Discussion Forums > Technology

10-core Ivy Bridge-EP Sample Tested

<< < (5/6) > >>

kitamesume:
tho any detailed explanation of whats the possible cause of the insane rise in temps? might be that 3d-transistor isn't so great with high frequency?

Tatsujin:

--- Quote from: kitamesume on May 02, 2012, 05:16:58 AM ---tho any detailed explanation of whats the possible cause of the insane rise in temps? might be that 3d-transistor isn't so great with high frequency?

--- End quote ---
I'm gonna speak the newb language: The design is different than the Sandy Bridge. >_>;; ... That's all I understood ... But it runs very hot if you overclock it. I'd rather not even bother overclocking this CPU. You already get Turbo Boost.

vuzedome:
It's a different chip but they're still using the same conventions when overclocking it.
Well I guess something is bound to go wrong, be it on Intel's side or OCers'.
On stock this processor is already more than capable, consumers should't worry about it at all.

kitamesume:
i got a rough hypothesis of whats causing it, intel did mention that they'd gang transistors in parallel which is supported in 3d-transistors to increase current supply.

my theory is, well actually its a fact, that transistors cannot perform identically even if they're the same(see how does Astable multivibrator begin the Initial power-up), impurities makes them have delays longer or shorter than the other transistors which means ganging them wont make them operate at 100% sync.

now when operating them, if like for example 10 transistors is turned on simultaneously but in actual half of them activate earlier than the other half, this means theres 50% decrease in path or in other words theres twice the electrical resistance in contrast to them fully activated. at lower frequencies however this issue isn't so noticable, but at higher frequencies the transistor will spend more time at this "delayed" effect which has twice the electrical resistance.

its just a hypothesis =P

bork:
Die is smaller so the heat density is up.
New thermal mounting method, not soldering the die to the thermal spreader.
Net result is about a 10C higher temp at full load.

Navigation

[0] Message Index

[#] Next page

[*] Previous page