There's no reason we can't -- the mappings we expose are basically
equivalent to persistent/coherent, already.
Improves mesa-demos drawoverhead (no state change) performance by
5.21362% +/- 1.25078% (n=11).
case PIPE_CAP_TEXTURE_MULTISAMPLE:
case PIPE_CAP_TEXTURE_SWIZZLE:
case PIPE_CAP_GLSL_OPTIMIZE_CONSERVATIVELY:
+ case PIPE_CAP_ALLOW_MAPPED_BUFFERS_DURING_EXECUTION:
return 1;
/* lying for GL 2.0 */
case PIPE_CAP_TGSI_BALLOT:
case PIPE_CAP_TGSI_TES_LAYER_VIEWPORT:
case PIPE_CAP_CAN_BIND_CONST_BUFFER_AS_VERTEX:
- case PIPE_CAP_ALLOW_MAPPED_BUFFERS_DURING_EXECUTION:
case PIPE_CAP_POST_DEPTH_COVERAGE:
case PIPE_CAP_BINDLESS_TEXTURE:
return 0;