Subject: Re: What now, tech?
From: marc ainger (email@example.com)
Date: Sat Dec 14 1996 - 10:46:12 EST
Jean Piche wrote:
>would love to see benchmarks, but I am about to make the claim that an
>r5kSC running Csound pumps out more oscilators than a pair of i860s
>running FTS.... The Kyma system probably comes close on a 8x64k setup...
>Again, this is very imprudent speculation on my part... The point is that
>its most likely getting too close to call. Costwise, high-end cpus are much
>cheaper than specialized hardware solutions...
>IRCAM/FTS does run on the SGI, albeit not as smoothly as on the i860. SGI
>has in fact replaced the i860 as the platform of choice at IRCAM.
At the last ICMC (96, Hong Kong), IRCAM gave a presentation in which they
revealed that the Mac Power PC was currently their platform of choice for
further MAX/FTS development. They have done some performance testing,
which they admitted was rough, and, came to the conclusion that a MAC 9500
PPC at 132 MHz was almost three times more powerful than a single ISPW
(i860, 40MHz), and slighly faster than an SGI Indy (R4400, 200 MHz). This,
of course, was before the release of the O2.
But the Mac PPC is currently being released in versions with multiple
processors, including a 4 processor version from Daystar which IRCAM is
also developing for. They are also taking advatage of some Open Doc
implementation possibilities -- I refer you to the two articles in the ICMC
Proceedings (pp 185-189).
There are tradeoffs here, too-- a multiprocessor PPC will run as much if
not more than an O2 (although the price will come down, and there are none
of the support problems which have been mentioned in relation to the O2);
also, I would be concerned about the stability of the software, especially
since it is a "closed", ie, non-hackable, system (no source code
Anyway, this is not an endorsement either way, strictly FYI-- I like the
This archive was generated by hypermail 2b27 : Wed Jun 11 2003 - 13:12:40 EDT