Compaq Computer''s Alpha chip has long been the workhorse of high-end genomic and bioinformatics research, so Compaq''s recent decision to sell the 64-bit Alpha technology to Intelóand to eventually standardize on Intel''s Itanium chipócame as a surprise to many in the industry.
But Compaq was quick to quell any doubts among its high-performance computing customers in the life sciences that it has lost faith in the Alpha technology, which it has long touted as being the best performing system for bioinformatics applications.
"The assumption that Compaq is abandoning the Alpha processor in favor of the Intel Itanium processorÖis not true," said Lionel Binns, worldwide life and materials science group manager for high-performance technical computing at Compaq. "The Alpha processor will live on for many years yet."
Compaq spokesman Dick Calandrella added that the move actually validates Compaq''s confidence in the technology. "There''s always been a question about our commitment to Alpha since we acquired it [from Digital Equipment Corp. in 1998]. Most of the customers that we talk to are pleased that it will now be manufactured forever."
While Compaq does indeed intend to make the next generation of the Alpha family, the EV7, its last, the Alpha technology will be incorporated in the new generation of Itanium microprocessors scheduled to appear in the 2004-2005 timeframe. Compaq said that Alpha design features like SMT, on-chip interconnects, and high-speed I/O processors should remain in the new Itanium chips. Compaq will also port its Tru64 Unix, OpenVMS, and Nonstop Kernel operating systems and development tools to the Itanium processor family.
Binns noted that Compaq customers would benefit from an anticipated reduction in the cost of processors.
"By joining forces with Intel, Compaq will inject critical technology into the Itanium processor family roadmap while increasing the investment on key technologies above the chip level in our high performance technical computing systems," wrote William Blake, Compaq''s vice president of high performance technical computing, in a letter to Compaq customers.
Phil Butcher, head of information technology at the Sanger Center, which currently has between 500 and 600 AlphaServer systems, noted that Intel''s market penetration would eventually level the playing field for 64-bit technology, "and the differentials will be in operating system features, systems built around the new IA-64 technology, and other novel technologies" that Compaq could provide.
"The announcement last week was surprising but does make sense for Compaq and customers," Butcher said.
Compaq''s decision falls in line with its recent announcement that it would restructure its operations to focus on software and services rather than hardware. But the question remains whether such a move is necessary in the life sciences market, where the company has maintained a strong position despite heated competition from IBM and Sun. Compaq''s AlphaServers currently power some of the field''s most high-profile bioinformatics projects at Celera Genomics, GeneProt, the Sanger Center, the Whitehead Institute, Genentech, and the Institute for Genomic Research, and a recent International Data Group report indicated that Compaq had a considerable lead over its next nearest rival in the life sciences technical computing market. But a shift toward services in this market will actually place Compaq behind IBM, which has been aggressively promoting its software and services in the space for almost a year.
Furthermore, the decision to phase out the mature Alpha architecture in favor of the still-unproven Itanium chips could be risky. Binns conceded that the Itanium chip "is not yet proven in the life sciences market," but noted, "Alpha will continue to serve our life science customers better than any competing processor until our newly announced joint development collaboration with Intel comes to fruition."
Compaq also intends to continue using the Alpha chips in its collaboration with Celera and Sandia National Laboratories to develop a supercomputer capable of achieving 100 TeraOPS by 2004. "The decision will have no effect on operations," Calandrella said. "Our customers will continue to use the Alpha in all projects without interruption." Butcher said Compaq has committed to supporting existing Compaq customers up to 2009.
"The decision has been very warmly accepted by the life science market at large," said Binns.
And despite some speculation in the industry that the Alpha phase-out may have played some role in
Myriad''s recent choice of Sun Microsystems over Compaq to provide the compute power behind its proteomics work, Binns said he was unaware of any recent orders being pulled as a result of the decision.
"I haven''t heard one dissenting voice" from Compaq''s current life science computing customers about the decision, said Binns. "As far as I understand, Myriad''s decision was based on cost."
CÈdric Loiret-Bernal, CEO of GeneProt, which recently opened a large-scale proteomics facility powered by 1,420 Alpha-based Tru64 Unix processors, said, "We feel very comfortable that the new roadmap will ensure that we have access to the best available technology today and tomorrow."
"Compaq has always been very good in support of its products and I see no reason for this to change," said Butcher.