Quote Originally Posted by saaya View Post
(snip)
did you guys see the tsmc yield news from digitimes?
0.1-0.3 errors per square inch? fermi = ~ 1square inch...
so every 3-10 fermi chips there is one error? then yields would be 70%+ for fermi and 90% for rv870... weird huh? i wonder if he just made that number up or if he misunderstood it... maybe 0.1-0.3 errors per square millimeter?
that would mean 2-8 errors per fermi chip, on average... and around 1-3 errors per rv870 chip on average... nah thats too much, right?

weird...
It could mean square centimeters......
Linky