Just to clarify... who thinks something is wrong with these temps?

Yeah. So... you see, I tried playing a round of Black Ops 2 (with everything but that, TS3, Steam, Tweetdeck and HWMonitor closed) on my machine. Unlocked fps, max graphics (16xQ CSAA), getting 180 fps constant. CPU is undervolted by -50mV on dynamic core voltage offset, which is why the max power draw didn't hit 47W. I think something be wrong. Temps were hovering around 80-82 for cores 1 & 2, 74 for core 3 and 70 for core 4. I am pretty bloody sure it didn't do this
1 - when I got my computer back a month ago
2 - before I sent it for repair in July

My PC loves to give me problems, I know, but hey, gotta ask. I thought it reached a new level of ridiculousness tonight when I had 3 streams open and was playing hearthstone (something I usually do) and cores 1 and 2 were hovering around 89-92 degrees.

Please note, I am using max fans (fn + 1). Yes, hot air is blowing out the back of my case. Yes, my laptop's back is raised, and I do have a fan blowing at it (which I usually do). In case one might say I'm somehow obstructing the airflow to the CPU, please note the GPUs' heat.

 

Have you ever cleaned out your vents?

over time, dust will build up on the heatsinks and slowly clog the fans.

If not, open up the back of your laptop (4 screws I think), while HOLDING DOWN THE FAN FROM SPINNING (like with the end of a pencil) blow air into the fans to blow the dust out, and blow out the vents.

Then with a lint-proof rag, wipe down the heatsinks to wipe off any dust collection they may have.

That should lower your overall thermals.

Otheriwise, it may be time to repaste.

 

Yes, it's all entirely clean. I checked last week. Also, I just tried an offline CS:GO bout. 3 rounds with bots. Game had unlocked FPS, running on a single card. Here's the results.

 

Have you used a 0.5mm thinner (takes it to 1 mm IIRC) pad on the cpu cooler vrm section?

 

Haven't touched CPU cooling. All I know is, before I sent the PC up in July to Mythlogic, it was a lot cooler CPU-wise. I had -80mV instead of -50mV then, but +30mV (not even hitting stock) is NOT a 15-20 degree difference in temperature.

Also, when i got it back, it was a bit cooler. I was thinking that somehow my paste job was bad and it had a pump-out effect, which is odd because it's supposed to be IC Diamond, but I didn't adjust a thing. I tried cleaning the fans out twice, but they were clean as a whistle. As you can see from the GPU temps in both links (though second one was single-GPU only) the temps are ice cold, especially for 780Ms in my climate. If dirt was a problem, those should be hot too.

*sigh* I guess I'll have to hunt someone down with some paste and run a repaste and see if that helps it for now until I can get some CLU. I've never seen such bad heat with ICD before, unless they used Arctic Silver MX-2 or something. As it is, I can't even play half my games or do anything while I watch a livestream, especially in the daytime.

Edit: I also wanna mention that my heatsink is the same as before I sent it for repair. Mine has a slight bend at the top right; not enough to obstruct cooling in any way (only affects the tip, kind of caves in exactly 1/2 of 1 air passage; nothing to blink twice at), but enough to make it a trademark. One would need to softly hammer a new heatsink to mimic my old one.

 

It is sad to see so many people talking about temps. I wonder if this is Clevo being lazy (this would be a shame - I recommended the brand to loads of people for cooling and price) or some 'scheme' by Nvidia and Intel to fry their components into an early grave so people don't do horrific things like keep their machines for 5 years.

I hang around on a Polish Clevo forum and there are mega threads with people literally milling their own copper pipes and fins with dozens of workshop pics. The most active threads are "paste x dropped my CPU temp from 100' to 97'!! YEAH!!!") That just can't be right.

 

Haswell is a hot beast, they do run on the warm side.

 

Well this is extremely abnormally hot for me, to be fair. I'm accustomed to (without doing anything different), like 15-20 degrees less heat on my CPU.

 

Well I managed not to throttle with 3.9ghz on the setup I tweaked.

 

What machine, CPU and apps? 5 years ago resellers were advising people who were not video editing for a living in Central Africa to RMA any machines where the CPU went above 70. Today some resellers appear to be telling people not to worry about 100' because even though it will indeed fry itself to death at those temps, that will only occur after the machine as a whole becomes obsolete (whenever that is).

 

95 degrees is the cutoff point at which mobile throttling occurs. The danger point is 105C or higher. My CPU usually doesn't cross 80 unless I'm streaming or playing BF4, and sometimes even when streaming it won't cross 80 XD.

Haswell is indeed quite hot, but IC diamond and Liquid Ultra and Artic Ceramique can handle the heat quite well under normal circumstances. I just wanted to be certain I was not going crazy is why I posted this thread... also to check if the CPU voltage I was getting was normal.

 

I'm not sure I agree with arbitrary red lines. I guess 110 would be perfectly fine for 5 seconds a year over 20 years. And probably regularly hitting 80 will mean something is xx % more likely to fail 1 year sooner.

Either way, imho going to uncle's welding shop with a bunch of copper pipes should not be happening for the ordinary gamer. People saying that unless they repaste every 8 months with some magic gold gel at 49.99 per syringe they get throttling - should not be happening...

 

Yup, those thermals are abnormally high.

I have the 8258-S with the 4810 i7. My CPU tends to hover between 62-78c depending on the game or task at hand.

The absolute max I've slammed it with some insane games that do ridiculous number crunching on physics that are non-graphical (thus the GPU can't crunch them).

In which case I saw a stunning 86c in my i7 after 6 hours of gameplay.

I've since dropped the 10,000 units in the field to the non-modded (cough) 6000 units and I don't get above 76. Guess there's a reason there was a cap

 

Well once again, I'm saying that I haven't heard much of that. I live in a place with 32-36 degrees celcius ambient temp with 70-100% humidity and with max fans, IC Diamond ($15 a tube) and just putting folded toilet paper blocks under the feet at the back of my laptop I very rarely crack 80 degrees... except for BF4, which heats things unrealistically. I honestly don't get that game. Aside from that, I've never had heating problems much, and my situation is quite extreme. If the people you are used to conversing with have much more extreme problems than I do, then... I don't know. I don't get how a large number of people can do something wrong, but something is wrong.

 

Well you know what it's like on the interweb. 20 people get a batch of warped CPU heatsinks or something and it sounds like its prevalent / severe enough to be contributing to global warming. THAT SAID I've been around these forums for each of my past 4 Clevo purchases and discussion of thermals has usually been more like "look at those dumb alienware customers - pay $800 more for same specs and they're 10' hotter!". Now it's more "is 90' the new 70' , or should I buy some copper and a blowtorch?"

By the way, the temps I mentioned in previous posts were ofc just arbitrary figures to make a point, I have no idea how common 80' really is on Haswell Clevos nor how many flight simmers regularly get throttled, so to speak.

The funny part is, I need a decent CPU to marry with a 980M GTX for X-Plane 10, and it just seems like there's no good choice at the moment...

 

Ha. The 4910MQ would be fine for you, but you'd need a system capable of fully cooling it. Haswell is a lot hotter than Ivy Bridge, and we don't actually know why. It just is. There's no escaping it. 80 deg is normal for haswell being pushed. 90 is not desirable in any way, but it is not exactly a "danger" either. If you mod in some cooling or are in a very cool room with an AC or whatever, you could even avoid 80 deg under some load. You could further avoid this with Liquid Ultra thermal paste; it's the clear-cut winner for paste jobs. Sometimes up to 15 deg C improvements over bad paste jobs, and usually a minimum of 5 deg C less under load than the best of IC Diamond paste jobs. And IC Diamond is pretty much the best non-liquid metal paste you'll get, so... yeah. It's very long lasting too if done right (which mine obviously was NOT).

Some warped heatsinks will never cool correctly though you are indeed correct. If a few people got some of those and no QA testing was done before sending it out, I definitely would understand your view.

Anyway, a P170SM-A or P177SM-A would be fine for cooling. I think one of the ASUS G series might also work, but they appear to love using HQ CPUs so you won't get anything nearly as strong. I also don't like recommending ASUS laptops. I know more than two people who've had them and have had breakdown of cooling capabilities over under a year of owning them. I also see lots of horror stories from this place. If you go AW, a AW18 or AW17 should work out well for you, and keep things cool enough.

 

Thanks D2 :thumbsup: I was going to go with Asus 751 this time round simply because Clevo no longer seemed to be on the ball with cooling and the Asus is a lot quieter (often a hallmark of competent thermal design), has better sound and is sexier.

However the off-topic discussion in the 980M release thread made me rethink the previous 'no-brainer' of 4GB (Asus) vs 8GB (Clevo) of vRAM. Seems lazy coders will make 8GB necessary sooner than I thought...
Also, we don't know enough about Win10 and apparently it's a load of work to chase up drivers for Win7 with the Asus. So looks like a 5th Clevo (P170SM-A) to add to my collection tomorrow. Clevo should really issue me with a loyalty/points/rewards card so I can get some air miles or something, lol.

Hopefully with a cooler GPU (980M) the CPU temps on the 4910 will be under less pressure!

 

It won't matter for a P170SM-A. The P370EM had shared cooling with the CPU and the Slave GPU. The P37xSM and all P37xSM-A machines needed more cooling for the 780M, so the CPU got shafted a bit, basically. You COULD use the EM type coolers, but they will make the slave hotter and the CPU cooler... for 780Ms or 880Ms it's likely not a good idea, but for the 980Ms being 680M-level cool OR COOLER, sacrificing slave cooling for better CPU cooling appears to be a nice reality. And now, I go to repaste my CPU with some Arctic Ceramique that I had my friend emergency-drop for me, because when the following picture is what happens when you're just watching a single livestream AT MAX FANS, you know something's wrong. Also, not in the picture, max GPU temp was 50 deg on master and 42 on slave, with averages of 40 and 38 respectively.

 

SOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO I emergency-got some Arctic Ceramique 2 from a friend and repasted. The pictures I am about to show you are AS-IS when I pulled the heatsink off the chip. WITHOUT me doing any cleaning at all. Ready? Here we go.



Edit: YES YES I KNOW MY CAMERA IS BUTTS. It's a 2010 nokia and the only way I have of taking pictures.

 

How the heck? Your heatsink is either incredibly flat or all your paste pumped out...

Sent from my HTC One_M8 using Tapatalk

 

Funny thing is that this happened twice; once with MX-2 (where it flat out evaporated) and the second time with this. Also, yeah, my heatsink looks like a 800 grit mirror-sheen lapped heatsink. THIS however was supposed to be ICD. Which is not meant to do that. But the first 4 months I had the PC this never happened... I originally tried repasting just to check, and I *DID* have a not-so-good paste the first time, because temps tanked before it evaporated. But this time I got it back and temps just got worse and worse. I don't even know anymore. My PC, man.

 

That doesn't look like IC Diamond to me. IC Diamond is very thick.





That was when I first got the machine. You can even tell by the dark color of it.

Sent from my HTC One_M8 using Tapatalk

 

That's exactly what I thought. It FELT like when we removed the MX-2 before. If they gave me MX-2 and told me it was ICD after denying my request for CLU... I've got no energy left to even be annoyed.

 

Dafuq is the only thing I got to say

And I concur, that is most definitely NOT IC Diamond. Shame on you Mythlogic.

 

Looks like MX-4 to me but I have no experience with MX-2.

Its certainly not ICD, GC Extreme, or any other high end paste...

BUT at least you know what the problem is and can hopefully fix it. My only way to have low temps is to hold back my processor from its full capabilities because the heatsink is so bad...

 

Well either way... I just have no energy for this anymore. I'm tired of being lied to by them. I'm fairly convinced that my GPUs have ICD on them, because they work like a charm. Hell, I find it difficult to actually get them to cross 80 degrees even when OC'd. Which is a good thing, considering the heat of my day. However I do more CPU intensive tasks, and my CPU was the thing I specifically asked for CLU on, specifically saying CLU would be perfectly fine on my GPUs... I can't use AS MX-2 or MX-4 on my CPU. It evaporates in a month or so. Last time there was even LESS paste when we used MX-2... I just caught this one before it got un-usable. I'm just fed up. I deserve better. I deserve a working machine and up-front information. I don't even know anymore.

 

I'd paste that up yourself at this point, use a nice thin line with a couple of practices to make sure you get it spot on and you should not have to worry about it again.

 

Well I did that, though I plan to use CLU later on. At this point I think I'm probably on my own with it. Best I try to fix anything other than flat out hardware failure on my own.

 

Pssst there's a reseller forum just so you know

 

Hopefully you have better luck with CLU than I did. We have the same heatsink and mine has too little contact to use CLU, machine overheated in no time and I took the heatsink off and it barely had connected between the heatsink and the CPU die application.

Sent from my HTC One_M8 using Tapatalk

 

Just to clarify one thing in the thread (and no we aren't at all trying to argue )

We did use IC Diamond on both the CPU and GPU. Its really the only thing we have in the whole place. We don't even stock anything else except for Liquid Ultra (which we would suggest not doing in mobile, as there isn't enough pressure), and you would KNOW if we put that on =)

 

Well, seeing as how this is the second time some sort of paste evaporated and I was able to run 3.5GHz and a lot of programs without going heavily past 90 degrees on basically bare contact, it might work for me.

 

You might think that but I'm doubtful. The retention mechanism itself is the issue but let me know.

Sent from my HTC One_M8 using Tapatalk

 

It's hard to tell from the picture due to the lighting and quality I think it is ic diamond but it's very odd to pump like that.

 

Maybe it was beside the bottle of vaseline (j/k)

Man, that makes me shudder... vasaline petrolium jelly as thermal paste. How fast to set it on fire?