they had problems with coffee lake on 10nm - not enough time to revamp litho - so they went with 14nm
ice lake since it's later had the time to revamp lithos and design to work with 10nm
10nm was delayed sooo long - because of problems and coffee lake failed on 10nm(bad lithos, bad power regulation needed new design) and why it's on 14 nm now
when you shrink you have to redesign the chip - the transistors and the power delivery system - your running lower voltages - and closer and closer to cutoff voltages
it's essential to have good power regulation - to keep power smooth and consistent - power spikes are more critical to leakage and failure - because of the close nature of the parts and power dips will cause losses/hangs as well
for the most part i think intel had it's pants down - not really looking at a full redesign of the chip and why it's been delayed -to fix up the completed design of the chip
Coffee Lake is going to impact Intel’s margins
https://semiaccurate.com/...impact-intels-margins/ as a note Nvidia went thru this with fermi and why they built a chip design simulators(for each generation) after that
and why Nvidia has been revamping power delivery system with each generation along with the complete redesign of the chip
something AMD should learn to do ...a GPU that east 500 watts??? when overclocked
post edited by Xavier Zepherious - 2017/08/16 09:41:54