diff options
author | Rob Clark <[email protected]> | 2014-01-07 10:55:07 -0500 |
---|---|---|
committer | Rob Clark <[email protected]> | 2014-01-08 16:30:18 -0500 |
commit | c0766528baaef48902c87bbdaa4f5926c472269b (patch) | |
tree | f825706059f50c37a9fda7961ec596b0fb6a65da /src/gallium/drivers/freedreno/freedreno_draw.h | |
parent | bfb44c24bc1eff850d47984b2cb60c957ffc143d (diff) |
freedreno/a3xx: support for hw binning pass
The binning pass sorts vertices into which bins/tiles they apply to.
The visibility information generated during the binning pass can be
used to speed up the rendering pass by filtering out vertices which
do not apply to the current tile. See:
https://github.com/freedreno/freedreno/wiki/Adreno-tiling#optimized-approach
This brings a significant fps boost. A rough assortment of tests
(supertuxkart, etracer, tremulous, glmark2 'build' test, etc) seems
to yield a ~35-45% fps improvement.
For now, to be conservative, the binning pass is not enabled yet by
default. To enable it use:
FD_MESA_DEBUG=binning
So far I haven't found anything that breaks with binning enabled,
but I'd like a bit more testing before I enable it as default.
Signed-off-by: Rob Clark <[email protected]>
Diffstat (limited to 'src/gallium/drivers/freedreno/freedreno_draw.h')
-rw-r--r-- | src/gallium/drivers/freedreno/freedreno_draw.h | 23 |
1 files changed, 16 insertions, 7 deletions
diff --git a/src/gallium/drivers/freedreno/freedreno_draw.h b/src/gallium/drivers/freedreno/freedreno_draw.h index 190c0e52d24..e8bb420889e 100644 --- a/src/gallium/drivers/freedreno/freedreno_draw.h +++ b/src/gallium/drivers/freedreno/freedreno_draw.h @@ -38,19 +38,21 @@ struct fd_ringbuffer; -void fd_draw_emit(struct fd_context *ctx, const struct pipe_draw_info *info); +void fd_draw_emit(struct fd_context *ctx, struct fd_ringbuffer *ring, + enum pc_di_vis_cull_mode vismode, + const struct pipe_draw_info *info); void fd_draw_init(struct pipe_context *pctx); static inline void -fd_draw(struct fd_context *ctx, enum pc_di_primtype primtype, +fd_draw(struct fd_context *ctx, struct fd_ringbuffer *ring, + enum pc_di_primtype primtype, + enum pc_di_vis_cull_mode vismode, enum pc_di_src_sel src_sel, uint32_t count, enum pc_di_index_size idx_type, uint32_t idx_size, uint32_t idx_offset, struct fd_bo *idx_bo) { - struct fd_ringbuffer *ring = ctx->ring; - /* for debug after a lock up, write a unique counter value * to scratch7 for each draw, to make it easier to match up * register dumps to cmdstream. The combination of IB @@ -64,7 +66,7 @@ fd_draw(struct fd_context *ctx, enum pc_di_primtype primtype, OUT_PKT3(ring, CP_DRAW_INDX, 3); OUT_RING(ring, 0x00000000); OUT_RING(ring, DRAW(1, DI_SRC_SEL_AUTO_INDEX, - INDEX_SIZE_IGN, IGNORE_VISIBILITY)); + INDEX_SIZE_IGN, USE_VISIBILITY)); OUT_RING(ring, 0); /* NumIndices */ /* ugg, hard-code register offset to avoid pulling in the @@ -76,8 +78,15 @@ fd_draw(struct fd_context *ctx, enum pc_di_primtype primtype, OUT_PKT3(ring, CP_DRAW_INDX, idx_bo ? 5 : 3); OUT_RING(ring, 0x00000000); /* viz query info. */ - OUT_RING(ring, DRAW(primtype, src_sel, - idx_type, IGNORE_VISIBILITY)); + if (vismode == USE_VISIBILITY) { + /* leave vis mode blank for now, it will be patched up when + * we know if we are binning or not + */ + OUT_RINGP(ring, DRAW(primtype, src_sel, idx_type, 0), + &ctx->draw_patches); + } else { + OUT_RING(ring, DRAW(primtype, src_sel, idx_type, vismode)); + } OUT_RING(ring, count); /* NumIndices */ if (idx_bo) { OUT_RELOC(ring, idx_bo, idx_offset, 0, 0); |