--- crowd-funded eco-conscious hardware: https://www.crowdsupply.com/eoma68
On Fri, Apr 28, 2017 at 12:31 PM, Bill Kontos vkontogpls@gmail.com wrote:
Where does the 40% improvement come from ? I remember watching a video from Berkeley student where the improvement was "only" 11%.
i read a paper somewhere where they'd put a RISCV64 design into silicon @ 40nm, ran it at 1.5ghz... but i have, if i am honest, a "fuzzy memory"
The kirin 960 was a complete fail on their part. They downgraded the manufacturing node to one that has a lower frequency sweetspot but lower cost as well... still 14nm for whatever it's worth these days. On top of that, and especially on the gpu side they have very aggressive clock steps, even missing the efficiency sweet spot and resulting in peak draws of up to 10 watts. All for benchmarks on reviewing websites, most reviews just run a benchmark once and show the result but do not test overheating. Totally not suitable for eoma, no matter how much of a sucker for high performance I am.
there's always the EOMA200 standard - one for the future (several more years into the future).
l.