12:40 Triang3l: gerddie: Hey, can you please join /linuxgamingdev at Discord? I'll probably be asking lots of questions soon
12:41 Triang3l: like, I already have one about SSBO loads in SFN and CB/TC coherence within one invocation
12:42 Triang3l: And we'll need to keep in sync regarding locations of things in the source tree and the architecture
12:45 Triang3l: I'm currently aiming to separate SFN and r600_asm/eg_asm (at least eg_asm specifically) from gallium/drivers/r600
12:46 Triang3l: so they can be used in the Vulkan driver
12:47 Triang3l: It will have a very different layout of resource bindings and kcache buffers also, but I'm hoping to be able to resolve most of this via NIR lowerings
12:47 Triang3l: but this also means that it will need to be separated from r600_pipe.h
12:48 Triang3l: specifically, due to the extreme scarcity of resource slots when everything needed for D3D11 is allocated, I'll be merging some parts of what's currently in bufinfo into one buffer that will also contain push constants
12:50 Triang3l: like draw ID for shader draw parameters, tessellation ring layout, sample locations, and also (ewwww AMD just why…) cubemap array layer counts
12:51 Triang3l: (the latter will probably be tightly packed, and also accessed relatively to the base offset for each shader stage also passed alongside push constants, to support VK_EXT_shader_object)
12:52 Triang3l: Overall, by the way, I'm moving everything that can be common from gallium/drivers/r600 to amd/terascale (currently not the files themselves so I can still rebase, but in meson.build at least)
12:53 Triang3l: "terascale" is probably not a very nice name since AMD has almost never used it apparently (so far I've only been able to find a reference to it in an RDNA presentation)
12:54 Triang3l: but r600/vulkan just wouldn't make any sense because it's r800/vulkan
12:55 Triang3l: Asm and SFN things will be in amd/terascale/compiler probably similarly to amd/compiler for ACO
12:56 Triang3l: overall, by the way, I'd also like to separate SFN from everything TGSI-related, but maybe that's not that important, NVK still uses codegen that has some references to TGSI headers, even though it's outside Gallium
12:58 Triang3l: like TGSI semantics (in favor of NIR VARYING, SYSTEM_VALUE and if needed FRAG_RESULT)
12:59 Triang3l: and of course to separate r600_shader_from_nir into a function that calls lowerings/optimizations needed by SFN internals themselves and Gallium-specific lowerings (and the Gallium-specific r600_pipe_shader)
15:42 mlankhorst: Any idea why my wife's laptop fails to resume from suspend, then acts dead so completely that the GPU won't post (and laptop won't turn on) unless I leave the laptop on for 5-10 minutes, then reboot -- where all of this does not happen if I leave a video in firefox open as main screen? (guessing vdec related) I'm sorry if this sounds specific, was about the RMA that laptop. :)