Return to BSD News archive
Xref: sserve comp.os.386bsd.questions:6510 comp.windows.x:60594 Newsgroups: comp.os.386bsd.questions,comp.windows.x Path: sserve!newshost.anu.edu.au!munnari.oz.au!news.Hawaii.Edu!ames!elroy.jpl.nasa.gov!swrinde!cs.utexas.edu!math.ohio-state.edu!howland.reston.ans.net!pipex!uknet!festival!castle.ed.ac.uk!richard From: richard@castle.ed.ac.uk (Richard Tobin) Subject: Re: XFree86 2.0 Experiences References: <CFvyn1.732@aib.com> <CFxApF.C01@festival.ed.ac.uk> <CG044M.185@aib.com> Message-ID: <CG0wn3.LME@festival.ed.ac.uk> Sender: news@festival.ed.ac.uk (remote news read deamon) Organization: University of Edinburgh Date: Fri, 5 Nov 1993 14:37:50 GMT Lines: 27 In article <CG044M.185@aib.com> dwex@aib.com (David E. Wexelblat) writes: >Well, look at it this way. Assuming no raster-op and no planemask >(i.e the vast majority of the cases), setting a pixel in 8bpp means >loading a byte. Setting a pixel at 1bpp means read a byte, modify >the bit, write the byte (of course, this gets optimized when spans >are 8 or more pixels long). In dumb-VGA 4bpp mode, you have to >do this read-modify-write 4 times. The thing I get worst (ie most annoying) performance on is xterm scrolling. This is a case where the mono server wins big. Presumably this is a case where the 4 bit server could approach twice the speed of the 8 bit server. >>>>(4) The mono server doesn't work at all. >>>Run it in "generic" mode. Thanks for the tip. It now works fine for up to 800x655. On a separate point, the Link Kit is excellent. I thought I was going to have to get the whole MIT distribution to try any experiments, but the Link Kit makes it really easy. One small bug - at line 175 in include/xf86_OSlib.h LINKKKIT is spelt with 3 Ks. -- Richard -- "Beat your ploughshares into swords and your pruninghooks into spears" (Joel 3:10)