• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Intel Broadwell-K running too hot!

Soldato
Joined
17 Jun 2004
Posts
7,744
Location
Eastbourne , East Sussex.
http://www.eteknix.com/intel-broadwell-k-running-too-hot-possible-cancellation


According to the official roadmap, Intel will release Broadwell-K and Skylake-S mid-2015. Both are built on the 14nm process where the former is for overclockers and the latter for the mainstream users, and that is something a lot of our readers are looking forward to. However, new information points towards some bad news for all of us waiting for these – Broadwell-K is having overheating issues. The CPU was originally set to a 65W TDP, but it has been unable to reach that. Instead, it’s ranging about 88W instead and even up to 95W on some models

and

The really bad news is the rumoured complete removal of the Broadwell-K series by Intel executives – at least initially. It might simply not be economically possible to bring it to the market at this point
 
Not sure what to believe with Broadwell anymore.

I've been seeing 95W TDP broadwell-K on roadmaps for months - though recently this changed to 65W on the latest leaked slides.

Guess we'll know in a month or two.
 
Probably gimped pasted chips like the Ivybridge and haswells. Pretty much expected this tbh.
 
Probably gimped pasted chips like the Ivybridge and haswells. Pretty much expected this tbh.

It will only get worse with the chips getting smaller and smaller on lower process nodes. We're destined to have pasted chips until Intel decide to release 6 and 8 core mainstream CPU's (non enthusiast). I can't see this happening until Zen is released, and only if Zen blows intel's current offerings out of the water.#

That said, there's nothing inherently wrong with pasted chips - they just run a few degrees warmer than soldered chips. It's only an issue for us overclockers, and we have x99 soldered chips to jump onto if we really want soldered chips.
 
Last edited:
True, they only run hot if you abuse them with certain programs. Programs that Intel themselves advise you not to use. I've owned both ib and hw. In normal use even when clocked they run pretty similair temps to previous chips I've had. 6 core on mainstream would be nice, but tbh I cant see it happening for a while.
 
True, they only run hot if you abuse them with certain programs. Programs that Intel themselves advise you not to use. I've owned both ib and hw. In normal use even when clocked they run pretty similair temps to previous chips I've had. 6 core on mainstream would be nice, but tbh I cant see it happening for a while.

which programmes? anything which pushes a cpu to 100% useage? like video encoding?
 
Check your V-Core because my temps on my K series Processor were really high and i found that my BIOS was an a version were the V-Core was set wrong, if this will help
 
88-95W TDP is nothing to worry about really it's pretty much the same as Haswell, it's just a shame that they missed their target of 65W TDP at this early stage of the new process.

If it's die size and their rubbish TIM method causing overheating then they need to ditch heatspreaders and do what AMD do with their GPU's, bare core with a metal shim around the outside to protect it, although they'd probably need a socket redesign for that and I don't think they trust people with bare cores.
 
If they changed the paste to something like cool labs liquid metal, they could keep the existing design. It has been shown that just doing this on a delidded chip and keeping the thick adhesive on the ihs results in very significant temp drops.
 
88-95W TDP is nothing to worry about really it's pretty much the same as Haswell, it's just a shame that they missed their target of 65W TDP at this early stage of the new process.

If it's die size and their rubbish TIM method causing overheating then they need to ditch heatspreaders and do what AMD do with their GPU's, bare core with a metal shim around the outside to protect it, although they'd probably need a socket redesign for that and I don't think they trust people with bare cores.

its not the same? the top end haswell-k are coming in at 84w with the bulk at 65w on 1150
 
EK waterblocks make a kit that allows direct die cooling, and msi include a drlid die kit with their high end z97 xpower board. Cant say if either work with air coolers though.
 
its not the same? the top end haswell-k are coming in at 84w with the bulk at 65w on 1150

4770K is 84W TDP, 4790K is 88W TDP less than 10W TDP difference assuming the 88-95W TDP reported in the article is worst case (ie. the highest end parts), so Broadwell won't be any different to Haswell if that's the case. The 65W TDP Haswell's are just higher end parts with reduced functionality, lower clockspeeds/voltages etc or smaller dies to begin with.

EK waterblocks make a kit that allows direct die cooling, and msi include a drlid die kit with their high end z97 xpower board. Cant say if either work with air coolers though.

You have to remove the socket though no way Intel are going to support something like that.
 
Last edited:
If they changed the paste to something like cool labs liquid metal, they could keep the existing design. It has been shown that just doing this on a delidded chip and keeping the thick adhesive on the ihs results in very significant temp drops.

Completely impractical for mass production purposes, so I doubt we'll see anything like that.
 
4770K is 84W TDP, 4790K is 88W TDP less than 10W TDP difference assuming the 88-95W TDP reported in the article is worst case (ie. the highest end parts), so Broadwell won't be any different to Haswell if that's the case. The 65W TDP Haswell's are just higher end parts with reduced functionality, lower clockspeeds/voltages etc.

i7 4790s still turbos to 4ghz and is a 65w part? matches the non k 4790.

(ofc the K versions isn't technically a haswell part anyway - its devils canyon)

http://ark.intel.com/compare/80806,80807,80808,80809

so taking out the devils canyon part - the 84w and under parts all perform in a very similar way - whilst the DC parts doesn't have VPro or SIPP
 
CLP / bare dies etc are super-unlikely - they'd go solder long before either of those ideas.


Devils Canyon is a name for some parts within Haswell, so they most certainly are Haswell.

intel name the part devils canyon (see ark link) - if it was haswell they would have named it as such - but the maker of the chip didn't.
 
They're basically speed binned 4770k's. Towards the end of 4770k production the clocking ability of 4770k's got very poor. Also from my experience the DC chips run much hotter than the older models as I've owned both.
 
Well with that crap TIM bet its boiling. For us on X99 its not problem solder all the way baby :D
 
Back
Top Bottom