Another OLPC man goes his own way
Submitted by srlinuxx on Sun, 08/24/2008 - 12:30.
Another stalwart of the One Laptop per Child Project has gone his own way - after telling project founder Nicholas Negroponte that he (Negroponte) had failed to go beyond the stage of a prototype.
Aaron Kaplan, founder and an active member of OLPC Austria, in a post on the OLPC News website, wrote: "Nicholas: in case you read this. Intel won. Big time. It will be hard to compete against the Dells and Intels of this world. We all knew that and you did not accept the fact. However where you could have been unique, outstanding and special was - guess what - with education software out of the labs of MIT and surrounding institutions and projects..."
Kaplan was referring to two news items - one that OLPC had refused to sell 900 test XOs to the government of Chile because it was considered too small a number. The second was a news item that Intel and the Portuguese government had inrtoduced production of the Magalhaes, "a Classmate-based computer that will be produced in Portugal and distributed to Portuguese children on primary education for €50 (free or at €20 for students on social aid), as well as exported to other countries."
OLPC redesign unavoidable ? Classmate same giveaway timing ?
OLPC was designed like any PC. Classmate design is obsolete, so Intel sold the manufacturing rights to recover whatever small returns.
We all know the inefficiency of FSB and chipset data transfer rate using PLL atm. We planned USB adoption with Craig Barrett(Intel chair) many years ago, not to push firewire. USB network replaces chipsets, using packet switching method same as telephone system.
The new design has to be using faster USB 2.0/3.0 (copper network with repeaters) serial data transfer speed in the cpu design change. Give up on all that obsoleted protocols.
OLPC will have to work with cpu design firms, fabless cpu manufacturers(Xilinx, analog devices, via, Mips, etc. for a design win) on realtime single core(split into multi 8 bit cores by chip select shift register manipulation) cpu with huge L2 cache. $100 computer design is just that massive integration to cut out printed circuit board and other parts not needed; using silicon chip that is a computer by itself(I/O by USB ports).
The hand writing is on the wall. Massive integration into silicon chip is the only cheap manufacturing method for computers that are ultralight for cost consideration.