*OFFICIAL* M18xR1/R1/18 Owners "Are my CPU/GPU Temps OK Thread" - Ask Questions/Get Help Here

Heihachi ,

ok run Batman for a hour and Resident evil 5 for 3 hours on overclock settings you give me , plus i am running with thermal compound supplied with 3 pipe HS ... starting temps are 44 degrees and highest in all that time is 76 degrees

i hope that helps you ?

plus i am running a notepal U3 as well lol

 

Thank you, but i need your temps after successful run of 3dmark11

 

Prime 95 for 20 mins 86 degrees on stock compound

 

Thanks according to your info its almost no difference between 2piped vs 3piped heatsink.

 

just downloading 3Dmark11 now ! just paid for Pro as well lol

well i dont know about that , as i have never had 2 pipe . i am upgrading the CPU to 3920xm the weekend , so will use icy diamond and see temps then

 

prime95 gave me a max of 82C after 35 min with arctic cooling MX-2 on my 3720QM at 40/39/38/38 (36h after the install so the MX-2 compound had some time to settle)
and as you can see in the pic below I was at 79-80C most of the time on core #1 (which is always the hottest for me)

Sure seems like its quite similar, the 3 pipe has 3 smaller heatpipes compared to the 2 bigger one from the 2 pipes and overall they seem to cover the same area. After all the CPU die is kinda small ...

I'm thinking better thermal paste would help bring the temps down even more since I have lower with a 2 pipe heatsink on the 3720QM OC'ed for a longer run of prime. Sure you guys on the 3820QM can run an extra 100mhz faster but it should not affect temps that much (or so I'm thinking)
Icy diamond should be a 'lil better then my old MX-2 paste (I probably have had the same tube for 5 to 6 years now and its finally almost empty!)

The nice thing about the mx-2 is that it is supposed to last for 7 to 8 years before you need to reapply new paste and while I cant confirm if that is true, I can say that I never had to repaste anything I've done with it unless I had done a bad job at applying it in the 1st place.

 

xmadror, when you apply a thermal paste on cpu die you make it like 2 pea dots or 1 line across the whole cpu die?

 

since its a rectangle shape, I applied one line in the middle on about 2/3 the length of the cpu die.

 

So I thought I had this temperature thing licked but apparently not. I was playing League of Legends and my FPS would drop from 250+ to 30. I ran a few benches with Hardware
monitor only to see my CPU hitting 99c and my GPU almost hitting 90c. I think I have a temperature problem :-/



I also noticed that my fans were not running on high... I hope there is not real issue here :-(

 

are you running a fan controller software ? and yes i would say them temps are very high ! any Ocing ?

 

The screen shot shows HWiNFO64 running. That may be why your fans did not kick on. I use HWiNFO64 100% of the time, but you have to pay close attention to how the fans act when using it. Sometimes they don't run. Either set up a custom fan profile or click on the fan icon and select the "System Auto" button, and click on it to be sure the fans will operate before doing anything intensive like gaming or benching. See attached screen shots.


(sidebar gadget is also HWiNFO64)

 

I don't have the temps on mine but When I play just about any intensive game, after about 15 or 20 minutes the system slows to a crawl. I found that If I use Speedfan to manually crank the fans up I have no problem at all. I did the GPU throttle flash. I am ok with manually cranking the fan though speedfan is not always reliable. Does anyone have any other suggestions?

 

My suggestion is to stop using SpeedFan and use HWiNFO64 instead. I have had significant issues with it causing fans to not run frequent. It works OK to force fans to run at 100%, but its fan profiles often do not work properly. I also found it cause severe DPC latency when I had GTX 580M SLI. Both are reasons I stopped using SpeedFan.

HWiNFO64, if set up correctly, rarely has an issue. (See screen shots in my previous post.)

 

My CPU is running 4.5Ghz across all four cores. The GPUs just have the modified vBIOS that bumps the voltage and increases the clock slightly. When I got the severe throttling of the GPUs HWinfo64 was not running. I will force the fans on and see if that makes a difference.

 

HWiNFO64 had to have been running when you posted the screen shot showing the temps were high, otherwise there would have been no screen shot.

How long has it been since you cleaned out the fans and heat sink radiators? Depending on the working environment, they can get enough dust and lint accumulated to raise temps in a month or less of frequent use.

Edit 1: Throttling is normal (and a good thing) when temps are higher than they should be. Assuming the fans are operating correctly, you should not be experiencing GPU throttling. My GTX 580M cards never, ever, ran hot... even when OC'd and overvolted. And, they never had a throttling problem with P0 and P1 set at equal speeds. If cleaning the fans and heat sinks has already been performed, you may have an issue with needing to re-paste or the thermal pads may not be optimally placed. The left GPU should have a back plate covering the RAM on one memory bank that runs the length of the card. The right GPU back plate should have a heat spreader that covers both banks of memory.

For the CPU, you can change the thermals with the unlocked vBIOS. Raise the the throttle trigger from 80°C to 90°C and set the thermal max to 120°C and that may circumvent CPU throttling. (Raising it to 120°C for the max will not change the point for thermal shutdown. That will still occur at 100°C, but it will help to circumvent CPU throttling as it approaches 100°C.)

Edit 2: Also, as it is summer time, what are your ambient temps? If you are working in a warm environment, everything may be working perfectly and the environment may simply be too warm for intensive CPU and GPU utilization. If your ambient temps are much above 70°F, this could be what is causing issues, especially if you are overclocking. Running CPU and GPU beyond specs demands a cooler working environment. You'll need to either change that condition or use less aggressive settings.

 

I was running HWinfo64 while benching after I noticed a problem but I first noticed the problem while playing League of Legends, which I don't think is a very demanding game. I would get 250+ FPS and then out of no where it would revert to P12 state (according to Nvidia Inspector) and produce only 30FPS for a few minutes before returning to normal. I ran Hwinfo60 with fans on full and ran 3DMark11 this morning and got a score of 7160 with GPU temps at 44 max which is great. I went to run the bench again and I got a blue screen error 124. I have these SSDs installed and every now and then I would have an issue but I don't know what it is given the temps I am sometimes reaching.

 

The 0x124 STOP error is generally attributable to the FlexVID (voltage) being too low, Pri Plane (amps) being too high, or a combination of both. Of course, increasing that can exacerbate temps. What are you using for Flex and Pri Plane?

See edits to my previous post.

I am defining Flex and Pri Plane for the benefit of readers that may not know what we are talking about. I know that you know what those terms refer to.

 

I forget what I set the pri plane and Flex at so let me check but I am indoors and it is 78F inside so that should be fine. I did not know about the back plates, is that on the underside or behind the keyboard side? I should look at some pics when I re-pasted the CPU. I had thought that paste was sufficiently applied and the initial few weeks were fine where the temps never went over 85c and I was able to overclock it (while on a cooler) to heights I haven't been able to (4.8Ghz) but now 4.5Ghz is the most I can do stable... well it use to be stable, doesn't seem so now. I will post back shortly.

Pri plane = 1050

Flex = 15

 

No, 78°F is way too warm for that kind of overclocking. You need to be in the 70°F to 72°F range, or lower. Yes, the back plates are on the underside of the GPU. These are what the heat sink screws are attached to. You've seen my 580M benchmarks... off the hook awesome... and I never saw 85°C GPU temps benching on normal air or gaming. I think the highest I ever saw was 79°C to 80°C and that was due to high ambient temps. You'll not get any stability with your CPU beyond 4.5GHz with your current ambient temps and you'll have to lower those clock speeds as the weather gets warmer.

A laptop cooler is not going to help because they are ineffective. Moving more hot air doesn't lower the temps. Even using my U3 with precisely placed fans and an aftermarket higher CFM fan added for the CPU, the benefits of a cooler are not much greater than simply propping up the rear of the laptop to improve air flow. It's more of a novelty item than it is of great benefit.

Try lowering Flex to 10 or 12, set Pri Plane to 950, and put your multipliers at 44x. Also lower your BCLK to 100.2 (in the BIOS, not XTU) and this should help. The difference in performance will be negligible.

What paste did you use? After trying many different brands, I have found that anything except for IC Diamond seldom lasts more than a month or two of overclocking and intense gaming/benching. They all start out effectively cooling and degrade over time... some degrade more rapidly than others. Also important when re-pasting is to be careful to tighten the CPU heat sink screws in the X pattern, not overtighten them, and do so gradually so the heat sink sits flat on the die when you are done. If you tighten one screw even a quarter turn too far by accident, the heat sink will be cocked and not mate perfectly with the die. This is an easy mistake for pros and newbs alike... it's happened to me more than once and I take extra time and effort now on this step. It's a pain in the tush, but worth it.

 

I used the Liquid Pro Liquid metal thermal paste. I will try your settings above when I get back from errands but my BCLK is at 99.98 already.

Well I haven't had much time for testing but the temps seem to be down a bit. I will run benchmarks later today and see what happens. Thanks for your help thus far Mr.Fox

 

I am on my Galaxy Nexus posting, so bear with me.. lol

These were my GPU Temps after running World of Warcraft's beta, MoP. Recalling these figures from memory, I have a Screenie at home.

M18x R1

GPU #0 - 6990M
Diode - 73°C
Core - 75°C
MemIO - 82°C
ShaderIO - 100°C

GPU #1 - 6990M
Diode - 78°C
Core - 78°C
MemIO - 86°C
ShaderIO - 78°C

GPU #0 is the same with X-Fire disabled too.

This was about 3 hours into gaming that I watched the temps. Running Fur mark gives slightly higher temperatures. Nothing is blocked, fans are cleaned out very well.

Should the Shader / TSO#2 be that high under load compared to the rest of the card? Its nearly a 30°C difference on gpu#0 and under 10°C on GPU#1.

 

I am thinking on your primary gpu the 100C shader is due to one of the thermal pads being missing or applied improperly from factory.

 

OK, I have the screenshot at hand and these are the exact temps:

GPU #0 (Lowest to Highest is a range of 22*C MAX)
Diode - 78*C
DispIO - 79*C
MemIO - 86*C
Shader - 100*C

GPU #1 (Lowest to Highest is a range of 10*C MAX)
Diode - 83*C
DispIO - 82*C
MemIO - 92*C
Shader - 83*C

Even though its a tad hotter than I had posted prior, the fact the Shader is 22*C Higher than the Diode (Shader and Diode are equal on GPU#1) means that something is wrong from what I understand. Pads missing or applied wrong is what I was thinking, but didn't wanna jump to conclusions quite so quickly.

I am not getting any thermal shutdowns though, which is good, but it feels abnormally hot on the chassis.

As far as I know, isn't the MemIO supposed to be the hottest sensor by atleast 10*C normally?

 

So, this just started happening all of a sudden? Had you made any changes with pads and paste to trigger this? If the thermal pads were not applied correctly from the factory it should have been having issues all along. It could be simply a matter of degrading pads and paste through use over time.

 

Thanks for responding Mr. Fox, my machine came back from the depot recently. The CPU runs cooler, the GPUs seem to run cooler according to the thermal diode sensors, but GPU #0 runs hotter on the shader sensor.

I ran furmark to stress it and both cards acted in a similar way as in the WoW Beta, but slightly hotter.

Any recommendations to make resolving this go smoother than India-Support-land takes care of it? Thanks again for responding pretty quickly Mr. Fox.

P.S. the machine does not crash but gets extremely hot to the touch on the keyboard and at other spots, fans forced high don't phase it, not good.

 

Well, there is no telling what they might have done at the depot. Did the repair involve GPUs? In any case I would recommend getting some Fujipoly thermal pads from FrozenCPU and some IC Diamond thermal paste and opening it up. There are some old threads in the M18x and M17x forums with photos of everything that needs pads, but you may need to search the forum using Google to find them. Click the link in the left side of my signature for more info.

Here is a photo of a 580M and 6990M heat sinks showing placement of thermal pads from the factory.

 

The repair involved replacing the motherboard (Bios update failed, corrupted it, and computer rebooted with a corrupted bios).

So I am assuming that everything was taken off, repasted and put back on.

 

IC Diamond has cure time?

Used IC Diamond and did an x pattern - prime95 temps keeps rising till 97 celsius during 25mins of prime95, then i repasted checked contacts - same thing. What's wrong? Does IC diamond needs a cure time?

 

If you are in a warm climate, this may be entirely normal. An ambient temperature of 70°F (21°C) or less is appropriate for optimal cooling and system function. Using a laptop cooler and manually boosting fan speeds will offer reduced benefit if the air being pushed through the heat sink radiator is already warm.

What speed is your CPU running? What are your voltage and amperage settings? These will also have a huge effect on your results. If your ambient temperatures are higher than those mentioned, you may need to use less aggressive settings to maintain control of your CPU temperatures.

There is a cure time for IC Diamond, with minor improvement occurring shortly after application. You would not see a significant reduction with cure time.

The stock Dell thermal interface material (TIM) would have been about as effective as using IC Diamond, and little more durable. Their cooling properties are similar and the stock TIM achieves a very uniform application. Is there a reason you did not use it? Did you remove it before applying the IC Diamond? Using an "X" pattern may have caused an air pocket in the paste, which could lead to uneven heat dissipation. It is best to apply a thin bead the length of the CPU die to ensure even distribution of the paste over the entire die surface. It is also very important to tighten the screws evenly and gradually in a cross pattern so the heat sink ends up mating perfectly flush with even pressure. If you tighten one heat sink screw just a small amount further than the others, this can also lead to ineffective cooling as the CPU die and heat sink will not have even contact at all four corners of the die. This is very easy to do by accident, even for those with a lot of experience. It sometimes requires a little bit of trial and error to achieve the best result in aligning the two components.

 

To be honest there is a little to non improvement from the 2 pipe CPU heat sink to 3 pipe one why ? because you have two big pipes on one and 3 little pipes on the other one but with the same amount of copper on both if you gain 2C lower on the 3 pipe that's about it won't expect more there is tweaks to improve the temps more but you need washers on the retention springs to create more pressure and you can also lap the heat sink to mirror finish that alone will give you better results than any other mod

 

I used 2 pea blobs method, not the entire line. Is that ok?

 

Probably so. If you have a reason to take it apart, try the thin line method next time. When you have a space between the blobs it increases the chance of having an air pocket, but it does not necessarily mean there will always be one.

The other thing is that you have to look at this strategically. So, you hit 97°C after 30 minutes of wPrime? That would most likely be far more CPU stress and higher temps than you would see in normal use. I would not worry too much unless you are hitting 97°C doing normal stuff, like gaming, web browsing, burning DVDs, editing videos and photos, etc.

Also, from a strategic perspective, a 2°C to 4°C improvement using, for example, the 3-pipe heat sink is a big deal. Along with a drop of 1°C to 2°C using a laptop cooler or propping up the rear of the laptop, dropping another 1°C to 2°C by increasing fan speeds, a 3°C to 5°C improvement by working in a cooler environment, another 3°C to 5°C improvement from cleaning the dust out your heat sink radiators, (if they need it,) etc... My point here is that these are all very small changes when you don't think bigger picture, but the cumulative effect of doing all of the right things for cooling is huge. All it takes is a change of 1°C to mean the difference between continuing to run or thermal shutdown. If you're not getting hot enough to cause thermal shutdown under an extreme load like running wPrime for 30 minutes or longer, I would have to say that your system is working correctly.

 

Hey speedy can you please post a pic for the washer/retention spring mod? Thanks in advance bro.

 

Ok, after some cure time i got 92 celsius on CPU with prime95 30 minutes (3piped heatsink) is that ok? Ambient temp is like 28 celsius.

 

Hey bro... stop running prime95 and see how your system behaves playing games and running benchmark programs like 3DMark11 or Vantage. I would say yes, 92°C is fine for running prime95 that long, but what really matters is real-world experience. The prime95 software is taxing your CPU more than normal, so that is why I say you're good to go at 92°C. Any temp that is low enough to not cause thermal shutdown would be OK running prime95 for 30 minutes.

 

my 18x r2 with 3720 and single pipe cooler is running about 75c during gaming and its been repasted by hidevolution.com, does that temp seem normal? pretty cool enviroment, not much heat and its proped up in the rear. maybe time for a 2-3 pipe cooler or repaste? just curious if these temps are normal for gaming. (battlefield 3 and Tera Online)

 

There's nothing wrong with 75°C during gameplay. You don't really need to do anything with that temperature. If you want to have lower temps, then you need to get a better heat sink. Simply re-pasting probably will not change anything from where you are at right now. The one-pipe unit is OK for a non-XM CPU under normal use scenarios, but it's not as effective as one of those better heat sinks.

 

whats the cooling difference between a 2 pipe and 3 pipe? i cannot find a 3pipe other than ebay, and i can get a decent deal on the 2 pipe from Hidevolution

 

It's not expensive. Just buy it from Dell. Speedy says they are about the same. They probably are during normal use without an XM, but I can overclock with better temperature control with the XM CPU. Johnksss found his system ran much cooler with the 3-pipe than it did with the 2-pipe heat sink also. On a 3720QM CPU it may not be an earth-shattering improvement, since they normally don't run fast enough to get extraordinarily hot.

Here's the info for ordering a 3-pipe from Dell: 0MXJ5 Heat Sink Assy, CPU QC XE M18XR2. I comes with a thermal compound already applied to the surface which mates with the CPU die. No aftermarket paste is needed. The stuff Dell uses will last longer than aftermarket paste and it does a good job of cooling.

 

if that was meant to be a link it did not work, i guess im going to just buy the 3 pipe, is this considered a void of warranty doing the install myself? i'm not a novice at computer assembly, but just curious.

 

Nope, you can do anything you feel like doing to your Alienware. Nothing voids the warranty as far as upgrades are concerned. User-induced damage is not covered under warranty. Be sure to hang onto your old parts in case you ever need to send it to the depot (heaven forbid) or you may not get your better parts back... especially if it needs to be exchanged with a new system.

As far as doing your own repairs, unless you are dealing with a novice Alienware Support rep they will generally provide POS (parts only service) if that is your personal preference.

Not a link... just changed the font color so it would stand out. I copied and pasted the part info from my http://forum.notebookreview.com/ali...dge-cpu-heatsink-7970m-crossfire-upgrade.html thread.

 

with about 45 minutes of metro2033 i'm reported max temps of 85, that doesn't sound normal to me, does it?

another oddity is the processor is running prime 95 right now at 100% load and its showing the cpu clock in realtemp to be between 2990ish to 3049ish, i thought this was supposed to go upto 3.6ghz, this is the 3720qm, btw.

 

You need to adjust the long duration power limit (and maybe short duration power limit too) to get the cpu to sustain 3.4ghz on all core otherwise it'll back down to 3.0 after a short while.
I think 55W should get you there but it might need a 'lil more then that.

to get an idea of what I'm talking see here:

 

Can anyone with 3720QM or 3820QM with 400mhz overclock with 3piped heatsink tell his cpu temps? Mine during Dirt Showdown was 87Celsius isn't that too much for that game? (i have 3 piped hs)

 

is there anything wrong with setting the mutli's to 38/38/38/38? with a 3720.

 

No, enjoy the speed, but if temps are going too high your card will throttle and reduce turbo boost.

 

Both of your observations are normal and there is no reason for concern. Hitting 85°C simply means Metro 2033 and your CPU are both working properly, and that temperature is not a problem.

With a non-Extreme CPU you will not have the same level of control over clock speeds. You can get the new unlocked BIOS for the R2 and see what adjustments are available, but what you are seeing is EIST (Enhanced Intel Speed Stepping) an c-states operating normally. With an XM CPU you can work around those things with an unlocked BIOS, but you may experience a thermal shutdown at 100°C if the CPU is not allowed to downclock itself to circumvent heat build up.

 

i just feel the top of the keyboard area of the laptop where the power button is, getting pretty uncomfortably hot, and it bothers me that it may effect its longevity, so im considering getting the 3pipe cooler. with stock TIM and keeping a moderate overclock

 

Like I've said before, I never go above 80-81C (even in prime95) and depending on the game I usually get something between 67-75C
I have a room temp of 24-25C in hot days and I use a CF storm notebook cooler to help some with a 2 pipe cooler and 40/39/38/38 turbo multi.

you could go up to 40/39/38/38 it will not run hotter because of that since the 40 and 39 multi will only be used with 1 and 2 core usage and what really heat up the cpu is when the 4 core are used at the same time. Also make sure your Flex VID override is at 0. (unless your cpu need some extra voltage to be stable but I doubt it will)

Never really felt that with mine even when I only had the 1 pipe heatsink. Do you prop up the back of the laptop ? or use a notebook cooler ?
In any case the 3 pipe heatsink should lower the temps by more then 10-12C(thats the drop I had with the 2 pipe).

 

i'm waiting for a call from Frank L from alienware sales to sell me a 3 pipe, and i do not use a notebook cooler and i do have it propped up...