For Sun chip chief, an empty roadmap is a clean roadmap
SPARCs fly
Posted in Servers, 15th April 2004 22:34 GMT
Free whitepaper – Reliability analysis of the APC Symmetra MW Power System
Sun is putting a brave face on the cancellation of the UltraSPARC V, and Eagle, an entire family of servers designed for the processor. The Millennium chip - so called because it was originally due to be ready in 2000 - had just taped out, but two weeks ago became the latest cancellation from Sun. The multicore UltraSPARC VI Gemini, VI and VII have already fallen off the roadmap, but microprocessor chief David Yen is looking on the bright side.
"If you look at [sic] from this our perspective, Sun's processor roadmap will be a very clean one," Yen tells eWeek.
You can't argue with that.
The roadmap isn't entirely empty, but it now resembles your reporter's fridge before the twice-weekly trip to Molinari's Delicatessen. A few stale items long past their sell-by date wait their turn to be spun into something that can only be deemed edible with a lot of marketing sauce. But unlike the fresh home-made spinach pasta and sweet olives, microprocessors don't materialize so easily.
Sun has systems based on the Niagara and Rock processors it gained through the acquisition Afara systems, which are based on a SPARC core, and IV-Plus, which is an UltraSPARC IV with more cache. There's also Andy Becholsteim's skunkworks project to consider, based around AMD's Opteron processor. But it looks like Sun will be increasingly dependent on any relationship with Fujitsu, which has been developing its own SPARC-based processors for a decade. The two companies were said to be in talks last year, but no formal partnership has yet been announced. We're assuming Sun is serious when it says it wants to stay in the hardware business.
eWeek speculates that the USV may be revived, but sources in the know say this is unlikely. The Millennium processor would struggle to be competitive today, and Eagle servers using the chip were not slated for production until 2007, more than a decade after the project began. As the project slipped, Sun engineers joked that the chip wasn't late - it was simply 996 years early. ®
Related stories
Sun shelves UltraSPARC V in favor of the great unknown
Sun tries UltraSPARC IV temptation
'What do we stand for now?' ask Sun staff
Sun sees a lot of cache in its future
Sun shelves UltraSPARC VI in favor of The Rock
Sun puts co-founder back to work
Sun and Fujitsu: a relationship with Sparc
Free whitepaper – Fundamental Principles of Air Conditioners for Information Technology

The Register Agile Data Center Summit
New storage architectures make SSDs more cost-effective
Dell PowerEdge R710 solution with VMware ESX vs. Dell PowerEdge 2850 solution
Analyst Keynote: The Register Agile Data Center Summit

Toshiba plans new enterprise: High capacity 3.5-inch HDDs
IBM greases mainframe app pipe
Acer, Asus dominate Euro netbook biz
Quantum's small tape libraries get big