SAN FRANCISCO--Intel has some tough decisions ahead of it at the
high end of the server world. It's got two processors, Xeon and
Itanium, selling into slightly different areas of the mainframe/supercomputer world.
These aren't your classic high-volume runners, so the company has
over time rejiggered its engineering teams under server development
VP and GM Rory McInerney to work on both. It's a way to maximize
engineering resources and share technology insights between the
"Power, I/O technology have come from Xeon into Itanium. RAS
(Reliability, Availability and Serviceability) has come from Itanium
into Xeon," McInerney (pictured) said in an interview.
The modular approach has reached the point where the only
difference between Xeon and Itanium is in the cores. The packaging, I/O, EDA tools, the chip set are common to both. In the
cores, the difference rests with the fact that Itanium runs the HP/UX instruction set
architecture while Xeon runs Windows and Linux.
In the scheme of things, this market lacks the
volumes of markets that Intel's Atom processor enjoys. For Itanium, there's not a huge customer base;
Hewlett-Packard, which rolled out new servers this week based on
Itanium, remains the big kid on the block.
Which begs the question: How much longer can Intel continue to
support two core designs running two ISAs at the high end?
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.