 15 min. While playing stalker clear sky I overclocked the HD2600 to 600 core and 800mhz RAM. And the GPU temp never went above 92C.

Keep in mind my system has the T7500 that seems to run somewhat hotter then the T5750 and also the room ambient temperature makes a huge diff. My room was at 22C.

 

Mine was very cool. Hours of playing Mass Effect caused lots of hot air to blow out.

The 2600 ran ME and GRID much better than a 9500m GS in another notebook I reviewed.

You should get yours checked out.

 

There is nothing to check out my temperatures are fine. You on the other hand need to provide some specific numbers as opposed to 'cool'. Use HWMonitor 32bit version for example.

 

WOW! Ur temps are freaking fine!, u even overcloked that gpu!
I have a M-153xl, with the T-8300 and the hd 2600
While playing Cod4 or left4dead, gpu can go up to 96 degree, and cores around 74

Whould you recommend me to do the same hardware modification? Have you tried a cooling pad? ( bottom of laptop is really hot , which is made of plastic)(I lifted up my M-153 by like 2 inch to increase airflow)
THX!

 

Well what is the ambient temperature in your room?

 

20 Celsius degree

 

Waiting for ur in-advance-appreciated answer!

 

I have an M-6864FX and have had problems with the heat being uncomfortable.

I work for a company that does manufacturing on routers and other networking equipment and so heat management is something our electrical engineer works with all the time. I pinged him about it and we got a sample of a product from Chomerics that is designed as a gap filler. The product is Chomerics THerm-A-Gap 976 chomerics.com for info.

I like the idea of a gap filler because it's fully cured and doesn't have any setup time to work efficiently. That and this compound it is so highly conductive, almost 2x more conductive than arctic silver. So I used it to replace the CPU, GPU and NB.

Ambient room temp: 21.1C (70F).
GPU: 90C MAX. Holding stable at 88C.
CPU: 83C MAX. Holding stable at 83C.

So given other results I've seen here and on other forums, I'd have to say this stuff is possibly the best option for the GPU, and better than stock for the CPU as well.

I think I can probably do a little better with the CPU with something like AS5 because of the very tight spacing not being quite enough for a gap filler to be fully effective.

Feedback welcome.

 

the best and more easy trick to reduce heat in my M-6862 laptop is undervolt
after hour of intense gaming with crysis the CPU is around 70C in a ambient room with 27C, is hot too, but more healthy for laptop and my hands
under normal use CPU is 32-45C with 5 apps open and running.

without undervolt under intense gaming the processor is on fire!... around 85-90C

would be a nice option to lower the voltage to the GPU too.

details of undervolt in my case
CPU fixed at 2.0ghz and FID 12.0x
CPU voltage VID 1.0750v (stock is 1.250v)
all time in maximum performace (conected to AC power)
with battery put profile in performance on demand and enable throttle in CPU

and.. no BSOD, not freezed apps, vista x64 run stable and responsive... with less heat!!!

 

Yaquiz, you have a centrino right?(penryn core) If so , how did you undervolt ? RMclock doesnt recognize my cpu.. I use latest version.

 

right, is a centrino with cpu Merom-2M
my RMclock version is 2.35.0 built on 2008/02/29
with pro features enabled and RTCore64.sys compatible with vista x64

the only strange thing is that RMclock report that cpu 0 is (ex.) 35C and cpu 1 is 5C more all the time :S

 

Keep in mind copper is most likely still far more head conductive then the material you used. So a copper strip with AS5 or whatever the most heat conducting paste you can find will probably meet or beat the thing you're using.

 

Thx for answer,
in RMClocks, in cpu info , does the program gives corrects infos? name of cpu core revision? (because it says unknown for both of them)

And which guide did you used for undervolting?

 

this is the guide for undervolting.

 

This is a pain in the but it lowered my max temps after undervolting by 11 degrees celsius. First I bought some copper from lowes which turned out not to be copper after cutting it. Then I found some pennies 1980 and earlier(Newer ones are zinc/copperplate.) Strong sissors/tin snips to cut' em into the shape of the bridge and gpu, sanded with course grain then fine grain. (Reverse wrapping my finger with tape made it much easier for grip.) There is a screw under the left palm rest which I loosened because after sanding flat it was still a bit to wide, to release pressure on the chips.

 

Yep seems copper plates are a much better solution then gateways' thermal putty junk. I can only guess they didn't put copper there in the first place due to it being fairly expensive now-a-days.

I wonder how much cooler will the system run if I just take off the cpu/ram/GPU/HSF cover and run it that way. The cover seems to be designed to allow very little airflow in there on the other hand it keeps the dust out.

 

the covers like these ones have more air flow than the stock covers.

http://www.laptopking.com/kingpartdetails.asp?ref=2553&category=misc

we need guinea pig...who volunteer?

 

That looks like a good idea but im not 100% sure that cover is exact fit... if someone could spend the time to verify it