Unified Core is just a name. Royal Core was also a thing, except the "thing" kept changing. SMT got "turned off" and back on in less than a year. 18A was the node to make IFS great again after 3 years of kung-fu focus.
It's just like the Core vs Netburst scenario. They called Core as combining Netburst with Pentium M, but it was a natural evolution of Pentium M. A total pivot with an entirely new architecture doesn't happen. Zen was new for AMD, but many fundamentals were shared with pre-Bulldozer chips and Intel cores as well. The saying why RYC did not do so well was that they didn't have experience making a core built for actual customers.
Same thing is likely going to happen here. Arctic Wolf is an evolution of Skymont, and Unified Core is a grown up version of Arctic Wolf.
No, i don't think so because not only did Intel stuff up Atom, they stuffed up modems as well.
I'm saying being in 2 years earlier would have made the difference. They would have been forced to put more resources into advancing if they had enough marketshare.
Skymont doesn't have enough 1t perf compared to the ARM camp now and Lion Cove is just too power hungry.
The biggest barrier for Medfield wasn't performance as much as application compatibility. 2 years earlier would have had enough x86 compatible Android binaries. One big complaints for Intel based phones were that some just did not work well.
Performance would have mattered in that regard too of course. Because they had the same in-order core for 4 years. It was Moorestown in 2010 and Medfield in 2012, with the exact same core, only difference in that Moorestown was shoddy for phones and Medfield was decent. LG was in line to use Moorestown, but it never materialized, because the platform was crap.
Remember the in-order Atom was competitive in performance with the out of order Cortex A9 chips of that time. If 2012 it was acceptable, in 2010 it would have made real impact.
Could the Atom core development been accelerated even further in that case? We can't say for sure, but many times Intel has said they wanted to keep a 10x performance gap between Atom and Core. If they took 1/4 of the whole market? Then this mentality would have changed quick. Also remember the anemic 4EU graphics in Bay Trail Atom? It was rumored by those in the know that in the planning stages it had
half the EUs. The mobile market forced them to do better.
Every day, every month, every year it passed they lost the opportunity to take part of that market.
Also being actively in the market forces them to do things they wouldn't have done otherwise. I argued that Intel had zero motivation for drivers because they don't directly deal with gamers. As soon as ARC dGPU came out, their driver effort ramped up greatly. It's a feedback loop between seller and buyer.
Everything else is minute detail. They missed the big picture by not pivoting to mobile earlier and having capable chips available so vendors could just choose Intel rather than Intel going "pick me, pick me" after everything was pretty much over.
Facebook
Twitter