[Beowulf] IBM Sequoia
Many of your questions may have already been answered in earlier discussions or in the FAQ. The search results page will indicate current discussions as well as past list serves, articles, and papers.
Kilian CAVALOTTI kilian.cavalotti.work at gmail.comWed Feb 4 10:06:19 PST 2009
- Previous message: [Beowulf] IBM Sequoia
- Next message: [Beowulf] IBM Sequoia
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
On Wednesday 04 February 2009 14:23:01 John Hearns wrote: > Yes, the Hpcwire article confirms 1.6Pbyte. > http://www.hpcwire.com/features/Lawrence-Livermore-Prepares-for-20-Petaflop >-Blue-GeneQ-38948594.html It also examines the core count: > "However, since Sequoia will be composed of 98,304 compute nodes and > contain a total of 1.6 million cores, one can surmise that a Blue > Gene/Q node will contain 16 cores. Whether this is implemented as one > 16-core chip or two 8-core chips (or even four quad-core chips) > remains to be seen." Seems right. According to http://www.cse.scitech.ac.uk/disco/mew17/talks/Keable_IBM_MEW17.pdf (slides 25 and following) BG/Q will feature a "similar number of nodes", but with "many more cores". And according to http://www.theregister.co.uk/2008/07/11/ibm_power7_ncsa/, POWER7 chips will feature 8 cores (and 4 threads per core!). Pack two of thoses in a compute card, 32 compute cards per node card, and 32 node cards per rack, it gives 1.52 million cores in 93 racks. Which marketing writes as 1.6 million, the count is right. :) But even so, with a TDP of say 100W for the POWER7 chip, it's about 19MW of heat to cool down. Just for the CPUs... Liquid nitrogen anyone? Cheers, -- Kilian
- Previous message: [Beowulf] IBM Sequoia
- Next message: [Beowulf] IBM Sequoia
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the Beowulf mailing list
