07:01 karolherbst: glehmann: so.. I found a correctness issue on asahi with lower_fminmax_signed_zero. When the numbers are denorms, the feq returns true, but the integer max/min could screw things up and I can fix this with this: https://gitlab.freedesktop.org/karolherbst/mesa/-/commit/37d68f953d04d63bc396a52a956101f1866ee460 any better ideas?
07:01 karolherbst: ehh
07:01 karolherbst: it returns false
07:02 karolherbst: ehh no, true
07:02 karolherbst: Seeing issues like this in the CL CTS: ERROR: fmax: -8386560.000000 ulp error at {-0x1.ffcp-139 (0x800007ff), -0x1.fffffcp-127 (0x807fffff)}: *-0x1.ffcp-139 vs. -0x1.fffffcp-127 (0x807fffff) at index: 3763
07:20 karolherbst: probably better to fcanonicalize the result actually
07:55 glehmann: yeah I think either solution makes sense
07:55 glehmann: karolherbst: but the whole lowering is garbage anyway: https://gitlab.freedesktop.org/mesa/mesa/-/work_items/14937
08:12 karolherbst: glehmann: okay... well at least with my fix the CL CTS is happy enough about this
15:56 alyssa: pendingchaos: I was expecting my A-b to be added :P
19:46 robclark: hmm, seems like we are missing a fourcc modifier for dx/vk standard tiling?
20:28 jenatali: Does anybody actually support that?
20:28 jenatali: From the DX POV it's basically dead
20:50 robclark: hmm, ok... I think dx defined two things, one that only requires 64k block size (which vk also has), and one that defined the layout within a block?
20:50 jenatali: Correct
20:50 robclark: I guess if no one supports the latter there isn't much need for fourcc modifier
20:51 jenatali: Yeah. We had high hopes, but even the vendors that did support it ended up dropping support because of perf
20:51 jenatali: WARP still supports it...
23:08 alyssa: Would've been nice but alas