You'll have to examine quite a few of each to break it down. Still, if the processors are arriving damaged and the only difference is the thickness of the substrate, my money would be the substrate. [Yet another reason I like to do this on-site when I'm creating/installing kit. The I know whose at fault, thank you. And speaking of which, mounting them on my six-core Xeons. Much older though.]
Alert after Intel Skylake chips, mobo sockets 'warp under coolers'
Beware if you're fitting third-party coolers to your Intel Skylake processors: it's reported that the chips and their sockets can be damaged by the weight of the heat sinks. Apparently, the Skylake parts have a thinner substrate than their Broadwell predecessors, and are more likely to bend and break when heavy-duty coolers …
COMMENTS
-
-
Monday 7th December 2015 00:02 GMT Sorry that handle is already taken.
Re: Oh, look. Another mainboard.
The last few generations of Intel's full-fat x86 CPU haven't resulted in much of a performance improvement anyway, so there's almost no need to upgrade your CPU(/MB/RAM) unless what you've got pre-dates Sandy Bridge or you absolutely must have the latest and greatest. Also, if you're overclocking the headroom has decreased.
Just buy a shiny new video card and be happy!
-
-
Monday 7th December 2015 11:28 GMT Nigel 11
Re: Oh, look. Another mainboard.
DDR4 is the main reason you might want to upgrade. The CPUs aren't significantly faster but DDR4 increases memory bandwidth and consequentially reduces idle cycles waiting for memory accesses.
IMO most folks upgrade PCs for the same reason they upgrade cars: a desire to boast. (My car is twelve years old and my work PC is six years old. Both are completely adequate for my needs, which in the case of the PC does not include "running" Windoze-anything and in the case of the car does not include carrying two tons of assorted extra junk around).
-
-
-
-
Saturday 5th December 2015 00:35 GMT Infernoz
Are people really still using huge heat pipe CPU cooler towers instead of water cooling?
You can get economically priced pre-assembled CPU water coolers which the beat pants off the biggest heat pipe CPU cooler towers. A much lighter and low profile water cooler CPU water block is probably a lot less likely to damage a CPU, socket or even board in transit than a heavy bouncing and wobbling heat pipe CPU cooler tower.
My pre-assembled CPU water cooler has been running 24/7 for several years, no problems, and the radiator is a hell of a lot easier to dust than a huge heat pipe cooler heat sink!
-
Saturday 5th December 2015 01:11 GMT Anonymous Coward
Re: Are people really still using huge heat pipe CPU cooler towers instead of water cooling?
pre-assembled water coolers barely beat good tower coolers. In fact, some of the better tower coolers will outperform the pre-assembled water coolers and are cheaper to boot! Not to mention, the likelyhood of pump failure after 5+ years. A good tower cooler is a much better option.
-
-
-
Monday 7th December 2015 13:32 GMT Captain Scarlet
Re: Holy war time!
That's odd, mine didn't smell, it just started ticking after air had got in. Replaced it with a bigger pre-assembled one and instantly regretted not checking the lengths of the pre-assembled hose (Queue several hours of GET IN THERE trying to fit the radiator where I wanted it in the case, then sod it, the radiator got screwed to wherever I could screw it in).
I think I will go back to air cooling next time
-
-
Sunday 6th December 2015 12:28 GMT Anonymous Coward
Re: Holy war time!
it's easy to tell when a fan is dying, but hard to tell when a pump is failing.
This is just due to cost cutting. There are plenty of cheap flow sensors available and plenty of designs to do something with the output. Pumps usually don't fail in the space of a few seconds.
It should also be easy enough to put a thermistor on the cpu cooler block and connect it to a cheap meter off eBay, running 5V to it via a suitable resistor.
Years ago I had a load of process plant running which used in-line flow sensors. There is no risk of leakage because the impeller is magnetically coupled to the sensor, no bearing goes through the case.
-
-
Sunday 6th December 2015 06:45 GMT P. Lee
Re: Are people really still using huge heat pipe CPU cooler towers instead of water cooling?
>pre-assembled water coolers barely beat good tower coolers
True, but I think the point is that water coolers move the weight off the chip and motherboard. A few percent difference in performance is not that important compared to the damage a large tower cooler can do, especially in in transit.
-
-
Monday 7th December 2015 07:25 GMT Pascal Monett
@ Steve Davies 3
Noise ? If your water cooling system is making noise, then you haven't bought the right one.
I have a model of the Corsair Hydro series, and the only way I can tell that my PC is running is that the LEDs on the front fans are on.
I used to have a full rig with pump and cooling for the CPU and graphics card and I couldn't hear that either. Hard disks make more noise than a proper cooling system.
On the other hand, I have witnessed low-quality pumps that make quite a bit of noise.
I hate noise.
-
-
-
-
Saturday 5th December 2015 20:29 GMT ADRM
Re: Are people really still using huge heat pipe CPU cooler towers instead of water cooling?
"You can get a Hyper Evo 212 for £24. Show me a pre-assembled water cooling system that cheap".
Got 5 Cooler Master 212 in our 5 main PC's. All with push pull fans and very quiet too. These are all AMD machines and one has been running with a Phenom II 965BE not over clocked for almost 4 years now. 30C idle and 44C under full load Prime 95. As someone who used to repair all kinds of electronics I have seen the damage water can do so in my mind water cooling is a big no no (At least for my machines, if you do fine but it would worry me).
-
Monday 7th December 2015 09:56 GMT BinkyTheMagicPaperclip
Re: Are people really still using huge heat pipe CPU cooler towers instead of water cooling?
A Hyper Evo 212 cooler doesn't look large enough to warp the CPU - it's the huge lumps of copper with a backplate that cause problems. A low end integrated water cooler is reasonably price competitive with a high end air cooler - I suspect in that case the air cooler will actually win, but be louder.
I stuck a water cooler on my main system purely because I was very short of space, and an air cooler blocked adjacent PCIe slots.
-
-
Saturday 5th December 2015 23:21 GMT Anonymous Coward
Re: Are people really still using huge heat pipe CPU cooler towers instead of water cooling?
"You can get economically priced pre-assembled CPU water coolers which the beat pants off the biggest heat pipe CPU cooler towers."
Really? My not-actually-that-large 120mm air cooler at worst "keeps up" with most water coolers (Corsair, etc)
" A much lighter and low profile water cooler CPU water block is probably a lot less likely to damage a CPU, socket or even board in transit than a heavy bouncing and wobbling heat pipe CPU cooler tower."
Agreed.
-
Saturday 5th December 2015 23:22 GMT Sgt_Oddball
Re: Are people really still using huge heat pipe CPU cooler towers instead of water cooling?
Until you warp the cooling plate and have coolant pissing all over your mobo (mobo and cpu just fine, cpu was hot enough to boil off the fluid before serious damage, so back to my 7 year old heat sink I went. Still runs at 5ghz, still doesn't go past 70 under load).
-
-
-
Saturday 5th December 2015 13:28 GMT Timbo
Re: Sounds familiar.
You can get under-side re-inforcing plates (that "spread the load" on Intel-type mobo's) quite cheaply...though you might need to change the fan too due to the different fixing arrangement.
I had an issue with a mini-ITX Intel mobo, using a relatively light-weight Intel-made fan/heatsink...and within months, noticed that the mobo PCB was straining under the force of pressure from the cooler, pressing it onto the CPU.
Found an "X" plate, which went under the mobo, and which had small sockets that went through the holes in the mobo (for the Intel cooler) and by changing the fan (to one that uses proper screws, instead of plastic clips that "spread" when the cooler thumbscrews is turned), and the mobo is now perfectly flat and running at a cooler temperature too.
-
-
Tuesday 8th December 2015 01:59 GMT I Like Heckling
Re: Sounds familiar.
Same here... bought a titan fenir back in 2010 when I was OC'ing a Phenom II 955BE. Earlier this year I swapped out the MB for a Asus sabretooth and an FX8320BE and it still idles at around 36ºC and never gets above 55ºC when at full load gaming. Also swapped out the crosfire pair of 5770's for an R9280X and the whole rebuild cost me £200 after I resold the old parts on fleabay.
I also bought a Noctua cooler for the mediaserver and stepped the fans down so it runs quieter... and that is barely audiable across the room at all... and when the TV is on you can't hear a thing. That aircooler is coping very nicely with one of the hotter FX cpu's running at 4.2ghz and still never gets above 62ºC when transcoding for up to three other devices whilst also playing a movie/tv show locally... even if it's transcoding HEVC x265 stuff.
I used to buy air coolers everytime something better came along... at one point I had a drawer full of them. Had the very first heatpipe cooler, the first Jet cooler and so forth. I was so OCD about getting low temps whilst overclocking that if it went over 50ºC under heavy use I consider the cooler inadequate.
But 18yrs of building my systems, and the fact I've stopped overclocking them because it now seems pointless... and this Titan Fenir is still going strong.
I doubt it would cope with one of the FX9xxx series though... I think a 220w load is more than most basic tower coolers can handle. But if I was going to build a rig with one of those, I'd probably do a custom water cooled setup.
-
-
-
-
Saturday 5th December 2015 13:35 GMT Timbo
"Intel provide a cheap solution that works. Use it."
err....I did do as you suggest and used an Intel-solution, but it didn't work so well, resulting in a bent pcb. (see other reply I made).
I would suggest that in some cases, 3rd party solutions can sometimes be a better option, simply because they overcome limitations in stock designs, which might not have been foreseen, when the product was originally designed.
-
-
Sunday 6th December 2015 06:58 GMT P. Lee
>Stock coolers tend to be loud as jet engines under continuous heavy loads.
Not even under load. My 3930k was running at 60-75 idle with an Intel air cooler
It's now down to low 30's with a prebuilt water loop. It doesn't need to be that low but my ambient often hits the high 30's in summer so a low base is useful.
-
-
Saturday 5th December 2015 23:20 GMT Anonymous Coward
"why not just use the Intel-supplied fan and heat sink? You know its going to work and it won't cause damage. Using third party heat sinks is simply an exercise in vanity, stupidity or futility."
Because running Metro 2033/Last Light caused my i7-4790K to hit 95C and thermal throttle, resulting in uneven performance (stuttering). Replacing the stock cooler with a Noctua U12S reduced load tempertures by over 30C, removed the performance variation I was seeing, and was quieter. When run under sustained high load, my (underventilated, stock cooler) i7-2600K work PC throttles and down-clocks under load, too
The stock coolers are fine for "normal" use, but for extended periods of high load, they're not...
-
Sunday 6th December 2015 12:37 GMT Anonymous Coward
A feature
"When run under sustained high load, my (underventilated, stock cooler) i7-2600K work PC throttles and down-clocks under load, too"
Perhaps the idea is to reduce the thermal load on the under-designed motherboard?
Commercial computers tend to be like base model cars; yes they have a top speed of 100mph, do not try this for long distances unless you enjoy engine swapping. If you put in a stronger engine you may simply find out what is the next weakest link in the chain.
-
Monday 7th December 2015 11:03 GMT Nigel 11
Re: A feature
Commercial computers tend to be like base model cars; yes they have a top speed of 100mph, do not try this for long distances unless you enjoy engine swapping.
Not these days. If they get too hot they'll throttle back (reduce clock speed) to avoid damaging the CPU by overheating. So more like a car that'll do 100mph sprints but will refuse to go that fast for hours on end on a hot day, however hard you press the pedal.
(Some hybrid cars are exactly like this. Can't/won't maintain top speed up a long incline, to protect the electric motor or battery.)
-
-
-
-
Saturday 5th December 2015 14:10 GMT John Geek
to intels' credit they do have excellent publicly available engineering documentation, errata, etc.
even a couple gens ago (i5-3570 here), most of the big tower heatsinks exceeded Intels' heatsink weight recommendation.... but the stock Intel heatsinks were barely adequate for 100% CPU compute jobs, and had noisy fans (quiet at idle, but noisy at 100% CPU heats), so I'm using one of those big tower heatsinks with a sideways fan. I'm not shipping my system, it was assembled on site, so its fine.
-
-
Sunday 6th December 2015 08:47 GMT Kyorin
That statement is almost completely false (in the UK at least). Nearly all of Intel boxed processors come with an Intel heatsink/fan. There are some models that don't, but these tend to be the K models or high end models, but the vast majority of their boxed CPU's do come with them.
Now if you'd said Intel don't make their own motherboards anymore, I would have agreed, but even that is not entirely true, they do still make their own server boards I believe.
-
-
-
Sunday 6th December 2015 13:05 GMT Anonymous Coward
why would you have a massive 50lb heatsink
You wouldn't. 50lb is the maximum loading. If the computer is dropped and hits a hard surface, that could easily be exceeded.
In the days of metal can TO5 and TO18 transistors, the maximum safe impact was stated as around 300g, which sounds a lot till you realise that this could be exceeded by a transistor falling off a desk onto a concrete floor. In the same way, for a 500g cooler to apply 25kg of force on a motherboard only requires 50g. That means that if there is 1cm of give due to packaging, dropping it from 50cm will exceed the limit. If the motherboard itself is dropped directly onto a hard surface, with perhaps only 1mm of give, a 50mm drop will do it.
-
Sunday 6th December 2015 19:23 GMT Zmodem
Re: why would you have a massive 50lb heatsink
if you drop your box you would be expecting everything to be broke, so a bent socket means nothing, most cases would be broke, and half of your PC would fall out, you have to buy a rack mount case or spend £200 on a sturdy ATX
if a socket could handle 24kg, the motherboard would need case mounts on all 4 corners first
-
Monday 7th December 2015 11:20 GMT Nigel 11
Re: why would you have a massive 50lb heatsink
if you drop your box you would be expecting everything to be broke
I've seen it happen three times, moving PCs between offices. Every time a corner of the case hit the floor and the case got bent out of right-angle alignment so afterwards it wobbled on the desk and it was a pain to get its covers off or back on. Every time the system still worked perfectly (including the hard disks, which I find rather more surprising than the CPU/Motherboard).
So no, I would not be expecting a metal-case system with a stock Intel cooler to be broken if it gets dropped while powered down. I'd be cautiously optimistic about it.
Of course, if you buy those nasty systems with a mostly-plastic case, the case will probably be in several smashed pieces and there's not a lot of point testing the electronics after that.
-
Monday 7th December 2015 11:27 GMT Zmodem
Re: why would you have a massive 50lb heatsink
if you had a single heat sink that was 24kg, the rest of the PC would be atleast 6kg which would put it over the health & safety limit of 25kg for men, and you would have you get a trolly if you don`t do weights and can lift 25kg in 1 hand
office workers are soft and get footrest etc all paid for under health & safety in the office level 1
-
-
-
-
-
Monday 7th December 2015 13:53 GMT Andy The Hat
I doubt conventional heatsinking is the problem. It's hanging a massive heatsink off the board when the board is mounted vertically ... the rotational moment is huge so tends to try to rip the top mounts out and press the bottom ones through the board.
Water cooling works well (a Seat Ibitza heater matrix is a good size for near silent cooling :-) ) but some boards are designed such the the airflow cools the surrounding components so a water cooler, whilst great for the processor, is not good for the local area of the board ...
-
Thursday 14th January 2016 12:30 GMT David Lawrence
That might explain it.....
Last week a friend brought me a broken PC to look at. It had a lovely Motherboard (big, shiny AsRock thing with pretty blue cooling fins - very swish), a high-spec Core i7 and a simply massive CPU cooler with huge fins and two very large fans strapped either side of the fins.
Turns out the Motherboard had failed, but the CPU had survived so a new (cheap) mobo and he was good to go.
The case was a tower, so the mobo was vertical and the cpu cooler must have been applying considerable sideways/downwards force. Yes it had an X-plate under the mobo but I still think the enormous/crazy/excessive/heavy CPU fan might have caused the damage.
Now I'm wondering how long that new (cheap) mobo might last.....!