Bug 106800

Summary: intelReadPixels: fallback to CPU mapping in PBO case for Z24S8 attachments
Product: Mesa Reporter: Markus Wick <wickmarkus>
Component: Drivers/DRI/i965Assignee: Intel 3D Bugs Mailing List <intel-3d-bugs>
Status: RESOLVED MOVED QA Contact: Intel 3D Bugs Mailing List <intel-3d-bugs>
Severity: normal    
Priority: medium    
Version: git   
Hardware: x86-64 (AMD64)   
OS: Linux (All)   
i915 platform: i915 features:
Attachments: Apitrace dump of citra showing the performance issue

Description Markus Wick 2018-06-03 22:56:52 UTC
Created attachment 139997 [details]
Apitrace dump of citra showing the performance issue


calling glReadPixels for the GL_DEPTH24_STENCIL8 attachment and the GL_DEPTH_STENCIL / GL_UNSIGNED_INT_24_8 format seems to stall the GPU, even if a PBO is bound. 92% of my CPU time is spend in intel_miptree_map.

I guess i965 lacks a copy shader for the tiled Z24S8 attachment to the linear TBO buffer.

Tests was done on: Mesa DRI Intel(R) UHD Graphics 620 (Kabylake GT2)
and mesa: 4affeba1e9eb426a1ba13a3e8ced4673c4bb9b34

An apitrace dump which highlights this issue in the last frames is attached.
Comment 1 Tapani Pälli 2018-06-07 08:25:52 UTC
glReadPixels path for blorp (and tiled memcpy path) has been enabled for color buffers only, this case falls back to Mesa frontend.
Comment 2 GitLab Migration User 2019-09-25 19:11:46 UTC
-- GitLab Migration Automatic Message --

This bug has been migrated to freedesktop.org's GitLab instance and has been closed from further activity.

You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.freedesktop.org/mesa/mesa/issues/1730.

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.