Supercomputer maker Cray said Monday that it is planning to release a line of products based on the Advanced Micro Devices-powered Red Storm machine it is building for the Department of Energy.
Cray Forecasts Red Storm for Masses
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
rewrite os x from the ground up so that it runs in amd only in LONG mode, and apple shall survive should ibm discontinue high-end powerpc support
…that Cray, a wholly owned subsidiary of SGI, would be producing an AMD64 system that serves as an appealing alternative to the IA64-based Altix?
>a wholly owned subsidiary of SGI
It is has being sold back to another company. SGI doesn’t have a big stake on Cray anymore.
Bascule wrote:
…that Cray, a wholly owned subsidiary of SGI, would be producing an AMD64 system that serves as an appealing alternative to the IA64-based Altix?
Well, I guess all money is good money
And wasn’t the relations between SGI and Cray a bit shaky?
Gein
Will Cray gradually stop producing their own chip and instead become something like the <cough> Dell <cough> of supercomputers ?
When Virginia Tech can build a G5-based supercomputer for $5 million in a few weeks, why is it that laboratories such as Sandia keep wasting taxpayers money on products that will cost 18 times more and will take months to assemble ? Is that the kind of efficiency we can expect from the Energy Department ? I hope one day the Congress will tell those guys they shouldn’t behave like kids in a candy store. At least, the Japanese don’t use their hardware to watch useless stuff (nuclear explosions) in virtual reality.
When Virginia Tech can build a G5-based supercomputer for $5 million in a few weeks, why is it that laboratories such as Sandia keep wasting taxpayers money on products that will cost 18 times more and will take months to assemble?
Interconnect. These systems will most likely utilize HyperTransport/NUMA to achive maximum memory bandwidth, with memory shared amoung processors. This is great if I/O overhead would prevent maximum processor utilization by their modelling code if a slower interconnect mechanism were used.