Sunday, August 25, 2013

SPARC at 25 Years - GREAT Panel of Computer Industry Legends


I watched this FANTASTIC panel on SPARC - Scalable Processor ARChitecture that had ALL of the Sun Founders - Scott McNealy, Vinod Khosla, Andy Bechtolsheim and Billy Joy as well as Dave Patterson of UCB and the first VP of Engineering Bernie Lacroute and Anant Argrawal who led the first SPARC processor design team and Rick Hetherington who was a microprocessor designer at Sun.

If you are a fan of Sun OR like computer processor architecture discussions this is worth an hour and half of your time.

I was at Sun prior to SPARC, so I remember how all of this came about from a field standpoint.  At the time, the hot chip was a 25MHz Motorola 68020 that was 4 MIPS.  When the first SPARC chip came out, it was 10 MIPS, 16MHz and was manufactured by Fujitsu and was on a pair of 20,000 gate array.  It was codenamed Sunrise.  The first system was a SPARC 4/260 which was a desk side system.  What I remember about the introduction of SPARC was they wanted to have SPARC Ambassadors out in the field.  At the first meeting, Sun's microprocessor engineering team wanted us to demonstrate the advantages of SPARC by showing how SPARC register windows worked with this round piece of paper that had these windows cut in them and you rotated the paper around to show how the INS, OUTS and LOCALS would move.  I remember telling the folks in engineering that while this was interesting, what most customers really care about is end performance as opposed to HOW it was achieving the performance.  Some of Sun's customers did care though.  As Scott used to like to say, "the ideal Sun customer is someone who goes to the bike store and pays more to get a box of parts so they can assemble it themselves."

The really fascinating part of this is not only the history, but the exchange between these computer industry legends.  I especially enjoyed the back and forth between Dave Patterson and Bill Joy around the 1:17:00 mark.