Received: by karazm.math.UH.EDU id AA26256 (5.65c/IDA-1.4.4 for glove-list@karazm.math.uh.edu); Wed, 23 Oct 1991 13:00:54 -0500 From: J Eric Townsend Message-Id: <199110231800.AA26256@karazm.math.UH.EDU> Subject: Re: Transputers... To: newton@neworder.ils.nwu.edu (Dave Newton) Date: Wed, 23 Oct 91 13:00:54 CDT Cc: glove-list@karazm.math.uh.edu In-Reply-To: <9110231722.AA00815@neworder.ils.nwu.edu>; from "Dave Newton" at Oct 23, 91 12:22 pm X-Mailer: ELM [version 2.3 PL11] >> The only question is whether 100Mbps interconnect is fast enough. > Um, I would certainly hope so. That two-page article about the VPL thing >was using straight ethernet for their interconnects. If 100M isn't fast >enough, there's a serious software problem, I'd have to say. Ether isn't fast enough for parallel computing.. Intel found this out with the iPSC/1. The iPSC/2 and iPSC/860 use proprietary "Direct Connect" technology that runs dedicated bit-stream lines w/ almost no protocol from one node to the next. -- J. Eric Townsend - jet@uh.edu - Systems Wrangler, UH Dept of Mathematics vox: (713) 749-2126 '91 CB750, DoD# 0378, TMRA# 27834 PowerGlove mailing list: glove-list-request@karazm.math.uh.edu "allow users to create more impactful documents" -- from an Apple press release