#asahi-gpu · 2023-08-23 20:39
i509vcbI guess blame my code for uploading the uniforms?#asahi-gpu · 2023-08-23 20:39
i509vcbhttps://gist.github.com/i509VCB/1e8f18e69ba980cd37e641eed56a8900#asahi-gpu · 2023-08-23 20:39
i509vcbThe shader looks fine, I can see the load from the right uniform registers and the store into the buffer where the dEQP test compares the result#asahi-gpu · 2023-08-23 20:28
alyssaASAHI_MESA_DEBUG=trace AGX_MESA_DEBUG=shaders is my go-to#asahi-gpu · 2023-08-23 20:28
alyssaif you need to see NIR before/after every pass, NIR_DEBUG=print but it's very verbose#asahi-gpu · 2023-08-23 20:28
alyssaAGX_MESA_DEBUG=shaders#asahi-gpu · 2023-08-23 20:27
i509vcbsomething a bit simpler, dumping nir?#asahi-gpu · 2023-08-23 20:22
i509vcbI guess I'll have to guess where I went wrong since the dEQP says I have the wrong values#asahi-gpu · 2023-08-23 20:22
alyssaMakes the "game porting toolkit" announcement even more obnoxious when I now *know* that samplers are necessarily broken in there.#asahi-gpu · 2023-08-23 20:21
alyssaI am really peaved about this. What a stupid corner to cut.#asahi-gpu · 2023-08-23 20:21
alyssaThat means DX12 is impossible on this hardware#asahi-gpu · 2023-08-23 20:21
alyssaI just discovered that, due to a horrendously small sampler heap, it is impossible to implement standards conformant bindless samplers on M1#asahi-gpu · 2023-08-23 20:21
alyssa----------------#asahi-gpu · 2023-08-23 20:21
alyssacould hack decode.c#asahi-gpu · 2023-08-23 20:21
alyssanot really#asahi-gpu · 2023-08-23 20:16
i509vcbis there a way I can dump the state of uniform registers?#asahi-gpu · 2023-08-23 18:34
alyssagood ok i like this.#asahi-gpu · 2023-08-23 18:33
alyssathen the special cases disappear#asahi-gpu · 2023-08-23 18:33
alyssaalyssa you just need to add a load_texture_handle_agx intrinsic#asahi-gpu · 2023-08-23 18:33
alyssaAHAHAHAH#asahi-gpu · 2023-08-23 18:17
alyssagrep for agx_usc_uniform to see how that works#asahi-gpu · 2023-08-23 18:16
alyssaalong with each shader, the driver specifies a list of GPU addresses to be preloaded into uniform register ranges#asahi-gpu · 2023-08-23 18:09
alyssai509vcb: agx_usc_uniform#asahi-gpu · 2023-08-23 17:36
i509vcb(well I mean what buffer to read from for indirect)#asahi-gpu · 2023-08-23 17:35
i509vcbfor indirect, probably the same thing but I just tell where to load the values from?#asahi-gpu · 2023-08-23 17:33
i509vcb(how do I go from the parameters passed into vkCmdDispatch to putting those in the uniform registers I have reserved for that sysval)#asahi-gpu · 2023-08-23 17:32
i509vcbI believe I've done lowering for base_vertex and num_workgroups correctly. But I need to have these values be set before the main shader executes somehow? I'd imagine I need to do something when recording the command to load the uniform registers with the expected values?#asahi-gpu · 2023-08-23 09:29
_jannau__done#asahi-gpu · 2023-08-23 09:25
eric_engestromjannau: I think you can close https://github.com/KhronosGroup/VK-GL-CTS/pull/405, I think it might take a while for the khronos maintainer to get around to that#asahi-gpu · 2023-08-23 00:18
alyssaagx_nir_lower_sysvals aligns 64-bit to 64-bit boundaries so go with that#asahi-gpu · 2023-08-23 00:18
alyssa2. uhhhh not 100% sure off hand. I would expect alignment is needed#asahi-gpu · 2023-08-23 00:17
alyssa1. correct#asahi-gpu · 2023-08-22 23:57
i509vcbbut for a 64-bit value, it appears the two registers just need to be neighbors? (u4 and u3 for example?)#asahi-gpu · 2023-08-22 23:56
i509vcbJust to confirm, it not possible to do unaligned reads? i.e. read a 32-bit value from u1h and u2l#asahi-gpu · 2023-08-22 23:37
alyssa0 = mov u5#asahi-gpu · 2023-08-22 23:37
alyssameans#asahi-gpu · 2023-08-22 23:37
alyssa32 %0 = load_preamble (base=10)#asahi-gpu · 2023-08-22 23:36
alyssain 16-bit units#asahi-gpu · 2023-08-22 23:36
alyssai509vcb: uniform register index#asahi-gpu · 2023-08-22 23:24
i509vcbregister index of sorts?#asahi-gpu · 2023-08-22 23:19
i509vcbwhat does the "base" mean in load_preamble?#asahi-gpu · 2023-08-22 20:37
alyssasure#asahi-gpu · 2023-08-22 20:19
i509vcbConceptually I guess I can just consider the whole thing one pass even though I do a sort of inner pass to figure out what sysvals need to be lowered and another inner pass to actually lower#asahi-gpu · 2023-08-22 19:46
alyssa"the passes" probably you want a single pass for all the sysvals#asahi-gpu · 2023-08-22 19:45
alyssashrug#asahi-gpu · 2023-08-22 19:45
i509vcbSounds about right#asahi-gpu · 2023-08-22 19:45
alyssai mean#asahi-gpu · 2023-08-22 19:44
alyssawho cares#asahi-gpu · 2023-08-22 19:44
i509vcbWell relative to the rest of the passes#asahi-gpu · 2023-08-22 19:44
alyssaagx_pipeline, for now