#
✕ nick:i509VCB
#asahi-gpu · 2023-10-17 19:43
i509vcbFor CmdDrawIndirect, I'm not sure where I'm supposed to tell agx about the stride?
#asahi-gpu · 2023-10-17 17:37
i509vcbella-0: when I last had out of host memory it was because I forgot to enable swap when installing lol
#asahi-gpu · 2023-10-17 15:46
i509vcbNow to get back to image stuff in the driver so txs works
#asahi-gpu · 2023-10-17 15:46
i509vcbJust watched the XDC talk, pretty cool stuff
#asahi-gpu · 2023-10-13 15:07
i509vcb(Or I'll figure out in a few hours)
#asahi-gpu · 2023-10-13 15:06
i509vcbMy arch install imploded so I moved :)
#asahi-gpu · 2023-10-13 15:06
i509vcbI assume the fedora spin has a new enough abi for agxv?
#asahi-gpu · 2023-10-11 05:44
i509vcbit looks like a plain ol shader on my amd system
#asahi-gpu · 2023-10-11 05:41
i509vcbProbably a driver bug
#asahi-gpu · 2023-10-10 00:01
i509vcbalyssa: probably of interest for us with agxv: https://gitlab.freedesktop.org/mesa/mesa/-/issues/9920
#asahi-gpu · 2023-10-09 01:31
i509vcbVK_EXT_external_memory_host on agx? yay or nay?
#asahi-gpu · 2023-10-07 01:46
i509vcb(Or more likely via -Z codegen-backend like embark is doing)
#asahi-gpu · 2023-10-07 01:42
i509vcb(specifically referring to this: https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25498)
#asahi-gpu · 2023-10-07 01:41
i509vcbTake the CLC magic, have rust output LLVM IR and pipe that into the LLVM to CLC thing to run Rust on agx itself
#asahi-gpu · 2023-10-07 01:40
i509vcbNow I do feel Rust could have a very cool use idea
#asahi-gpu · 2023-10-07 01:40
i509vcbI mean you *don't* have to use an allocator but that would likely be rejected from Mesa or need a really good justification
#asahi-gpu · 2023-10-07 01:39
i509vcbbut it is ugly
#asahi-gpu · 2023-10-07 01:39
i509vcbI do recall there was an example somewhere of using Box with some from_raw hackery to effectively have an allocator
#asahi-gpu · 2023-10-04 20:15
i509vcbLargely a question on kmsro devices in Vulkan
#asahi-gpu · 2023-10-04 20:14
i509vcbalyssa: https://gitlab.freedesktop.org/mesa/mesa/-/issues/9920 (when you return)
#asahi-gpu · 2023-10-03 01:16
i509vcb🦀
#asahi-gpu · 2023-09-30 22:52
i509vcb(although needing to build tokio for mesa just sounds like an npm nightmare)
#asahi-gpu · 2023-09-30 22:51
i509vcbAlthough the kernel figured it out so *shrug*
#asahi-gpu · 2023-09-30 22:50
i509vcbWell for writing proc macros
#asahi-gpu · 2023-09-30 22:24
i509vcbAlthough if someone wanted to use proc-macro in mesa then it would be pain
#asahi-gpu · 2023-09-30 21:11
i509vcbalyssa probably has an opinion how to do this
#asahi-gpu · 2023-09-30 05:37
i509vcbhmm now to figure out why the palm trees are uhh weird https://usercontent.irccloud-cdn.com/file/TGYv01le/Screenshot_20230930_003310.png
#asahi-gpu · 2023-09-30 02:44
i509vcbI'm tempted to throw a hack commit to just make it work and revisit in the future
#asahi-gpu · 2023-09-30 02:34
i509vcbSpeaking of which the vk_meta discussion for layer_id seemed inconclusive from what I recall?
#asahi-gpu · 2023-09-30 02:32
i509vcbMainly ash
#asahi-gpu · 2023-09-30 02:32
i509vcbOn Rust until meson supports crates from cargo we'd probably need to vendor some rust stuff...
#asahi-gpu · 2023-09-30 02:18
i509vcbSelf contained things like a vk_meta implementation could be a cool idea but you'd probably end up writing hundreds of lines of unsafe Rust with ash which wouldn't be much better than C
#asahi-gpu · 2023-09-30 02:13
i509vcbNot sure how VkEvent would be implemented. Assuming there isn't special stuff in agxv I'd guess a buffer you write to after some stage finishes?
#asahi-gpu · 2023-09-27 18:18
i509vcbAlthough what would we test write in Rust for agxv is the big question
#asahi-gpu · 2023-09-27 17:03
i509vcb(And probably quite a lot of work given how much stuff that would need to be bound...)
#asahi-gpu · 2023-09-27 16:59
i509vcbAlthough that's still unclear
#asahi-gpu · 2023-09-27 16:59
i509vcbif agxv is being rewritten in rust I'm definitely intrigued
#asahi-gpu · 2023-09-20 15:40
i509vcbFixing vk_meta there sounds reasonable, I imagine a few nvk specific details will be there anyways
#asahi-gpu · 2023-09-20 14:23
i509vcbI guess I could do tree wide changes but then I have to figure out what other drivers expect...
#asahi-gpu · 2023-09-20 14:22
i509vcbone thing I did notice is that layer has a gl_varying_slot, wouldn't that imply it's already a varying or is the common infrastructure not capable of expressing layer as a sysval currently
#asahi-gpu · 2023-09-20 01:01
i509vcbSomething about LAYER not having a sysval value feels weird
#asahi-gpu · 2023-09-20 00:01
i509vcbI don't see any layer id enum variant in gl_system_value
#asahi-gpu · 2023-09-19 23:55
i509vcbOr I guess I am adding new things
#asahi-gpu · 2023-09-19 23:54
i509vcbsysvals_to_varyings has no layer option?
#asahi-gpu · 2023-09-19 23:42
i509vcbThe meta shader uses an actual raw load_layer_id
#asahi-gpu · 2023-09-19 23:42
i509vcbAlthough I am definitely still confused as to why agx_nir_predicate_layer seems to never lower load_layer_id in the blit meta shader
#asahi-gpu · 2023-09-19 23:37
i509vcbSure I guess that makes sense
#asahi-gpu · 2023-09-19 20:38
i509vcbthis is going several layers deep in vulkan CreateInfo types...
#asahi-gpu · 2023-09-19 20:22
i509vcbdescriptor set binding of VK_DESCRIPTOR_TYPE_SAMPLED_IMAGE with descriptorCount > 1?
#asahi-gpu · 2023-09-19 20:09
i509vcbCmdBeginRendering will know the number of layers because of binding the framebuffer

plus ›