Possible for a heatsink to be broken/not working to full capacity?

[Armadillo]

So strange question, but is it possible for a heatsink to have something wrong with it that effects cooling. They are just lumps of metal, so I didn't think there could be much wrong, but mine seems to be underperforming.

At first I thought it was the case fans (https://forums.overclockers.co.uk/threads/tie-case-fans-to-cpu-temp.18772046/ ), but it turns out I was mistaken. I was tired and must have just misread it or read the average, it actually doesn't make that much of a difference.

Trying to overclock my 5820 and [email protected] gives these temps after stress test with Realbench.




78C on package, 74C on hottest core. Still just under 80C, but not by much. Spinning the case fans up to full speed, drops 3-4C.

That's the hottest it gets. Handbrake video encode is 73C package and Firestrike run gives 68C package. Stock is fine, in terms of being safe temps, hottest I've seen is 60C or so.

The cooler is a k2 mount doom and those temps just seem underwhelming based on the cooler and what other people get on it. It still has only the standard fans on it (1500rpm 120mm and 140mm 1200rpm), same config as Noctua has and again people seem to get far better temps and the two coolers are comparable.

Obviously, the first suggestion would to remount it, but it's not just these temps that have me asking the question, it's always seemed to have been underwhelming compared to reviews and other uses.

Originally I had it on my x4 955 and while it cooled, the temps were worse than reviews and others. Changed to an 8350 and couldn't even go past 4Ghz due to temp, again underwhelming compared to others. On the 8350 I mounted it 3 or 4 times with little improvement.

Don't think it's an issue with me or thermal paste etc as all the other coolers I've mounted (mostly hyper212s) all perform in line with reviews and other users.

Initially I just put it down to bad airflow. As both the 8350 and x4 955 were stuffed into a cm690ii with questionable airflow, but these results have me questioning that theory as the system is currently in a Corsair 750D airflow which is a massive case with plenty of room and unrestricted airflow. PC is just on a desk, next to a wall, but I think it's still got plenty of room (can take a pic if needed).

So just not sure why it seems so underwhelming .

 

[Holst1981]

Heatpipes can leak and become less effective.

The walls are quite thick on the Heatpipes so this would be unlikely,

Ive no idea if your cooler is any good but your temps are a bit high, but not dangerous.
I have a cheap ocuk aio and I get similar temps at 1.35v 4.5ghz.

Check you don't have any odd bios settings (esp voltage)

Make sure you are reading voltages and temps correctly, some programs report incorrectly.

Otherwise try another cooler or borrow one.

 

[Tealc]

Could be an ineffective heatpipe or two. You could test this by touching the heatpipe surface but may not be conclusive.

 

[doyll]

While there is a very slim possibility of having a defective heatpipe, it is extremely rare. I have been using heatpipe cooler for over 10 years and have installed hundreds ... have never had a defective heatpipe or had one go bad. But I have seen many bad TIM seats, poor case airflow causing high temps, dirty filters / grills / fins / fans and other causes, but never a bad heatpipe.

 

[Armadillo]

Heatpipes can leak and become less effective.

The walls are quite thick on the Heatpipes so this would be unlikely,

Ive no idea if your cooler is any good but your temps are a bit high, but not dangerous.
I have a cheap ocuk aio and I get similar temps at 1.35v 4.5ghz.

Check you don't have any odd bios settings (esp voltage)

Make sure you are reading voltages and temps correctly, some programs report incorrectly.

Otherwise try another cooler or borrow one.

I think the readings are correct. Overclock is off now, but just doing a couple of runs of handbrake with different programs monitoring and the temps and voltages are moreless the same. Hwinfo, Hwmonitor and Coretemp all report the same core temps. Only actual difference in temps seems to be package. With hwinfo reporting 5C or so more there than hwmonitor.

Looking at hwinfo forum, package temp being 4-5C higher than hottest core seems to be thing on haswell-e and hwinfo or at least a thing enough of time to have multiple threads on the forum. It's suppose to be an average of the last 256ms samples. Even disgarding the package temp and just looking at the hottest core. 74C at [email protected] still seems a little high, especially compared to what others report, like so.

https://forums.overclockers.co.uk/posts/30606560/

50C with a d15s@ 4.4. D15s is only slightly better than the d14 and the d14 and k2 are supposed to perform similar . Even just on core, looking at 14C difference there at 400Mhz less.


Voltages. I don't see anything out of the ordinary. I changed to the multiplier to x40 and just put to +0.100 as offset to get 1.15ish under stress test. All the other voltages I left as they were.Looking at the voltages hwinfo reports, I didn't see any difference on any of them other than the vcore/vid between stock and overclocked.

Could be an ineffective heatpipe or two. You could test this by touching the heatpipe surface but may not be conclusive.

I don't think I can get to the lower parts of the pipes . I'll have to have a look. I would assume the top of them sticking out will be no good, as after going through the fins they should have cooled down?

While there is a very slim possibility of having a defective heatpipe, it is extremely rare. I have been using heatpipe cooler for over 10 years and have installed hundreds ... have never had a defective heatpipe or had one go bad. But I have seen many bad TIM seats, poor case airflow causing high temps, dirty filters / grills / fins / fans and other causes, but never a bad heatpipe.

I think airflow is ok. 750D is a big case. I'll try and grab a pic at some point.

I would think a bad heatpipe would be rare as well, it's just it always seemed to underperform where I had it before as well, compared to other heatsinks I've installed. I suppose it could have been bad airflow in case before and now, not a great mount/thermal paste etc.

Is there anything else it could be. The base did look ok last time I checked it, but in this review on https://www.vortez.net/articles_pages/alpenfoehn_k2_cpu_cooler_review,5.html, it seems like their one/all of them have a slight convex base. I've never checked mine, with a big chip like a 5820k would a base like that cause issues or should the thermal paste take up the slack.

It seems resonable at stock. Stock handbrake, get me about 55C on a core and 60C (on what may be questionable) hwinfo package. Gaming is lower as handbrake seems to hammer it more than games. Although based on others seeminly getting similar temps when overclocked, I guess it could still be considered underperforming.

 

[Holst1981]

If your happy your not accidentally running a voltage that's way too high and it's stable I wouldn't worry about it too much.
Just keep an eye on it.

Sometimes it's best to leave things alone if they are working OK.

5820k at 4gig is pretty quick.
If you need more you could invest in a new cooler and try more voltage, but I doubt you need more at the moment.

 

[Disco_P]

74C at [email protected] still seems a little high, especially compared to what others report, like so.

https://forums.overclockers.co.uk/posts/30606560/

50C with a d15s@ 4.4. D15s is only slightly better than the d14 and the d14 and k2 are supposed to perform similar . Even just on core, looking at 14C difference there at 400Mhz less.


Voltages. I don't see anything out of the ordinary. I changed to the multiplier to x40 and just put to +0.100 as offset to get 1.15ish under stress test. All the other voltages I left as they were.Looking at the voltages hwinfo reports, I didn't see any difference on any of them other than the vcore/vid between stock and overclocked.

I think airflow is ok. 750D is a big case. I'll try and grab a pic at some point.

I would think a bad heatpipe would be rare as well, it's just it always seemed to underperform where I had it before as well, compared to other heatsinks I've installed. I suppose it could have been bad airflow in case before and now, not a great mount/thermal paste etc.

About that 50C temp at 1.35V.

I would suggest trying the tests with the case sidepanel off. A big case doesn't necessitate good airflow, and the 750D has an "oven" style front.

I would also suggest manually setting voltage rather than relying on auto volts and offset - mobos are notorious for overvolting when left to auto whilst overclocking. Which motherboard are you using?

It's possible that 1 or more heatpipes have been defective since purchase, they are filled with a liquid that could have leaked out, never been present or is from a bad batch.

 

[Armadillo]

If your happy your not accidentally running a voltage that's way too high and it's stable I wouldn't worry about it too much.
Just keep an eye on it.

Sometimes it's best to leave things alone if they are working OK.

5820k at 4gig is pretty quick.
If you need more you could invest in a new cooler and try more voltage, but I doubt you need more at the moment.

I don't think I even need it at 4 yet, for what I play it seems fine at stock, not hit a cpu wall yet. It's more well, it can be overclocked, so I should .

Voltage seems fine, but I just don't really have much overhead for summer. 80C seems to be what people are saying not to go over. And while I can't seem to get an agreement on whether that's core on package (some say package, some say core). It's only 2C headroom on package or if we ignore that because it may possible be wrong on hwinfo with haswell-e, then 74C on core is still only 6C, not much to work with for summer temps.

About that 50C temp at 1.35V.

I would suggest trying the tests with the case sidepanel off. A big case doesn't necessitate good airflow, and the 750D has an "oven" style front.

I would also suggest manually setting voltage rather than relying on auto volts and offset - mobos are notorious for overvolting when left to auto whilst overclocking. Which motherboard are you using?

It's possible that 1 or more heatpipes have been defective since purchase, they are filled with a liquid that could have leaked out, never been present or is from a bad batch.

Sorry, my fault. Always forget to say it's the 750D airflow version. Airflow version has a large open front/grill rather than the solid panel of the normal 750D.

Motherboard is a Gigabyte X99-SLI. I kept an eye on the voltage and with the +0.100 offset, the highest it went was 1.15V, multiple programs agreed, so I don't think it's over volting. It doesn't take a second to check, so at the weekend I'll try with manual voltaged 1.15 and case on/off see it makes a difference.

 

[Armadillo]

Forgot to actually update this. I tested a few suggestions before the weather got hotter and just forgot to update.

Fixed voltage instead of offset, temps were the same as they were before.

Tried with the side panel off/on to see the difference incase of bad airflow. Difference between side panel on & off was ~5C. I think that's a reasonable difference and doesn't point to bad airflow.

Touching the heatpipes. Can only touch the parts sticking out the top, couldn't really feel any difference between them.

Touching the heatsink itself, I'm not sure if this is normal or not as this is the only dual tower cooler I've ever had. The front tower was much cooler than the rear. Obviously, the front has unrestricted cool air, so I expected it to be slightly cooler than the back, but it was quite a bit cooler. Does that seem right?

Not tried a remount yet as it's a pain in the butt (have to take gpu out and move case as there is not enough room to lay it down), but was underwhemling on other cpus, while other ones I've mounted perform fine, I guess I could just be messing this one up all the time .

 

[EsaT]

Touching the heatsink itself, I'm not sure if this is normal or not as this is the only dual tower cooler I've ever had. The front tower was much cooler than the rear. Obviously, the front has unrestricted cool air, so I expected it to be slightly cooler than the back, but it was quite a bit cooler. Does that seem right?

Temperature difference should be in class of how much first heatsink warms air flowing through it.

Do you have any kind meter you could use for checking temperature?
Human senses aren't very good absolute meters.

If you really want to hammer that CPU try Linpack based tests
http://intelburntest.en.lo4d.com/

 

[doyll]

I agree with EsaT on the temp difference and needing a thermometer.

5c with side cover off is the max I will allow, but I monitor the air temp going into cooler for that 5c difference. Here is link to basic guide to case airflow with thermometer I use at bottom.
https://forums.overclockers.co.uk/t...-i-put-my-temp-sensor.18564223/#post-26159770

The back finpack feeling significant'y warmer than front indicates the cooler is transferring a significant amount of heat from CPU into the cooler and then into the airflow.

 

[Armadillo]

Whoops, completely forgot about this. No update as with the recent temps, it was too hot & I couldn't be bothered to mess around and then I just forgot .

Thanks for that, good info there .Don't have a thermometer, but looks like similar basic digital thermometers with a probe like that are available for £10ish, so I'll think I'll pick one up.

Actually getting to the bottom of it (bad mount, bad cooler, people telling porkies about temps ), is on hold at the moment anyway as I might change to a d15s or aio for practical reasons, not sure yet. Changed gpu and was reminded how much of a pain it is with this cooler, as I'm forced to use the top pice on this board for x16. Offset of the 15s might make it easier, aio will definitely, but more cost for same performance.

 

[ANDARIAL]

If you have more than 1 fan mounted on the cooler are they both going the same way?
I.E. one is not push and one pull is it?