i7 920XM in W510, dynamic clockspeed (clocks down) when hot.

[playersnoopy]

well if you're talking about going down to a 720/820QM then the difference is the Max TDP from 55 Watts for the 920/940XM to 45 Watts. That definitely partially accounts for things, not to mention things run slower so less heat... but hard to say if it's truly an overheating problem when it comes to running Prime95... even a lot of people using desktops won't run it for more than 4-6 hours since theirs will run pretty hot too. Notebooks just have a ventilation issue along w/ the smaller heatsinks. Lenovo especially... The only time to worry about things is if your system clocks down/heats up immediately when doing anything labor intensive, like playing a game or running your CAD programs. I can play the new Medal of Honor and never peak over 75C CPU, and 78C GPU, which is pretty good for a notebook. Even though I'm running my 940XM overclocked at 3.8Ghz and GPU overclocked to 690Mhz w/ GPU memory overclocked to 945Mhz. Was surprised at overall temps... the major difference between this and say other brands is idle temps, about 5-10 degrees hotter at idle than most.

[Adda]

Another update.

I bit the bullet and went with the replacement of having a slower CPU. Really going to miss the speed.

This time, same test, the 4 cores are around 79-81 degrees C, which got me thinking...given that nothing else got changed, maybe 93-94 C was "normal". Having said that, I just couldn't risk having other nearby components getting that hot, 90+ c is just not healthy for a laptop.

As for the heatsink & fan, didn't replace it, so I can't tell what the differences are, if any. The tech guy did say that, from the internal notes, both thermal units are supposed to function identically and there's no thermal/cpu requirement that say they should use one over the other.

I think you made a good choice, the W510 cooler is simply not good enough to deal with 55w CPU's it seems.
Besides, the 720QM still is a fast processor after all.

well if you're talking about going down to a 720/820QM then the difference is the Max TDP from 55 Watts for the 920/940XM to 45 Watts. That definitely partially accounts for things, not to mention things run slower so less heat... but hard to say if it's truly an overheating problem when it comes to running Prime95... even a lot of people using desktops won't run it for more than 4-6 hours since theirs will run pretty hot too. Notebooks just have a ventilation issue along w/ the smaller heatsinks. Lenovo especially... The only time to worry about things is if your system clocks down/heats up immediately when doing anything labor intensive, like playing a game or running your CAD programs. I can play the new Medal of Honor and never peak over 75C CPU, and 78C GPU, which is pretty good for a notebook. Even though I'm running my 940XM overclocked at 3.8Ghz and GPU overclocked to 690Mhz w/ GPU memory overclocked to 945Mhz. Was surprised at overall temps... the major difference between this and say other brands is idle temps, about 5-10 degrees hotter at idle than most.

I disagree, apps like Prime95 and FurMark will show how much heat your system is capable of producing, and if the cooler can't deal with it, you need a better cooler or more energy efficient components.
If an overclocked system can't run these tests indefinitely, the overclock can not be considered stable.
Remember, hot summer days will make the system heat up even more, so even though these tests make the system run hotter than during realistic usage, it will show you whether your cooling has the headroom to deal with hot weather.

That said, the cooler in W510's seems to deal well enough with 45w CPU's if the cooler is properly fitted.

[Chatbox]

Yeah, I think I made the right choice too.

Running the i7 920XM at close to 95c is just not good for the laptop at all.

The i7 720QM seems to be a good match for the W510 included thermal unit / HSF, and I'm now extremely happy with the stability.

That extra 10watts of heat is just too much for the stock HSF to handle. It just wasn't dissipating the heat away from the system fast enough, and that's why the temperature was just keeping on rising. I'm guessing the W710 doesn't have this problem though, is this right?

[playersnoopy]

I'm happy running my i7 940XM, I have it overclocked to 3.86Ghz on turbo boost and unless I plan on running prime95 for a little while it's stable to my needs. Play any game or run 3D apps and temps never hit over 85C on extreme cases, GPU around 80C max but most of the time temps are in the mid to high 70s. Crysis, StarCraft II, Final Fantasy XIV all runs smoothly... and the new Medal of Honor runs perfectly.

[Chatbox]

I'm interested in finding out if your lappy can survive the:
1. prime95 max heat test 8 threads, alone.
2. Furmark (1280x1024, window mode) heat test, alone.
3. Prime95 max heat test 6 threads, with furmark (1280x1024, window mode) heat test, both running together.

If possible, post some screen shots too, it'll be great to find out what the temperature trend is and how long it can survive for. Plus, if everything is stable, say for at least an hour, then I may consider fitting mine with the 940xm in it. Can you also post your HSF FRU number and running rpm as well, please? Thank you! Thank you!

[Chatbox]

I disagree, apps like Prime95 and FurMark will show how much heat your system is capable of producing, and if the cooler can't deal with it, you need a better cooler or more energy efficient components.
If an overclocked system can't run these tests indefinitely, the overclock can not be considered stable.
Remember, hot summer days will make the system heat up even more, so even though these tests make the system run hotter than during realistic usage, it will show you whether your cooling has the headroom to deal with hot weather.

That said, the cooler in W510's seems to deal well enough with 45w CPU's if the cooler is properly fitted.

Hi Adda, I agree with you on this one. There's a reason why a stress test is called a stress test. A system, any "stable" system, SHOULD be able to handle STRESS, any kind of stress that's thrown at it, as long as it doesn't exceed the technical design specs. The system, W510 in this case, needs to be able to dissipate 55w of heat and the heat from the GPU AND other internal components, without having to resort to self-protection mode (speed throttle down, system shut down...etc). Having said that, some users just don't have the need for having their system(s) facing such scenarios. Me, I just like planning for the unknown, and since it's a work system, I would definitely like to have it as stable as can be / should be. There's no point in paying so much for a system if the excuse is going to be "I don't need it running at full speed anyway...so I'm fine with it", with effectively means "I just paid for things that I don't use/need".

There are typically three things that a stability stress test can tell you (or any test for the matter):
1. What the system can handle/do.
2. What can't the system handle/do.
3. Can you do anything to make point 1 into point 2. And typically, having a "workstation" (which the W510 IS) means that many aspects of the system have been thoroughly tested, which makes it stands out from your typical laptops. It should be a more capable, more stable, more robust system...with a lot less weaknesses.

[Adda]

The W701 has two massive coolers, one for the CPU and one for the GPU, so I can't imagine it would have any heat problems.
I don't think it was a wise move by Lonovo to put the 55w i7's in the W510, they belong in 17"+ machines, in my opinion.

[Chatbox]

Throughout the entire week, I've learnt the following about the W510 in respect to the CPU, GPU and GPU memory cooling. Hope this information will be handy to some W510 owners out there.

1. If your system is hanging (cap lock will still respond, screen will freeze, but HD was still ticking and divx was still playing in the background - sound was coming out) randomly (in Windows, at BIOS startup, while loading Windows, during shutdown) even with a fresh windows installation, latest drivers from Lenovo system update and windows update, it's likely that (at least in my case) the GPU memory's thermal pads are not contacting the heating properly.
2. At a room temperature of 23 c, palm rest taken off (exposing the heatsink and fan memory and systemboard, running furmark 1280x1024 window mode, MSAA = 4x, extreme burning, the max temp is 69 c (reported by furmark). This is the best (lowest) temperature I managed to get the GPU to run the stress test at. Previously, it was at 78c (out of the box), then I got that down to 71c (I was happy then...system was not hanging randomly anymore...then boom, 16 hours of test later, it hung. Then trying again, now it's at 69c this time.
3. With the palm rest taken off, Windows and Furmark running, you can gently press around the GPU heatsink to see if applying pressure at particular areas would change the temperature of the GPU. For me, I found that pressing the copper pipe (the area directly at where the GPU is having contact with the heatsink) helps to lower the GPU a lot.
4. Sometimes, even with good temperature from the GPU, the laptop will still hang, and it's likely because the GPU memory chips is not getting rid of the heat properly via the thermal pads to the heatsink. Make sure the thermal pads are not too thin because that will limit the efficiency of the heat transfer. If the pads are too thin, you can "GENTLY" peel them off, and slowly just placing them back onto the chip surface. You'll noticed that the pads have gotten a bit thicker again, as they're kinda like sticky, rubbery foam. Also, it seems the thicker the pads are, the better they are at having decent contact when pressure is applied. I'm buying some new thicker thermal pads, and they should arrive within the next two days or so, and I'll be testing out this theory / observation.

This whole experience is telling me that the HSF design, and thermal transfer design is really fragile (works within a very small tollerance window). And I don't think Lenovo did a whole lot of testing when the systems were assembled. So many users are getting systems with such a varied build quality control (or the lack of).

[tklonowski]

I was thinking of purchasing a Thinkpad W510 (43192PU - i7 720XM - 1920 X 1080) in the near future and eventually upgrading to a i7 920XM cpu. The information regarding overheating in the CPU and GPU in nearly new W510 laptops is quite dismaying. Does anyone have an idea as to the frequency of these problems in the W510 and if these issues are a show stopper when considering the purchase of a Thinkpad W510? Thanks.

[Chatbox]

Having called up Lenovo support, and the guy over the phone said to me "With the overheating issue, we can send out a technician to you, but it could be due to a product issue." I'm guessing Lenovo is aware of this and judging by the tone over the phone, I'd say it's quite common (after all, people who call in must have one issue or another), and there's no real permanent fix for it.

Some people might not stress their components out like a few of us do...and are quite happy with the product right out of the box. And that's ok for their usage behaviour. For me, I'm just picky, just need to have the system able to handle whatever scenario I throw at it.

tklonowski, the w510 is a decent product, and with the 720xm it's a good match. 920XM was definitely killing my laptop (just my experience). Getting the thermal unit "just right" with the cpu, gpu and gpu memory was a lot of hassle (I've been trying to get the components temperatures as low as possible...), but if you're a handy person, then you can definitely fine tune it yourself and have a great unit working.

[Adda]

Interesting, now you can get W510's with dual core 35w CPU's.
If those models use the same cooler as the quad core equipped models, the cooling performance would be overkill, as it deals with 45w CPU's well enough.

Edit: Maybe the other W510 cooler is for these models.

[nullface]

I havent read the whole thing, but... seriusly?

How can one wonder why a laptop is getting hot when running 8 thread max heat prime95 for two houers?

8 thread max heat prime95 is a wildly abnormal work load (add furmark to strees the GPU and it gets ridiculous), one would never get these kinds of load in a normal work or even gaming environment.

Do a temp-log and do some intensive gaming and let us know what kind of temp's it hits, if it is still around 90C i see the problem, but as long as the 90C are only achived with prime95 i think this is just stupid.

[Chatbox]

I havent read the whole thing, but... seriusly?

How can one wonder why a laptop is getting hot when running 8 thread max heat prime95 for two houers?

8 thread max heat prime95 is a wildly abnormal work load (add furmark to strees the GPU and it gets ridiculous), one would never get these kinds of load in a normal work or even gaming environment.

Do a temp-log and do some intensive gaming and let us know what kind of temp's it hits, if it is still around 90C i see the problem, but as long as the 90C are only achived with prime95 i think this is just stupid.

Agree that this test is abnormal. No disagreement here. The point is, if the system can function WELL at this abnormal test, then you can say with HIGH confidence that it will be rock solid in normal work situations. It's a matter of having a highly stable upper boundary when the system is under stress.

As to what "normal" work is...everyone buy their laptop (thinkpad or not) for their own reasons and usage pattern, requirements and expectations. I just happened to have higher expectations (less tolerance) than some.

[playersnoopy]

So after trying several different brands of thermal compounds and applying it differently according to different websites. I finally came to one that worked best for me and decided to put Prime95 to the test like you. Initially it would climb hotter and hotter but only because my room wasn't properly ventilated; so I'm wondering if yours was?... As long as I have my door open and a fan turned on so the room is vented I can run Prime95 on 8 threads Max Heat no problem.

I am running i7 940XM OC'd to 3.87Ghz, but on 8 threads it can only run at 2.13Ghz, 16gb RAM, 240gb SSD, GPU OC'd to 980Mhz core/945Mhz Memory.
My temps Current Low High
Core #0: 90C 50C 92C
Core #1: 92C 48C 92C
Core #2: 91C 52C 91C
Core #3: 92C 50C 93C
Using Core Temp 0.99.7
I've had some thermal application overheat on me in 15mins, this has been running for almost 5 hours so I'm pretty confident with it, as the temps just go from 90-93 and back and forth, but averaging 91-92. Biggest issue is room temperature I guess, also blowing the air away from the exhaust area and into the room.

For the GPU I've only ran FurMark for about an hour but it's been at 82C for almost the whole time, peaked at 83C at some point. This is probably the greatest improvement I've had. I'll consider running it for a whole day maybe during next weekend or something. The only improvement I have left is seeing if I can find a different fant/blower to replace with a higher rpm fan. Looking at Sunon's site hope I can find something that can do the trick.

[Adda]

The test we have been running is Prime95 with 6 threads in maximum heat mode, leaving one core free for FurMark witch was run in extreme burning mode.
This will test if you cooler can handle the combined heat output of both CPU and GPU.

[playersnoopy]

FurMark in 1280x1024 windowed?

[Adda]

Yes, resolution makes no difference in temps though as far as I can see.

[playersnoopy]

lol that's a bit ridiculous on cpu load, I've never ran both furmark and prime95 at the same time... 6cpu load when using prime95 is pretty much the same thing as 8cpu since it still uses 4 cores, plus furmark usually utilizes only a single core at almost 85-90%.

If you guys have been complaining about not being able to do a stability test of FurMark and Prime95 at full load at the same time... then lol... have fun.

[Adda]

The W510 is a professional workstation, so it should be capable of chewing through anything you throw at it, at full speed and without breaking a sweat.

Edit: After tinkering with the cooler my W510 eats up and swallows this test without complaining.

[Chatbox]

Man, someone should sticky this thread. It's like the holy grail in W510 cooling techniques.

@playersnoopy, here's my testing environment:
1. Wooden (not sure what kind of tree/wood it is) table (no, don't cheat...don't say you have a nice expensive brushed steel table to help with your cooling...haha), laptop in the centre, nothing else is on the table.
2. No airflow in room (doors closed, no opened window, no fan blowing at the laptop, no laptop cooler).
3. Located away from sunlight.

I'd say you've got the GPU running at a decent temperature. Typically somewhere around 81 to 86, give or take.

About your initial unsteady temperature, it's expected that when you apply new thermal paste/compound you'll notice variable thermal characteristics (thermal fluctuation) as the paste is not settled yet. Usually (according to those in the overclocking scene...highly demanding in the cooling efficiency department) it takes about 100hours for the compound to provide a good thermal interface between the heat source (CPU/GPU in this case) and the heatsink. Also, numerous heat up and cool down cycles. The other thing you might want to take into consideration is that thermal pads (than paste) will be better in the long run as pads don't evaporate at high temperature over long period of time.

Lately (yesterday), as I've been fiddling and adjusting the heatsink so frequently, the thermal pads for the GPU memory modules are totally screwed now. I went and bought some replacement pads (0.5mm thick)...and retesting the laptop (probably another week before I can trust it with work data).

Trust me, I don't take a whole lot of joy doing this really...but it seems like a necessary step as the "out of the box" W510 just wasn't good enough (in my case) with its cooling ability.

[playersnoopy]

@playersnoopy, here's my testing environment:
1. Wooden (not sure what kind of tree/wood it is) table (no, don't cheat...don't say you have a nice expensive brushed steel table to help with your cooling...haha), laptop in the centre, nothing else is on the table.
2. No airflow in room (doors closed, no opened window, no fan blowing at the laptop, no laptop cooler).
3. Located away from sunlight.

lol why are you blowing up on me... I was only making a comment on whether or not it was getting too hot in the room after the 2 hours you were talking about... my last remark was to Adda. But if you are also having issues only when trying to run Prime95 AND FurMark then no easy solution. You can search for the threads about the issue of FurMark and Prime95 with throttling on the ASUS notebooks and try to disable the throttling, and go fromt here. I only said the part about the room because you said you got it to run for 2 hours... and that just seemed like a long time for those apps to run and a possible explanation could be the room temp.

I understand some compounds require burn in time and such, 2 of the type I had didn't require any, the other I've allowed a few weeks already. My fluxutations were with 1 specific type and just 1 of the cores weren't getting good contact and such. I only said that I finally got a good result.

I was actually trying to help along the way but oh well... mods can remove all my comments, I'll leave this thread alone.

[Chatbox]

My apologies, didn't mean to come across like that. It was meant to be read with a j/k kind of tone.

I really want to find out if any W510 out there can survive with the i7 9xx series procs, so maybe that's why it seems I'm always replying to you and stuff. You know, as my previous 920xm was an ES, I really am interested in finding out if the production ones are any better with their temperature.

[jketzetera]

As the W510 is marketed as a workstation computer offering uncompromised performance (not my words, Lenvo's words), it should as a minimum tolerate its CPU being fully loaded without resorting to thermal throttling that cripples the machine.

[Chatbox]

I give up..."my" W510 really does suck. On paper, it's awesome, but it's just not good enough in practice. It just can't function at peak performance for anything more than 30 minutes. The rthdribl test (full screen 1920x1080) will cause the system to shutdown within 15 minutes. The heat it generates will just kill it...basically a self-destruct..and power off. It's been a really disappointing experience.

[Adda]

It sounds like your GPU or video memory is overheating, are you sure the thermal pads for the memory has proper contact?

[Chatbox]

Think you're right about the vid memory getting overheated because it shuts down only for this specific test.

I think I might have made matter worse when I re-fitted the thermal unit multiple times, and the thermal pads are now somewhat deformed. So, in all fairness, it could be user error.

Need to find some new pads. Would you know if the pads are included if I order a new thermal unit from Lenovo?

Update: Lesson learnt so far -
1. Don't over compress the thermal pads. (I did, and that messed up a lot of things).
2. Use a decent thermal paste (I'm using the MX-2 at the moment, and it took a good 8-10 degress off the CPU and GPU).
3. Apply the thermal paste as a dot on the GPU, and a diagonal corner-to-corner cross on the CPU (that's just my way of doing it).
4. Make sure the heatsinks have a decent pressure pressing onto the CPU and GPU (don't over do it though...).

[Adda]

Yes a new cooler as new pads on it.
You could also try and peel off the ones on your current cooler, hang them by the fingers for a few secs, then they'll contract back in to shape and then you can reapply them.
I have done this quite a few times and the pads on both my coolers is fine, just make sure you have very clean and dry hands when doing this.

[Chatbox]

Hum...this sucks...(see the following).

Please wait while we find an agent to assist you...
All agents are currently busy. Please stand by.
An agent will be with you in a moment. Thank you for your patience.
You have been connected to **<rep>.
**<rep>: Thank you for choosing Lenovo chat. My name is <rep> . If you are placing an order online, please attach my Rep ID: nnnnnnnnn in the designated area at checkout near where you enter your credit card information. This will allow me to help manage your order from my end. My direct phone line is 1-866-426-9292 ext. <nnnnnn> if you wish to contact me. What may I help you with today?
Customer: I'm having overheating issue with my W510 (4318-CTO)
**<rep>: please contact tech support immediately at 1-800-565-3344
Customer: And it seems like it's a product design issue.
Customer: Tried getting a repair engineer sent out to me a couple of times in the past month. Same problem.
Customer: I'll contact them again...but doubt that will change anything.
**<rep>: I can provide assistance with your purchase of a replacement product and give you best pricing but unfortunately I cannot assist you with your technical issue.
**<rep>: Sorry
Customer: Does that mean the only option I have will be to see if a different model can fit my needs, and at the same time, will have to handle the current issue separately?
**<rep>: I am not sure if we got disconnected, but please contact me at xxxxxx@lenovo.com if you have further questions or if I can help expedite the ordering process by processing the order for you. If you are placing an order online, please attach my Rep ID: nnnnnnn in the designated area at checkout near where you enter your credit card information. Once you have processed the order, please email me your order number at xxxxxx@lenovo.com. This will allow me to help manage your order from my end. Thank you for choosing Lenovo!

[Chatbox]

The GPU memory is overheating again. The newly replaced thermal unit was doing the job quite well initially when it was first fitted into the laptop. However, now, two weeks later, it's overheating again.

The thermal pads' heat transfer efficiency seems to have changed...aged? Or humidity related?

[Colonel O'Neill]

See if your whole heatsink/fan assembly is clogged with dust or something. Optionally, replace the paste with AS5 (or, on the safe side, something electrically non-conductive like the MX-#).