diff options
| author | Jason Ekstrand <[email protected]> | 2019-07-31 10:42:24 -0500 |
|---|---|---|
| committer | Jason Ekstrand <[email protected]> | 2019-08-06 05:46:28 +0000 |
| commit | bc612536eb2f0f77725103e53077ad5a28036ac3 (patch) | |
| tree | 50eaea505ca1751c56d6ded905c90d4689e92d7b | |
| parent | 5544b2cbbd23df82192aea09d909b5cc2c1f1af9 (diff) | |
anv: Emit a dummy MEDIA_VFE_STATE before switching from GPGPU to 3D
There is an object-level preemption workaround which requires this.
However, even without object-level preemption, we seem to have issues
with geometry flickering when 3D and compute are combined in the same
batch and this appears to fix it.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=109630
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=111267
Cc: [email protected]
Reviewed-by: Kenneth Graunke <[email protected]>
| -rw-r--r-- | src/intel/vulkan/genX_cmd_buffer.c | 19 |
1 files changed, 19 insertions, 0 deletions
diff --git a/src/intel/vulkan/genX_cmd_buffer.c b/src/intel/vulkan/genX_cmd_buffer.c index 5b76ac61679..a91cffd8fc7 100644 --- a/src/intel/vulkan/genX_cmd_buffer.c +++ b/src/intel/vulkan/genX_cmd_buffer.c @@ -3827,6 +3827,25 @@ genX(flush_pipeline_select)(struct anv_cmd_buffer *cmd_buffer, anv_batch_emit(&cmd_buffer->batch, GENX(3DSTATE_CC_STATE_POINTERS), t); #endif +#if GEN_GEN == 9 + if (pipeline == _3D) { + /* There is a mid-object preemption workaround which requires you to + * re-emit MEDIA_VFE_STATE after switching from GPGPU to 3D. However, + * even without preemption, we have issues with geometry flickering when + * GPGPU and 3D are back-to-back and this seems to fix it. We don't + * really know why. + */ + const uint32_t subslices = + MAX2(cmd_buffer->device->instance->physicalDevice.subslice_total, 1); + anv_batch_emit(&cmd_buffer->batch, GENX(MEDIA_VFE_STATE), vfe) { + vfe.MaximumNumberofThreads = + devinfo->max_cs_threads * subslices - 1; + vfe.NumberofURBEntries = 2; + vfe.URBEntryAllocationSize = 2; + } + } +#endif + /* From "BXML » GT » MI » vol1a GPU Overview » [Instruction] * PIPELINE_SELECT [DevBWR+]": * |