On Thu, Apr 26, 2012 at 8:54 PM, Jose Fonseca <jfons...@vmware.com> wrote: > Otherwise I don't see anything wrong with this series. It seems a nice > cleanup/speedup. > > Brian's OOTO till Monday, so allow more time for him to comment. > > Also, once you updated the series, please provide it in a clonable git branch > for testing.
Hi, I pushed the branch in the main Mesa repository. r600g, softpipe and llvmpipe should work without regressions. I'll test r300g later. Summary: - st/mesa optionally uploads index and constant buffers - set_constant_buffer has been updated to support the BindBufferRange functionality (for UBOs) - a plain pointer to a user buffer has been added into pipe_vertex_buffer, pipe_index_buffer, and new pipe_constant_buffer, deprecating the user_buffer_create function; this results in a small speed-up in some CPU-bound apps. - all state trackers have been adapted not to use user_buffer_create (except for graw tests) FYI, these are the new commits which are follow-ups to the series sent to mesa-dev earlier: - st/mesa: remove more unnecessary code in draw_vbo - st/mesa: reorder code in draw_vbo - gallium: add void *user_buffer to pipe_constant_buffer - radeonsi: don't create temporary user buffer for r600_upload_const_buffer - gallium: remove pipe_resource::user_ptr - gallium/util: stop using user buffers in util_draw_texquad - st/xorg: fix compilation - wrong libkms include file - cso: cso_context should install u_vbuf by itself and not st/mesa - st/vega: use cso_draw_arrays - st/vega: don't use user_buffer_create - st/xa: don't use user_buffer_create - st/xorg: don't use user_buffer_create Feel free to review and test. Thank you. Marek _______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
