Hi, when I tried to compile Xdirectfb, it ended with no rule to make target `hw/directfb/libdirectfb.a` needed by `XDirectFB`
what could this be? I have DirectFB 0.9.20 DFBTerm is working I got the cvs version of XFree86 today On Thu, 5 Feb 2004, Denis Oliver Kropp wrote: > Quoting Colin Paton: > > Hi, > > > > I've had another play around with the CLE266 mpeg stuff (unlike the other > > night when I was trying to rush things and got it all wrong - D'oh. Ignore > > my previous posting about the HQV surface. More haste less speed....) > > > > Further to the suggestion from Ville Syrj�l� (Thanks for that) I decided to > > hack the driver to see if it could do YV12 blits as three separate Y,U,V > > blits. My results are encouraging - it can! > > > > It also seems to work quite well - I'm now getting about 6-8% CPU usage > > playing a video at 50fps, which is pretty good. > > ;) > > > I've created some proof of concept code. The uc_blit function has been > > altered so that it does 3 blits, but it also sets up the source and > > destination parameters (uc_set_destination and uc_set_source_2d now set some > > global variables which are referenced by uc_blit). > > I hope the "global variables" are stored in the device data struct > UcDeviceData, which resides in shared memory. > > > The code works but is rather ugly. The problem stems from the fact that the > > source and destination blit offsets etc are considered as part of the state > > of the card, whereas I now alter them for each blit. This mechanism is a > > good optimisation for 'normal' graphics drawing. > > Which optimization? > > > Is it possible to save register contents? One possible mechanism to > > integrate this more nicely into DirectFB is to add a uc_planar_blit call in > > addition to uc_blit, and set the function pointer to uc_blit or > > uc_planar_blit depending on the surface format. > > Yes, that would also make the Matrox driver a bit more readable. > > > In addition, I've added 'YV12' into uc_has_dst_format, which I believe might > > conceivably break functions that try to do a non-blit operation to a YV12 > > surface. > > You can check for the special case in uc_check_state(). > > -- > Best regards, > Denis Oliver Kropp > > .------------------------------------------. > | DirectFB - Hardware accelerated graphics | > | http://www.directfb.org/ | > "------------------------------------------" > > Convergence GmbH > > > -- > Info: To unsubscribe send a mail to [EMAIL PROTECTED] with > "unsubscribe directfb-dev" as subject. > -- Info: To unsubscribe send a mail to [EMAIL PROTECTED] with "unsubscribe directfb-users" as subject.
