Quote Originally Posted by ice_chill View Post
It's a bit like when Core 2 was shrank from 65nm to 45nm, nothing big but still a little architecture optimization.
65nm to 45nm was E6850 vs. E8600 - that was a boost in performance and raw MHZ oc...I am not sure yet if we'll see such big improvements.

Quote Originally Posted by RPGWiZaRD View Post
As usual from die-shrink "tock" at this point u can add the typical 4 ~ 10% IPC increase (around 15% in some special cases) but what is attractive is its performance/power consumption/overclocking. I pray it'll do like 5.5GHz @ 1.35v on an average sample and consume quite a bit less power than Sandy would do at like 5GHz 1.4v. This is just speculation but one can always hope right.
I like throwing my overclock speculation pre-launch so here's mine:
Worst sample: 5.3GHz @ ~1.35v
Average sample: 5.5GHz @ ~1.35v
Best sample: 5.7GHz @ 1.36~1.37v
i5-3750K / i7-3770K on good aircooling
If there will be numbers like that and it scales with cold --> big win...but I am not sure yet if it will really be that good...

Quote Originally Posted by RPGWiZaRD View Post
Yea 17% sounds a little optimistic (and the first test probably is using AVX) but when isn't a score produced by the manufacturer itself always a bit higher than what average joe gets in the same test? I bet we are more likely to see maybe around 10% or so in those tests. But OMG @ the internal GPU performance increase... sounds like it might be able to play a few even newer games at lower res quite fine.
There is a lot room for better bios and mem-tweaking for sure - thus initial results are not reflecting real world performance at this point imo