Thread replies: 15
Thread images: 3
Thread images: 3
File: core_i9.png (571KB, 1900x972px) Image search:
[Google]
571KB, 1900x972px
The new i9 series are just overclocked i7s with huge power consumption and heating issues.
>"I didn’t believe it at first, but the Core i9-7900X is using more than 70 watts additional power than the Core i7-6950X!"
>"But woe, if you have all the cores really work out properly! With Prime95 and Luxrender on all cores for the Core i9-7900X, we have determined absolute peak values for the power consumption. The Core i9-7900X achieves remarkable 48% more performance in the render score (Luxrender, console), but also takes up 58% more of the electrical power! This is almost the famous blue breaking bar."
>The system ran stable with 1.4 Vcore in 10 runs and the CPU took an average of 261 watts. Single peaks, however, were load-dependent at up to 293 watts, which is very close to the limit. A test carried out at 4.8 GHz with 20 instances of a shading-all-year calculation for a complex photovoltaic roof system including a yield calculation allowed the power consumption to rise to 335 watts. When starting Prime95 without AVX limitation then switched off the mainboard. The last, still logged value was 364 watts."
>"The real reason is Intel's incomprehensible fun brake in the form of unsuitable (but probably much cheaper) heat transfer paste instead of a sensible solder."
>"Up to 65 ° C Tcore with a heat spreader temperature of approx. 24 ° C results in a delta of more than 40 ° C. We reached this value with a power loss of just under 230 watts. As soon as you move above the 300-watt limit, which can be reached even with simple rendering programs from 4.6 or 4.7 GHz and the necessary voltages (depending on the chip quality), the chiller is hardly anything. At the maximum values of 300 watts reached by us the CPU is already permanently thermal limit of 100 ° C, shortly thereafter comes to the consequential shutdown."
Overclocked CPU couldn’t even be cooled with a $1000 compressor cooling solution.
>>
>>61021811
>look mom I posted it again :^)
>>
>>61021811
Who are you quoting and why did you quote twice?
>>
>>61022558
>when intel shills can't handle the heat
>>
File: skylakex.jpg (124KB, 400x625px) Image search:
[Google]
124KB, 400x625px
i just dont understand why Intel didnt let aymed have this year and make actual progress instead of rebooting broadwell a fifth time
>>
>>61024260
>and make actual progress
That's a decision they had to make years ago. No new arch until 2021.
>>
YEAR OF THE AMD CPUS!
>>
File: AMD_Reveals_‘Epyc’_Details;_Intel_Vows_to_Top_It_-_Barron_s_-_2017-06-21_02.16.39.png (324KB, 1191x1587px) Image search:
[Google]
324KB, 1191x1587px
>>
>>61025476
>Old man yells at clouds
>>
>>61021811
It goes up to 4.8ghz? Thats cool
>>
>>61025894
And requires a phase change cooler to keep it from melting down.
>>
>>61026049
Well thats obviously not a feasible oc then. But it's good to see that the oc is limited by temperature (and voltage) instead of stability.
Did they remove the heat spreader?
>>
>>61026117
Nope, because you shouldn't need to remove the heatspreader on a $1000 piece of silicon just so it operates in the appropriate temperature regime. But no, intel wants to save the $1-2 a processor it would cost to properly solder them to the heatspreader.
Its further compounded by the fact these things run hot and power hungry in the first place, and intel all but requires that their baseline stock cooling be a CLC or extreme high end air.
>>
i9 testing laboratories:
https://www.youtube.com/watch?v=UxQdS0pbpKo
>>
>>61026156
> because you shouldn't need to remove the heatspreader on a $1000 piece of silicon just so it operates in the appropriate temperature regime
You shouldnt need a phase change cooler to do that either. Clearly these clocks are not doable for the average person, but it'd be interesting to see if they pulled out every stop to see how high they could get the clocks.
Thread posts: 15
Thread images: 3
Thread images: 3