> > the other problem with the G450 is that when running with 2 heads, > > the memory isn't split 50/50; it's split 75/25. the second head only gets > > 8MB of video ram. ~~;( > > Have you tried either > Option "Crtc2Half" # [<bool>] > or > Option "Crtc2Ram" # <i> > ? Without either of these options crtc2 is limited to 8MB, since it > is more limited in the resolutions and refresh rates that it supports.
aha. so that's what that option does. I'll give it a try soon. it's not documented in the mga man page (yet), and the name isn't intuitive, so I didn't guess to try that option. :) > > does anyone know if the dual-head Radeons do anything similarly stupid? > I understand that the Radeon driver drives the 2 heads in a completely > different way from the Matrox driver; the XFree86 code for the MGAs takes > the easy way out and pretends that there are two independant frame buffers, > while the Radeon driver has two windows into one large frame buffer, > so this problem shouldn't arise. interesting. I see that www.compgeeks.com has dual-head Radeon VEs for $70. maybe I ought to get one and try it out. > > what about the G550? > Same drivers as the G400 and G450. Don't know how the Matrox provided > driver works on any of these cards. I used to use the Matrox binary drivers a long time ago, before X supported the G450 well. didn't like them vey much at all. :( as it is, when I get around to it, I'll try building CVS X with the matrox driver source patched in, to see if that fixes my DRI problems. I don't have much hope, tho. :( > > also keep in mind that you'll need to turn up the AGP speed on the > > second head in order to get a decent dotclock limit. otherwise your max > > refresh rates will *suck*. > Is the driver using the G400 limits, where the second head was very > limited ? no, it's not the G400 limits; at AGP 1x, the dotclock maximum is actually *lower* than a G400s. 112Mhz, I believe, where I think the G400 limits the dotclock to 120-something. I don't think my workstation at work (where I have a G400) supports more than AGP 1x, tho; so I can't test the G400 at higher AGP rates. Carl Soderstrom -- Network Engineer Real-Time Enterprises (952) 943-8700 _______________________________________________ Xpert mailing list [EMAIL PROTECTED] http://XFree86.Org/mailman/listinfo/xpert