18:23 fdobridge: <p​homes> I pushed implementations of VK_EXT_pipeline_creation_feedback and VK_EXT_pipeline_creation_cache_control on top of my pipeline shader cache MR
18:26 fdobridge: <p​homes> gfxstrand: I tested with NAK too. It requires some small conflict handling. Are you interested in a branch that is based on NAK?
19:51 fdobridge: <g​fxstrand> Good question.
23:33 fdobridge: <g​fxstrand> Do we know what the SKED cache is?
23:33 fdobridge: <g​fxstrand> I'm thinking it's the cache for QMDs but I'm not sure
23:53 fdobridge: <g​fxstrand> https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25595/diffs?commit_id=2c90acdefef5d22fc28884f4b5fc2b8357f91e1d
23:53 fdobridge: <g​fxstrand> Well, that was fun... It only reproduces if you run these three tests:
23:54 fdobridge: <g​fxstrand> ```
23:54 fdobridge: <g​fxstrand> dEQP-VK.pipeline.monolithic.dynamic_offset.combined_descriptors.all_offsets.reverse_order_64_compute_first
23:54 fdobridge: <g​fxstrand> dEQP-VK.pipeline.monolithic.dynamic_offset.combined_descriptors.all_offsets.reverse_order_256_graphics_first
23:54 fdobridge: <g​fxstrand> dEQP-VK.pipeline.monolithic.push_constant.compute_pipeline.simple_test
23:54 fdobridge: <g​fxstrand> ```
23:54 fdobridge: <g​fxstrand> Those tests are about 170,000 tests apart in a full "I want conformance" CTS run.