http://liliputing.com/2013/09/how-the-8-cores-in-samsungs-octa-core-chips-wil...Some further information of the state of ARM big/LITTLE multiprocessor scheduling (looks pretty good, doesn't it?) This functionality is absolutely vital to be working correctly and to be fully intergrated into the Linux kernel AND RELEASED before the next wave of ARM multicore processors hits the street.
.... and no, it wasn't a fluke that Samsung was chosen to be guinea pig the roll out of the octa-core big/LITTLE stuff. ARM always partners with a single roll out partner when rolling out something brand new, that way the bugs can be coralled much quicker and all the improvements made ready before the general roll out happens.
The roll out partner gets the technology six months to a year ahead of the pack to allow time for needed improvments to happen.
=======================
Note that Allwinner was the A-9 full SOC single core roll out partner, Samsung was the dual core/quad core 46/32nm A-9 partner, the 28nm quad A-9 core was rolled out with Rockchip and now the 28nm A-12 will rolled out late this year with Rockchip as well.
Of the lot of these product roll outs, the A-9 Allwinner was the most flawless and the 8 core big/little Samsung was the cobbiest. Still, all of them rolled out on time and went into full production on time, something Intel has yet to do without an extra year of vaporware tucked in there somewhere.
Getting all the big/LITTLE scheduling stuff completely right before rolling over to the A-53/A-12/A-57 generation was vital for ARM, as in 2014 ARM goes to war with Intel in a much more direct fashion.
From now on out, any bobbles or any dropped balls will count heavily on all three sides of the equation (ARM, AMD, Intel).
=========================
Also note that Intel is going to be ready to roll on RISC software if x86 isn't ready to go with a really compelling product.
Google has made sure that Intel products are well supported both by Android and by Chrome OS. Since MS's last reorg & CEO firing and Nokia buying, MS simply isn't considered something you'd want to "bet your company on" any more by any of the major players.
From now on MS must earn whatever laurels it gets the old fashioned way -- a better product executed in a better fashion. Intel is back in the game, so far MS is not ....
For ARM, the easy sailing is over. Intel has products that will play in the ball park and they have indicated a willingness to lower their sky high prices some.
Both teams are out on the grass now, practice time is definitely over. Let the 2014 game begin ....
Pages: 1
