Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!ucbvax!agate!shelby!portia!hanauma!rick From: rick@hanauma.stanford.edu (Richard Ottolini) Newsgroups: comp.graphics Subject: Re: ardent titan Message-ID: <5367@portia.Stanford.EDU> Date: 16 Sep 89 17:54:20 GMT References: <6495@ux.cs.man.ac.uk> Sender: USENET News System Reply-To: rick@hanauma.UUCP (Richard Ottolini) Organization: Stanford University, Dept. of Geophysics Lines: 24 In article <6495@ux.cs.man.ac.uk> toby@r2.cs.man.ac.uk writes: >I have an Ardent Titan to use, and I want to exploit its computing power, >really using its display just as a frame buffer. Does anyone know the most >efficient way for me to write to the display? (This is for a Mandelbrot >viewer). > >I've heard X is very slow for shipping pixels around. > >Thanks! >Toby a (1) Ardent shipped a new X server in June that was eight times faster than before. The old was based on the sample MIT server. The new uses device- dependent routines. It is now about the same speed as DEC 3100's X which uses the same CPU chip and is also optimized. (2) Both Ardent's new X, Dore, and some alpha-release scientific visualization tools call an underlying device driver called tg. There is a debate withing Ardent as to whether they will ever document this for TITAN users. You can get a sense of it by looking at the ioctl flags in /usr/include/machine/tigr.h. It seems to be a nasty, but very fast graphics driver involving queue and buffer synchronization. (We have two TITANS.)