Commit
30259856a8a82a55c030df1ad052e505c61144bc moved the state packets to
table generation time, but forgot to make this change. Apparently the
performance win there was about not reemitting the table pointers on
unrelated state changes.
No performance difference on cairo on glamor (n=118).
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
.mesa = _NEW_TRANSFORM | _NEW_PROGRAM_CONSTANTS,
.brw = (BRW_NEW_CONTEXT |
BRW_NEW_GEOMETRY_PROGRAM |
- BRW_NEW_GS_BINDING_TABLE |
BRW_NEW_BATCH |
BRW_NEW_PUSH_CONSTANT_ALLOCATION),
.cache = CACHE_NEW_GS_PROG
.mesa = _NEW_TRANSFORM | _NEW_PROGRAM_CONSTANTS,
.brw = (BRW_NEW_CONTEXT |
BRW_NEW_VERTEX_PROGRAM |
- BRW_NEW_VS_BINDING_TABLE |
BRW_NEW_BATCH |
BRW_NEW_PUSH_CONSTANT_ALLOCATION),
.cache = CACHE_NEW_VS_PROG
_NEW_BUFFERS |
_NEW_MULTISAMPLE),
.brw = (BRW_NEW_FRAGMENT_PROGRAM |
- BRW_NEW_PS_BINDING_TABLE |
BRW_NEW_BATCH |
BRW_NEW_PUSH_CONSTANT_ALLOCATION),
.cache = (CACHE_NEW_WM_PROG)
.mesa = _NEW_TRANSFORM | _NEW_PROGRAM_CONSTANTS,
.brw = (BRW_NEW_CONTEXT |
BRW_NEW_GEOMETRY_PROGRAM |
- BRW_NEW_GS_BINDING_TABLE |
BRW_NEW_BATCH |
BRW_NEW_PUSH_CONSTANT_ALLOCATION),
.cache = CACHE_NEW_GS_PROG
.dirty = {
.mesa = _NEW_PROGRAM_CONSTANTS | _NEW_MULTISAMPLE,
.brw = BRW_NEW_FRAGMENT_PROGRAM |
- BRW_NEW_PS_BINDING_TABLE |
BRW_NEW_BATCH |
BRW_NEW_PUSH_CONSTANT_ALLOCATION,
.cache = CACHE_NEW_WM_PROG
.mesa = _NEW_TRANSFORM | _NEW_PROGRAM_CONSTANTS,
.brw = BRW_NEW_CONTEXT |
BRW_NEW_VERTEX_PROGRAM |
- BRW_NEW_VS_BINDING_TABLE |
BRW_NEW_BATCH |
BRW_NEW_PUSH_CONSTANT_ALLOCATION,
.cache = CACHE_NEW_VS_PROG