summaryrefslogtreecommitdiffstats
path: root/src/intel/vulkan
diff options
context:
space:
mode:
authorJason Ekstrand <[email protected]>2017-03-15 11:58:52 -0700
committerJason Ekstrand <[email protected]>2017-03-28 14:57:08 -0700
commit6baae9625d26d282a72481598f9431fcad3211f6 (patch)
tree952fbf255ec64b3a11afe17e8f5a392c0eeb395d /src/intel/vulkan
parent0fe3dcce4c3e8b86a60beefe4c5adc760f2d59f8 (diff)
anv: Flush caches prior to PIPELINE_SELECT on all gens
The programming note that says we need to do this still exists in the SkyLake PRM and, from looking at the bspec, seems like it may apply to all hardware generations SNB+. Unfortunately, this isn't particularly clear cut since there is also language in the bspec that says you can skip the flushing and stall to get better throughput. Experimentation with the "Car Chase" benchmark in GL seems to indicate that some form of flushing is still needed. This commit makes us do the full set of flushes regardless of hardware generation. We can always reduce the flushing later. Reported-by: Topi Pohjolainen <[email protected]> Reviewed-by: Iago Toral Quiroga <[email protected]> Cc: "17.0 13.0" <[email protected]>
Diffstat (limited to 'src/intel/vulkan')
-rw-r--r--src/intel/vulkan/genX_cmd_buffer.c3
1 files changed, 1 insertions, 2 deletions
diff --git a/src/intel/vulkan/genX_cmd_buffer.c b/src/intel/vulkan/genX_cmd_buffer.c
index 5d923a8c08d..d0ddc29f008 100644
--- a/src/intel/vulkan/genX_cmd_buffer.c
+++ b/src/intel/vulkan/genX_cmd_buffer.c
@@ -2133,8 +2133,8 @@ flush_pipeline_before_pipeline_select(struct anv_cmd_buffer *cmd_buffer,
*/
if (pipeline == GPGPU)
anv_batch_emit(&cmd_buffer->batch, GENX(3DSTATE_CC_STATE_POINTERS), t);
+#endif
-#elif GEN_GEN <= 7
/* From "BXML » GT » MI » vol1a GPU Overview » [Instruction]
* PIPELINE_SELECT [DevBWR+]":
*
@@ -2160,7 +2160,6 @@ flush_pipeline_before_pipeline_select(struct anv_cmd_buffer *cmd_buffer,
pc.InstructionCacheInvalidateEnable = true;
pc.PostSyncOperation = NoWrite;
}
-#endif
}
void