Path: utzoo!utgpu!jarvis.csri.toronto.edu!cs.utexas.edu!tut.cis.ohio-state.edu!ucbvax!ucsfcgl!seibel
From: seibel@cgl.ucsf.edu (George Seibel)
Newsgroups: comp.sys.sgi
Subject: Re: Fast Fourier Transforms and the GVX
Message-ID: <13258@cgl.ucsf.EDU>
Date: 6 Mar 90 06:01:00 GMT
References: <1990Mar2.160357.10094@hellgate.utah.edu> <9003051643.AA14376@forest.sgi.com>
Sender: daemon@cgl.ucsf.edu
Reply-To: seibel@cgl.ucsf.edu (George Seibel)
Organization: Computer Graphics Lab, UCSF
Lines: 15

In article <9003051643.AA14376@forest.sgi.com> baskett%forest@SGI.COM writes:
]The new graphics subsystem is VGX, not GVX.  It is also refered to by
]the product literature as PowerVision.  It contains an "ImageVision
]Library" for FFT's, image rotations, warps, and convolutions directly
]in the graphics pipeline.  It does not use FFT chips in the graphics
]pipeline; it uses special code in the geometry subsystem to apply its
]128 mflops to image processing problems like FFT.  As to what FFT's are
 ^^^^^^^^^!
 You just got my attention here... Any chance of getting at these mflops
in a convenient way from high level code for general number crunching
purposes?   Is there enough precision to make it worthwhile?   You
wouldn't happen to have 1/sqrt(r) in there, would ya?

George Seibel, UCSF
seibel@cgl.ucsf.edu