On Tue, 20 Nov 2001, Ani Joshi wrote: >> Well, I've now confirmed that we are not on crack. ;o) People >> who claim they're using Radeon with DRI enabled and not having >> any 2D slowdowns, are indeed on crack though, as the source code >> itself, as well as developers have now confirmed to me that the >> 2D accel code is not fully implemented yet under DRI. > > >The radeon 2D acceleration while using DRI (using the CP >engine) is relatively compelte and should not really cause any >slowdown (I personally have not seen any 2d performance loss >with DRI enabled with regular radeon, 7500, or 8500).
Just to be sure that we are comparing apples to apples, I am refering to XFree86 4.1.0. Since I have brought this up now, I have been inundated with email from users of various Linux distros and BSD OS's confirming the 2D slowdown when DRI is enabled also. So I agree, it shouldn't be slow, but it definitely is in 4.1.0 at least. I see that the code in the CVS head adds more 2D acceleration, however I have yet to test it out and see if it solves the slowdown I am seeing. Users running in lower resolutions, or with other different variables may not see the horrible slowdown as bad as I am seeing it at 1600x1200 also. >If you look at the source, the CP engine accelerates solid >fills, screen copies, dashed lines, and solid lines. While the >latter 2 aren't extremely necessary, the first 2 are indeed >essential and are probably the only 2 primitives which really >matter when it comes to 2D acceleration with todays fast CPU's >(with perhaps the exception of expansion, which is arguable). Well, I can't speak for the CVS head right now, as I am not using it currently, but looking at the code in 4.1.0, there is no question the slowdown is caused by lack of acceleration. Comparing that code with CVS code shows that some work has been done since then. Wether it implements the necessary features to speed up the slowdowns I am seeing or not, I'm not sure. >I'll try again with cvs and see if I can reproduce this with my >radeon boards. Ok, be sure to run it in 24bit depth, at the highest resolution you can, preferably 1600x1200, or even better, at 1920x1440 if you can. Test it using 4.1.0, with and without DRI enabled. CPU speed doesn't have much effect speeding it up, as I'm using a dual 1Ghz P-III. A comparision between 4.1.0 and CVS would be great. I suppose I could produce an x11perf chart for comparison also. That would be interesting to compare with CVS also perhaps.. Thanks for the info Ani. -- ---------------------------------------------------------------------- Mike A. Harris Shipping/mailing address: OS Systems Engineer 190 Pittsburgh Ave., Sault Ste. Marie, XFree86 maintainer Ontario, Canada, P6C 5B3 Red Hat Inc. Phone: (705)949-2136 http://www.redhat.com ftp://people.redhat.com/mharris Red Hat XFree86 mailing list: [EMAIL PROTECTED] General open IRC discussion: #xfree86 on irc.openprojects.net ---------------------------------------------------------------------- _______________________________________________ Dri-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/dri-devel
