New Fujitsu Computer (VPP500)

Article View: fj.mail-lists.parallel

Article #473

Back to fj.mail-lists.parallel Previous Next

#473

From: Allan Gottlieb
Date: Thu, 01 Oct 1992 15:29

14 lines
921 bytes

I have read the reports posted here and even heard a Fujitsu talk at
PACTA'92 in spain last week but still have two questions.

1.  What is the cycle time?  I can do backwards calculations from
MFLOPS and number of pipes but cannot be sure how many results each
pipe produces each cycle.

2.  Is it really a 200x200 crossbar (I know it is more than 200)?
Fujitsu slides say crossbar but also show pictures of omega like
networks.  Specifically, can all permutations be realized without
blocking?

Thanks,
allan gottlieb

Technical Details

Message-ID: <316686@nttlab.ntt.JP>

Path: arc1-ca.pugleaf.net!sparky!uunet!sun-barr!sh.wide!wnoc-kyo!kuis!kudpc!nuis!tutgw!lab!nttlab!daemon

Show Full Headers

Complete Article Headers

Path: arc1-ca.pugleaf.net!sparky!uunet!sun-barr!sh.wide!wnoc-kyo!kuis!kudpc!nuis!tutgw!lab!nttlab!daemon
From: cmcl2!allan.ultra.nyu.edu!gottlieb@uunet.UU.NET (Allan Gottlieb)
Newsgroups: fj.mail-lists.parallel
Subject: New Fujitsu Computer (VPP500)
Message-ID: <316686@nttlab.ntt.JP>
Date: 1 Oct 92 15:29:11 GMT
Sender: daemon@nttlab.ntt.JP
Distribution: fj
Lines: 14
Approved: news@nttlab.ntt