#asahi · 2025-09-23 08:48
mazauscyber_: well, of course it is DNP -- it's only a PCB trace.#asahi · 2025-09-22 13:49
mazjannau: yeah, it feels rather expensive indeed. #asahi · 2025-09-22 13:46
mazfl0_id: unfortunately, I don't have any plan for a further run. I'd recommend you find other people interested in the EU and share the upfront costs.#asahi · 2025-09-22 13:41
mazfl0_id: ah, I didn't regenerate the files, only used what I had in the git tree for the v3.2 version (the last I've actually built)#asahi · 2025-09-22 13:39
mazfl0_id: in my case, it was only the connectors for which they now apply a surcharge. but it resolved the references with any apparent problem.#asahi · 2025-09-22 13:13
mazauscyber_: I just checked, and JLCPCB produces 5 populated boards for $66. with tax and shipping to UK, that's $106. still need to Picos to that. YMMV.#asahi · 2025-09-22 12:54
mazmake sure you check small details like component orientation -- not all PCB shops interpret that the same way, leading to funny things.#asahi · 2025-09-22 12:52
mazI've never used them, but if you are familiar with their process, go for it.#asahi · 2025-09-22 12:51
mazordering locally tends to be more expensive (at least for me), but taxes and shipping can sway things in a different direction.#asahi · 2025-09-22 12:50
mazauscyber_: it really depends *where* you're ordering it. you can get 5 PCBs from JCLPCB, fully populated, for about $50. depending where you are, shipping and taxes is what will cost you.#asahi · 2025-09-22 10:36
mazthe README file loads perfectly on my end.#asahi · 2025-09-22 10:36
mazyup.#asahi · 2025-09-22 10:35
mazthe original pico, not the wireless version, not the pico-2.#asahi · 2025-09-22 10:34
mazwell, even I can do it, so it is rather simple. just tedious.#asahi · 2025-09-22 10:33
mazauscyber: that's the correct thing to do. although you'll still have to assemble the Pico on top (but maybe you can convince the PCB manufacturer to do that as well?)#asahi · 2025-03-31 11:11
maz(it's Monday).#asahi · 2025-03-31 11:11
mazdon't*#asahi · 2025-03-31 11:11
mazfon't*#asahi · 2025-03-31 11:11
mazI love my xserve g5, but I think this is relevant here (or at all).#asahi · 2025-03-16 11:18
mazwhich of course is deadlock-prone.#asahi · 2025-03-16 11:18
mazmore or less. that's helped a bit if your interconnect knows how to delay snoops the right way.#asahi · 2025-03-16 11:17
maz(ffs, I'm even worse than usual today)#asahi · 2025-03-16 11:17
mazducks*#asahi · 2025-03-16 11:16
maz"but wait, transactional memory is awesome!" (me sucks)#asahi · 2025-03-16 11:14
mazsame*#asahi · 2025-03-16 11:14
mazISTR P8 having bizarre constraints on the SMT usage (save MM context mandatory for threads on the same core)#asahi · 2025-03-16 11:13
mazP10 definitely has SMT8.#asahi · 2025-03-10 18:18
mazvacuumed the underneath of the box, and it is now stone cold.#asahi · 2025-03-10 18:16
mazturns out that the air intake at the bottom had become clogged with dust after 3 years of hard labour.#asahi · 2025-03-10 18:15
maznot totally surprising for a build machine, but still warmer than it used to.#asahi · 2025-03-10 18:15
mazno idea whether this is a well known thing, but I recently noticed that my M1 studio was getting rather warm.#asahi · 2025-03-09 12:42
mazthe amusing thing is that at this stage, the biggest waste of cycle is the context-switch...#asahi · 2025-03-09 12:39
mazdepends on the model. in oversubscribed scenarios, you can get charged by the actual runtime.#asahi · 2025-03-09 12:38
mazchaos_princess: that's @cloud computing@ for you! :)#asahi · 2025-02-13 18:52
mazdftt#asahi · 2025-02-13 17:44
mazin my utmost generosity, I'll buy the now useless HW from you at a symbolic $/£/€...#asahi · 2025-02-13 17:39
mazI think clogification would have a nice ring to it.#asahi · 2025-02-13 16:38
mazyawn#asahi · 2025-01-17 10:46
mazpitust: the timers are working correctly. the vgic implementation is a nasty piece of shit.#asahi · 2024-11-26 11:41
maz"thermal shutdown"#asahi · 2024-11-26 11:35
mazdon't worry, we'll be back to the normal rainy grey very shortly!#asahi · 2024-11-26 11:25
mazj`ey: it's winter. you should push the frame rate higher! :)#asahi · 2024-11-20 08:05
mazare you up against something like that?#asahi · 2024-11-20 08:04
mazApple has some really funky extensions where they seed the algorithm on a per-EL basis, which we of course don't support. It appears that by some magic (probably thanks to what m1n1 does), this feature is disabled.#asahi · 2024-11-20 08:01
maznickchan: that's not explicitly disallowed in the spec. but if an implementation does, they get to keep the pieces and I'll happily ignore PAuth on their HW.#asahi · 2024-11-11 19:04
mazshit happens. it's just annoying that it is part of a release.#asahi · 2024-11-11 17:40
mazgiven that EFI pretty much mandates a flat VA:PA mapping.#asahi · 2024-11-11 17:39
mazand frankly, the whole page-size only matters if you are DMA-ing to the allocated memory.#asahi · 2024-11-11 17:38
mazthese guys should go out more and realise that not everything is an x86.