diff options
| author | Bas Nieuwenhuizen <[email protected]> | 2016-03-27 11:14:34 +0200 |
|---|---|---|
| committer | Bas Nieuwenhuizen <[email protected]> | 2016-04-19 18:10:31 +0200 |
| commit | 7a92c0842892bf55a82b7d95ab5a3b7dfbb83407 (patch) | |
| tree | 67bea5002d5768b610169a98f6e69e59f502dac3 /src/gallium/drivers/radeonsi/si_cp_dma.c | |
| parent | e764ee13ae21e3c1dbda24daeb2d08c5e7c81871 (diff) | |
radeonsi: do not do two full flushes on every compute dispatch
v2: Add more CS_PARTIAL_FLUSH events.
Essentially every place with waits on finishing for pixel shaders
also has a write after read hazard with compute shaders.
Invalidating L2 waits implicitly on pixel and compute shaders,
so, we don't need a CS_PARTIAL_FLUSH for switching FBO.
v3: Add CS_PARTIAL_FLUSH events even if we already have INV_GLOBAL_L2.
According to Marek the INV_GLOBAL_L2 events don't wait for compute
shaders to finish, so wait for them explicitly.
Signed-off-by: Bas Nieuwenhuizen <[email protected]>
Reviewed-by: Marek Olšák <[email protected]>
Reviewed-by: Nicolai Hähnle <[email protected]>
Reviewed-by: Edward O'Callaghan <[email protected]>
Diffstat (limited to 'src/gallium/drivers/radeonsi/si_cp_dma.c')
| -rw-r--r-- | src/gallium/drivers/radeonsi/si_cp_dma.c | 6 |
1 files changed, 4 insertions, 2 deletions
diff --git a/src/gallium/drivers/radeonsi/si_cp_dma.c b/src/gallium/drivers/radeonsi/si_cp_dma.c index 001ddd4bfae..38e0ee60d64 100644 --- a/src/gallium/drivers/radeonsi/si_cp_dma.c +++ b/src/gallium/drivers/radeonsi/si_cp_dma.c @@ -190,7 +190,8 @@ static void si_clear_buffer(struct pipe_context *ctx, struct pipe_resource *dst, uint64_t va = r600_resource(dst)->gpu_address + offset; /* Flush the caches. */ - sctx->b.flags |= SI_CONTEXT_PS_PARTIAL_FLUSH | flush_flags; + sctx->b.flags |= SI_CONTEXT_PS_PARTIAL_FLUSH | + SI_CONTEXT_CS_PARTIAL_FLUSH | flush_flags; while (size) { unsigned byte_count = MIN2(size, CP_DMA_MAX_BYTE_COUNT); @@ -296,7 +297,8 @@ void si_copy_buffer(struct si_context *sctx, } /* Flush the caches. */ - sctx->b.flags |= SI_CONTEXT_PS_PARTIAL_FLUSH | flush_flags; + sctx->b.flags |= SI_CONTEXT_PS_PARTIAL_FLUSH | + SI_CONTEXT_CS_PARTIAL_FLUSH | flush_flags; /* This is the main part doing the copying. Src is always aligned. */ main_dst_offset = dst_offset + skipped_size; |