X

SGI, Unisys to use Merced

The two companies detail plans to use Intel's new 64-bit chip in high-powered machines.

Stephen Shankland Former Principal Writer
Stephen Shankland worked at CNET from 1998 to 2024 and wrote about processors, digital photography, AI, quantum computing, computer science, materials science, supercomputers, drones, browsers, 3D printing, USB, and new computing technology in general. He has a soft spot in his heart for standards groups and I/O interfaces. His first big scoop was about radioactive cat poop.
Expertise Processors, semiconductors, web browsers, quantum computing, supercomputers, AI, 3D printing, drones, computer science, physics, programming, materials science, USB, UWB, Android, digital photography, science. Credentials
  • Shankland covered the tech industry for more than 25 years and was a science writer for five years before that. He has deep expertise in microprocessors, digital photography, computer hardware and software, internet standards, web technology, and more.
Stephen Shankland
3 min read
Silicon Graphics and Unisys have thrown their weight behind Intel's IA-64 architecture, detailing plans today for rolling out the new 64-bit chip in high-powered machines.

The two corporate computer companies announced new systems that will take advantage of Intel's first chip using the IA-64 architecture, code-named Merced.

Early in the year 2000, Silicon Graphics (SGI) plans to release a high-performance computer based on its ccNUMA architecture that can tackle technical tasks by performing up to 1 trillion mathematical calculations a second.

And Unisys announced plans for large corporate servers that initially will be able to use both Intel's Pentium II Xeon and Merced chips.

The first versions of SGI's system will be based on MIPS processors, but users will be able to swap out modules to put in IA-64 processors.

The systems are designed for customers performing intense computations, such as engineers modeling airplane designs or financial analysts simulating the global economy, said Ben Passarelli, product manager for high-end servers at Silicon Graphics.

The first systems based on Merced processors will be available in mid- to late 2000, SGI spokeswoman Ginny Babbitt said. Merced's successor, McKinley, will be used starting in early 2001. Meanwhile, SGI will continue selling systems based on the MIPS chip during an overlap period.

"The board-swap strategy is very positive," said Chris Willard, a research director at International Data Corporation. By being able to switch out MIPS processor boards with IA-64 processor boards, a customer still can keep the same input-output system and other computing infrastructures.

However, changing processors means that software has to change too, and that's a part of the transition that customers won't overlook.

"Lots of companies, including Silicon Graphics, have indicated that their future road map is to move to IA-64. The winners in this business are the ones that can address those conversion issues in a way that will be least painful to the actual buyers making the transitions," Willard said.

Passarelli said he expects two phases in the transition.

First, the early adopters who desire the fastest new systems will quickly buy the new Merced systems and begin porting their software to the new platform. Later, the more conservative customers--financial companies, for example--will wait until the operating system and software have been recompiled and tested for the new IA-64 platform.

That delay means a lot of customers will be interested in the new Silicon Graphics machines about the same time they'll be equipped with the McKinley chip, which is expected to ship about 18 months after Merced, Passarelli said.

Partly for the benefit of these more cautious organizations, Silicon Graphics will sell systems with the MIPS chip at least through the introduction of McKinley, Passarelli said.

The IA-64 architecture is good for Silicon Graphics' technical computing requirements because they can communicate with memory faster than existing chips and have more on-chip storage space devoted to performing floating-point calculations, Passarelli said. IA-64 makes it easier to keep the processor fed with data and calculating instead of waiting for the next data to come in.

The new systems, which use the SGI's ccNUMA architecture, will be available with as few as four processors and as many as 512, said Ben Passarelli, product line manager for high-end servers at SGI. The processors come packaged on modules that include memory and between one and four processors.

The ccNUMA architecture allows lots of processors to tackle computing tasks without getting bogged down with processors keeping track of the status of all the other processors. Passarelli said ccNUMA lets Silicon Graphics sell systems that can be expanded by adding more and more processors.

IA-64 also will be the foundation of Unisys's next generation of high-performance corporate enterprise servers.

Although the first such systems will ship with today's Xeon chips, users will be able to add Merced processors within the same system.

Unisys will use an architecture called Cellular MultiProcessing to tie together the components of its multiprocessor systems.

The first systems based on that architecture are scheduled to ship in 1999. They will have a "shared-everything" design, so that each processor in the system gets access to as much as 64GB of memory and 96 PCI ports.

The Unisys computers are designed for companies needing to run business-critical "data center" tasks requiring uninterrupted availability, security, high-volume communications, and the ability to add new processors.

Intel is an investor in CNET: The Computer Network, publisher of News.com.