[Beowulf] Re:
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.
Geoff Jacobs gdjacobs at gmail.comMon Dec 17 12:40:22 PST 2007
- Previous message: [Beowulf] Re:
- Next message: [Beowulf] large array to run
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
G.M.Sigut wrote: > On Sat, 2007-12-15 at 12:01 -0800, beowulf-request at beowulf.org wrote: > ... >> 2. Re: large array to run (Geoff Jacobs) > ... >> Message: 2 >> Date: Fri, 14 Dec 2007 21:46:42 -0600 >> From: Geoff Jacobs <gdjacobs at gmail.com> >> Subject: Re: [Beowulf] large array to run >> To: Craig Tierney <ctierney at hypermall.net> >> Cc: beowulf at beowulf.org, Mark Hahn <hahn at mcmaster.ca> > ... >> Craig Tierney wrote: >> <snippage /> > ... >> IIRC, the CDC 6600 (and compatibles) had a 48 bit integer format as >> default. So, yes it does vary. I guess I was rolling the dice a little. > > My memory is telling me, that while the arithmetics on the 6000 series > might have been done with 48 bits, the memory assignment was in words. > One integer took one word of 60 bits (i.e. 5 bytes of 12 bits each). > > George > Right you are. 5 12 bit bytes. I was confusing with the FP coefficient. Still weird, which was the whole point. Namely, more obscure examples of hardware and software will have differences in data representation. -- Geoffrey D. Jacobs
- Previous message: [Beowulf] Re:
- Next message: [Beowulf] large array to run
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the Beowulf mailing list
