#asahi-gpu · 2023-07-05 16:48
jannauthe failing cts run was slower on the ultra than a succesful one on m1#asahi-gpu · 2023-07-05 16:48
alyssaso..#asahi-gpu · 2023-07-05 16:47
alyssanominally the ultra but the CTS is completely single threaded CPU bound#asahi-gpu · 2023-07-05 16:47
alyssawho will win? =D#asahi-gpu · 2023-07-05 16:47
jannaum1 ultra x11_egl CTS is running as well#asahi-gpu · 2023-07-05 16:47
alyssaanyway, M1 CTS is running#asahi-gpu · 2023-07-05 16:47
alyssaeasier just to do the x11_egl run for now#asahi-gpu · 2023-07-05 16:47
alyssameh#asahi-gpu · 2023-07-05 16:47
jannauwe could do wayland and disable 10-bit formats#asahi-gpu · 2023-07-05 16:47
alyssaanyway, M1 CTS is running#asahi-gpu · 2023-07-05 16:46
alyssa(Strictly you could do a Wayland run with your patch and point to it and say "look it's your bug" but it would complicate things.)#asahi-gpu · 2023-07-05 16:46
jannauaye#asahi-gpu · 2023-07-05 16:45
alyssabut... should pass, probably, maybe? O:)#asahi-gpu · 2023-07-05 16:45
alyssaThat doesn't have your wide_color fix on the mesa side, and the 10-bit thing still hasn't landed upstream anyway, so unfortunately still an x11_egl run#asahi-gpu · 2023-07-05 16:43
alyssajannau: alyssa/mesa:agx/es31-v2#asahi-gpu · 2023-07-05 16:43
alyssakicking off CTS on M1 to break sure I didn't break anything in the meantime#asahi-gpu · 2023-07-05 16:39
jannauok#asahi-gpu · 2023-07-05 16:38
alyssajannau: Let me get you a non-stupid branch to use#asahi-gpu · 2023-07-05 16:37
jannaushall I start a full CTS run?#asahi-gpu · 2023-07-05 16:36
jannauthe lower core count versions have always cores in all clusters disabled#asahi-gpu · 2023-07-05 16:36
alyssathen https://rosenzweig.io/0001-asahi-agx-Set-coherency-bit-for-clustered-targets.patch should work everywhere#asahi-gpu · 2023-07-05 16:35
alyssak, yeah#asahi-gpu · 2023-07-05 16:34
jannaunum_clusters_total should be 2 / 4 / 8 for m1 pro / max / ultra#asahi-gpu · 2023-07-05 16:33
alyssadigs up her notes on cache bits#asahi-gpu · 2023-07-05 16:28
alyssathis.. adds to that suspicion >:)#asahi-gpu · 2023-07-05 16:28
alyssaI had a hunch already that bit 45 controls cache of some kind#asahi-gpu · 2023-07-05 16:27
alyssaspicy.#asahi-gpu · 2023-07-05 16:27
alyssaI mean, that's good but#asahi-gpu · 2023-07-05 16:27
alyssaAlright.#asahi-gpu · 2023-07-05 16:27
alyssaOof.#asahi-gpu · 2023-07-05 16:27
jannaualyssa: all atomic tests in es31 pass now on t6002#asahi-gpu · 2023-07-05 16:21
alyssaif that changes anything, let me know drm_asahi_params_global::num_clusters_total so I can do the proper runtime detection and not wedge t8103#asahi-gpu · 2023-07-05 16:21
alyssawith the atomics tests#asahi-gpu · 2023-07-05 16:21
alyssaGive that a try on t600x#asahi-gpu · 2023-07-05 16:21
alyssa_jannau__: https://rosenzweig.io/0001-HACK-Clustered.patch#asahi-gpu · 2023-07-05 16:17
alyssadid i wedge dcp too? powerful mesa patch#asahi-gpu · 2023-07-05 16:17
alyssaooh and green screen when shutting down#asahi-gpu · 2023-07-05 16:16
alyssaspicy#asahi-gpu · 2023-07-05 16:16
alyssa reason: AfFault,#asahi-gpu · 2023-07-05 16:16
alyssathat sounds like we're on the right track#asahi-gpu · 2023-07-05 16:16
alyssaI wrote a patch and it wedged my M1#asahi-gpu · 2023-07-05 16:12
alyssabit 45#asahi-gpu · 2023-07-05 16:12
alyssathat indeed sets 1 bit that isn't set on M1#asahi-gpu · 2023-07-05 16:11
alyssadottedmag: thanks#asahi-gpu · 2023-07-05 16:04
dottedmagm1max, if it matters#asahi-gpu · 2023-07-05 16:04
dottedmagalyssa: https://paste.debian.net/1285010/#asahi-gpu · 2023-07-05 15:54
alyssa(Or anyone else with an M1 Pro/Max/Ultra)#asahi-gpu · 2023-07-05 15:52
alyssa_jannau__: can you git clone github.com/dougallj/applegpu on macOS on t600x, run https://paste.debian.net/1285009 and send the output?#asahi-gpu · 2023-07-05 15:51
alyssa_jannau__: let's do process of elimination#asahi-gpu · 2023-07-05 15:11
_jannau__I think only atomics are blocking CTS on t600x, the heisenbug didn't reproduce there