#
✕ nick:alyssa
#asahi-gpu · 2023-02-18 19:46
alyssathat's enough hw for my brain rn
#asahi-gpu · 2023-02-18 19:46
alyssablah
#asahi-gpu · 2023-02-18 19:43
alyssapasstype of translucent punch through (?!?!?!)
#asahi-gpu · 2023-02-18 19:42
alyssaand still early-z testing is disabled in hardware?!
#asahi-gpu · 2023-02-18 19:42
alyssano zs_emit, but now we have at the beginning of the shader `gl_SampleMask = 1`
#asahi-gpu · 2023-02-18 19:42
alyssastill unk2->3
#asahi-gpu · 2023-02-18 19:42
alyssastill "spills"
#asahi-gpu · 2023-02-18 19:42
alyssawith [[early_fragment_tests]]
#asahi-gpu · 2023-02-18 19:41
alyssapunchtrough pass type + conservative depth any thanks to that depth write
#asahi-gpu · 2023-02-18 19:41
alyssatri merging is disabled
#asahi-gpu · 2023-02-18 19:41
alyssaearly-z is disabled
#asahi-gpu · 2023-02-18 19:40
alyssainterestingly, the frag shader gets a `gl_FragDepth = gl_Position.z` write after the store but before the st_tile
#asahi-gpu · 2023-02-18 19:39
alyssaunk 2: 2 -> 3
#asahi-gpu · 2023-02-18 19:39
alyssalina: ^^ that one is uapi
#asahi-gpu · 2023-02-18 19:39
alyssaso that's actually a "frag shader side effects" bit
#asahi-gpu · 2023-02-18 19:39
alyssa"set when frag shader spills" is set
#asahi-gpu · 2023-02-18 19:39
alyssaas for fragment shader side effects
#asahi-gpu · 2023-02-18 19:39
alyssa----
#asahi-gpu · 2023-02-18 18:52
alyssano still no early-z with 4 rts and MSAA
#asahi-gpu · 2023-02-18 18:52
alyssaoh but this was with MSAA 4x
#asahi-gpu · 2023-02-18 18:51
alyssaonly at 4 does early-z come off
#asahi-gpu · 2023-02-18 18:51
alyssaditto for 3 rts, with 0x444
#asahi-gpu · 2023-02-18 18:51
alyssabut with early-z/etc still enabled
#asahi-gpu · 2023-02-18 18:50
alyssawith 2 render tagets, 0x44
#asahi-gpu · 2023-02-18 18:49
alyssaoh, 4 render targets (rgba8) and blue has the first set of changes, just with unknown 0x4444
#asahi-gpu · 2023-02-18 18:48
alyssawhich makes even less sense
#asahi-gpu · 2023-02-18 18:48
alyssabut now the unknown is 0x2492
#asahi-gpu · 2023-02-18 18:48
alyssait's still no earlyz/unk2/unk3 and translcuent
#asahi-gpu · 2023-02-18 18:48
alyssauhh with 5 render targets (rgba8) and blue
#asahi-gpu · 2023-02-18 18:46
alyssaalso fragment_properties::unk3 from 0xf to 0x0 in this case
#asahi-gpu · 2023-02-18 18:41
alyssawith 8 render targets instead (and switched down to rgba8888) it's switched itself to translucnt and disabled early-z and unk2
#asahi-gpu · 2023-02-18 18:40
alyssaoh, huh
#asahi-gpu · 2023-02-18 18:36
alyssabut I wonder how that works with 8 rt?
#asahi-gpu · 2023-02-18 18:33
alyssawith blue on 4 render targets, 0x4444
#asahi-gpu · 2023-02-18 18:31
alyssawhere 0x1ffff means "everything is enabled"?
#asahi-gpu · 2023-02-18 18:31
alyssaso it's like a mask of what's enabled
#asahi-gpu · 2023-02-18 18:31
alyssaright, yes, blue mask is 0x4
#asahi-gpu · 2023-02-18 18:30
alyssaunless, hmm wait I did a green mask right
#asahi-gpu · 2023-02-18 18:30
alyssaIt seems like all the difference is from this oq2 unknown
#asahi-gpu · 2023-02-18 18:12
alyssanotably an... Opaque pass type? which seems broken?
#asahi-gpu · 2023-02-18 18:11
alyssaand still 0x2
#asahi-gpu · 2023-02-18 18:11
alyssawith no alpha to cov, we still have unk1 false and st_tile masked
#asahi-gpu · 2023-02-18 17:50
alyssaOcclusion query 2::Unknown from 0x1ffff -> 0x2
#asahi-gpu · 2023-02-18 17:49
alyssast_tile to be masked, of course
#asahi-gpu · 2023-02-18 17:49
alyssaShader::unk1 to be false
#asahi-gpu · 2023-02-18 17:49
alyssawith alpha to cov enabled, enabling writemask causes:
#asahi-gpu · 2023-02-18 17:49
alyssa---
#asahi-gpu · 2023-02-18 14:50
alyssa:+1:
#asahi-gpu · 2023-02-18 01:49
alyssa(I have a few other MRs up if you're feeling reviewy, but !21104 is the one that's causing rebase problems)
#asahi-gpu · 2023-02-18 01:47
alyssa(and then running that NIR through the backend)

plus ›