Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!ucbvax!SIERRA.STANFORD.EDU!lacomb From: lacomb@SIERRA.STANFORD.EDU ("Lloyd J. Lacomb") Newsgroups: comp.sys.sgi Subject: Re: (l)rectwrite Message-ID: Date: 6 Mar 89 17:14:57 GMT Sender: daemon@ucbvax.BERKELEY.EDU Organization: The Internet Lines: 17 Like Mike I have noticed the slow performance of the (l)rectwrite routines when I have to draw each line individually as opposed to drawing all the lines in one call. Since the explanation seems to have to do with the internal hardware/DMA configuration, I have an additional question about (l)rectwrite. Why are the (l)rectwrite routines fixed to only accept only data in 1st quadrant (i.e. y2 > y1) format rather than being versitaile enough to figure out that if y2 < y1, the image is in 4th quadrant form. It would seem that the routines could then pipeline the data in some hardware effcient manner to the memory rather than having the user issue the call for each line of the data. I suppose that the routines should be general enough to deal with 2nd and 3rd quadrant data but since that is another issue lets keep the request simple. Thanks for any help on this matter. Lloyd LaComb lacomb@sierra.stanford.edu