| --- Log opened Sun Feb 28 00:00:47 2021 |
| 00:29 | | catalyst [catalyst@Nightstar-ejd4sd.cable.virginm.net] has joined #code |
| 02:59 | <&ToxicFrog> | McMartin: weird. It works fine for me |
| 02:59 | <&ToxicFrog> | In both firefox and curl |
| 03:19 | | Degi_ [Degi@Nightstar-57biiq.pool.telefonica.de] has joined #code |
| 03:20 | | Degi [Degi@Nightstar-ddqiek.pool.telefonica.de] has quit [Operation timed out] |
| 03:20 | | Degi_ is now known as Degi |
| 03:35 | | Vornicus [Vorn@ServerAdministrator.Nightstar.Net] has quit [Connection closed] |
| 04:14 | | VirusJTG_ [VirusJTG@Nightstar-42s.jso.104.208.IP] has quit [Connection closed] |
| 04:15 | | VirusJTG_ [VirusJTG@Nightstar-42s.jso.104.208.IP] has joined #code |
| 04:32 | <&McMartin> | Man, Motorola chips |
| 04:32 | <&McMartin> | There's a whole class of illegal instructions that take different traps instead of the usual illegal instruction trap |
| 04:32 | <&McMartin> | And so these are used to produce extra instructions to be filled in by the OS. |
| 06:45 | <&[R]> | Instead of syscalls? |
| 06:58 | <&McMartin> | Well. Instead of the TRAP instruction. |
| 06:59 | <&McMartin> | But as doing this puts you in supervisor mode and bounces you through an interrupt vector, I think this means it still counts as a syscall. |
| 07:07 | | Vorntastic [uid293981@Nightstar-h2b233.irccloud.com] has joined #code |
| 07:07 | | mode/#code [+qo Vorntastic Vorntastic] by ChanServ |
| 07:08 | | himi [sjjf@Nightstar-v37cpe.internode.on.net] has quit [Ping timeout: 121 seconds] |
| 07:19 | <&McMartin> | Eesh. Kind of wild when a software developer obviously gives zero shits about windows distribution but the half-assed effort they make is still easier to get running on Windows than the Linux stuff |
| 07:22 | <~Vorntastic> | Man |
| 07:23 | <~Vorntastic> | I remember a year or three back i wanted to see if Ruby had gotten any better |
| 07:23 | <&McMartin> | To be fair, the system this guy clearly actually cares about is Amigas with PowerPC daughterboards |
| 07:23 | <~Vorntastic> | And so I went to go get it and, uh, it did not seem to have a Windows installer |
| 07:23 | | himi [sjjf@Nightstar-v37cpe.internode.on.net] has joined #code |
| 07:23 | | mode/#code [+o himi] by ChanServ |
| 07:23 | <~Vorntastic> | So the answer was pretty clearly "not enough to bother with in any case" |
| 07:59 | | Kindamoody[zZz] is now known as Kindamoody|afk |
| 08:34 | | abudhabi_ [abudhabi@Nightstar-afrfru.adsl.tpnet.pl] has joined #code |
| 08:35 | | abudhabi__ [abudhabi@Nightstar-r23qje.adsl.tpnet.pl] has quit [Operation timed out] |
| 08:37 | <&McMartin> | Pfft, I'd forgotten about that bit |
| 08:37 | <&McMartin> | "Building a RISC OS system [on a Raspberry Pi, as the Proper Inheritor of the Acorn Archimedes] and writing some software for it—or porting some software to it—sounds like a fun thing to do. I’ve named this “Project Mehitabel” after the name of Archie’s old companion, and because getting this all to work has been a bit like herding cats." |
| 08:39 | <&McMartin> | "Step 1: Does anything even work, anywhere, ever" |
| 08:41 | <&McMartin> | "We’ve got our two most important applications ready: a web browser and Minecraft^W^W^W^W^Wa text editor and Minesweeper" |
| 08:42 | <&McMartin> | It remains absolutely the greatest thing that when you shift into text mode the desktop starts scrolling off the top of the display, line by line, as you go. https://bumbershootsoft.files.wordpress.com/2017/06/riscos_02_shellcli.png |
| 09:22 | | Emmy [Emmy@Nightstar-l49opt.fixed.kpn.net] has joined #code |
| 09:28 | | celticminstrel [celticminst@Nightstar-8ih5mu.dsl.bell.ca] has quit [[NS] Quit: And lo! The computer falls into a deep sleep, to awake again some other day!] |
| 10:48 | | Vorntastic [uid293981@Nightstar-h2b233.irccloud.com] has quit [[NS] Quit: Connection closed for inactivity] |
| 10:51 | | Vorntastic [uid293981@Nightstar-h2b233.irccloud.com] has joined #code |
| 10:52 | | mode/#code [+qo Vorntastic Vorntastic] by ChanServ |
| 12:11 | | john_cephalopoda [john@Nightstar-qshh2j.dip0.t-ipconnect.de] has joined #code |
| 13:27 | <&[R]> | https://twitter.com/Sun_Ultra10/status/1365219185140391937 |
| 13:39 | | catalyst_ [catalyst@Nightstar-9naak6.dab.02.net] has joined #code |
| 13:42 | <@Tamber> | That's... impressive, for a variety of reasons, none of them good. |
| 13:43 | | catalyst [catalyst@Nightstar-ejd4sd.cable.virginm.net] has quit [Ping timeout: 121 seconds] |
| 14:54 | | Vornicus [Vorn@ServerAdministrator.Nightstar.Net] has joined #code |
| 14:54 | | mode/#code [+qo Vornicus Vornicus] by ChanServ |
| 15:23 | | celticminstrel [celticminst@Nightstar-8ih5mu.dsl.bell.ca] has joined #code |
| 15:23 | | mode/#code [+o celticminstrel] by ChanServ |
| 15:42 | < john_cephalopoda> | It's probably the cloud. https://xkcd.com/908/ |
| 15:49 | | catalyst [catalyst@Nightstar-hu950u.dab.02.net] has joined #code |
| 15:50 | | catalyst_ [catalyst@Nightstar-9naak6.dab.02.net] has quit [Ping timeout: 121 seconds] |
| 15:58 | | Vorntastic [uid293981@Nightstar-h2b233.irccloud.com] has quit [[NS] Quit: Connection closed for inactivity] |
| 16:42 | | Kindamoody|afk is now known as Kindamoody |
| 18:21 | | Netsplit Traal.Nightstar.Net <-> Krikkit.Nightstar.Net quits: @PinkFreud |
| 18:32 | | Netsplit over, joins: @PinkFreud |
| 19:55 | | catalyst_ [catalyst@Nightstar-ejd4sd.cable.virginm.net] has joined #code |
| 19:56 | | catalyst [catalyst@Nightstar-hu950u.dab.02.net] has quit [Ping timeout: 121 seconds] |
| 20:02 | | catalyst_ is now known as catalyst |
| 20:46 | | himi [sjjf@Nightstar-v37cpe.internode.on.net] has quit [Ping timeout: 121 seconds] |
| 20:47 | <&McMartin> | "Supexec(sub). XBIOS Call 38. Runs a subroutine in the processor's supervisor mode." |
| 20:47 | <&McMartin> | 1985 was a very different time |
| 20:54 | < catalyst> | what even is security |
| 20:54 | < catalyst> | probably a locked door |
| 20:55 | <&McMartin> | To be fair, the processor here is the MC68000, which does not meaningfully have memory protection and cannot be used to implement a Unixlike. |
| 20:55 | <&McMartin> | Not one that meets POSIX standards anyway. |
| 20:56 | <&McMartin> | (IIRC the 68020 is the one that finally gains the necessary MMU powers, but I could be off by a generation in either direction) |
| 20:56 | < Yossarian> | there are newer extended m68k and z80 processors and ISAs |
| 20:56 | <&McMartin> | Oh yeah. |
| 20:56 | < Yossarian> | but I'm not sure how useful they are, besides being apart of a car's EMU |
| 20:56 | <&McMartin> | AIUI, that's why I have to say the C in MC68000; that means The Chip In The First Macs and Amigas And The Sega Genesis/Mega Drive |
| 20:57 | <&McMartin> | And in, as is the case for the thing I was quoting there, the first Atari STs. |
| 20:57 | < Yossarian> | makes sense |
| 20:57 | | john_cephalopoda [john@Nightstar-qshh2j.dip0.t-ipconnect.de] has quit [[NS] Quit: Trees can see into your soul.] |
| 20:58 | < Yossarian> | a more competitive CPU choice for the end-consumer would be neat but yeah |
| 20:58 | <&McMartin> | But AIUI, the Atari ST line not only could run UNIX, but by the end of their relevance, the final versions of it were running 68030s and 68040s and were basically low-powered personal Unix systems, after the MiNT project got absorbed. |
| 20:58 | <&McMartin> | tbh I can't think of any use cases these days where an m68k-series chip would make sense where some ARM core wouldn't make more sense. |
| 20:58 | <&McMartin> | Legacy firmware I guess |
| 20:59 | <&McMartin> | The modern z80s are quite nice |
| 20:59 | < Yossarian> | I looked into it once and |
| 20:59 | < Yossarian> | they didn't seem competitive enough I suppose |
| 20:59 | | * McMartin nods |
| 21:00 | < Yossarian> | once of those "I could, but I don't see a reason why I should" |
| 21:00 | < Yossarian> | s/once/one |
| 21:00 | | * abudhabi_ remembers university assignments to code for the MC6800. Annoying! |
| 21:01 | < Yossarian> | although the modern Z80s are supposed to be able to address memory quite easily meaning wiring something to it to make it work should be easy, but I'm not sure if that applies to the newer memory modes |
| 21:02 | <&McMartin> | Yeah, the obvious use case for the modern Z80s is "we've been using, like, 1970s-era Z80As as an embedded microntroller for 40 years now and we'd like to keep making them" |
| 21:02 | <&McMartin> | "Oh they have a couple more pins now that's nice we'll just file those off" |
| 21:02 | < Yossarian> | like I said I think they're used in some car or engine control units but of which, I couldn't tell ya |
| 21:02 | <&McMartin> | That'd make sense |
| 21:03 | <&McMartin> | The only things I know of that used them were a (very long era of!) arcade boards and some disgustingly overpriced graphing calculators |
| 21:03 | <&McMartin> | And for the latter, a Z80 is Enough and an ARM would do the More Better. |
| 21:04 | <&McMartin> | re: the arcade boards though, I always kind of wondered why the Genesis had two wildly different coprocessors (an early 68k and a Z80) |
| 21:04 | < Yossarian> | oh yes, TI's Disgustingly Overpriced Graphing Calculators |
| 21:04 | <&McMartin> | That was matching Good Arcade Boards at the time, which had the same setup |
| 21:04 | <&McMartin> | ... and which lasted through well past the end of the Genesis, because, like, IIRC Street Fighter Alpha 3 was still using it in the late 1990s |
| 21:04 | < Yossarian> | I think my 84+ silver edition crapped out |
| 21:04 | < Yossarian> | which is annoying |
| 21:05 | <&McMartin> | That split-CPU design also makes the Saturn's disastrous architecture more comprehensible |
| 21:06 | <&McMartin> | Because "what the fuck are you doing requiring people to use parallel programming techniques that nobody had mastered then and arguably still haven't in 2021 in order to get proper performance out" |
| 21:06 | <&McMartin> | has the easy answer "we took the Genesis and asked 'but what if the coprocessor didn't suck'" |
| 21:06 | <&McMartin> | I have no explanation for the PS3. -_- |
| 21:07 | < Yossarian> | game library sucked tho |
| 21:07 | <&McMartin> | Yeah, because nobody could get it to run much better than half powre. |
| 21:08 | < Yossarian> | I think it had Phantasy Star: Offline |
| 21:08 | < Yossarian> | because the equipment to play it online would never be released, at least in NA |
| 21:09 | < Yossarian> | it started the trend of these JMMORPGs that get launched and are only available to play for a year or two until the servers go bye-bye |
| 21:09 | < Yossarian> | like Final Fantasy XVIVIVIVIVIIII or whatever |
| 21:10 | < Yossarian> | At least the South Korean MMOs keep the lights on nigh forever |
| 21:10 | < Yossarian> | but their release target is PC usually, so, that shouldn't surprise anyone |
| 21:15 | <&McMartin> | FFXIV is going very strong indeed, and even XI continues to run on cruise control |
| 21:15 | <&McMartin> | If I wanted to grouse about flash in the pan MMORPGs, I'd look to NCSoft's endless spinoffs |
| 21:27 | < catalyst> | amusingly FFXI and FFXIV are both going strong |
| 21:28 | < catalyst> | ya |
| 21:28 | < catalyst> | what McMartin said |
| 21:28 | < catalyst> | god I'm tired |
| 21:29 | < catalyst> | <&McMartin> I have no explanation for the PS3. <-- so far as I can tell, hubris given how well the previous generation went |
| 21:30 | < catalyst> | technically the PS3 was very capable, it uh, just required committing to a style of programming and development that didn't match how most studios work |
| 21:31 | <&McMartin> | Yeah, I come at this from a weird place |
| 21:31 | <&McMartin> | But I knew, very clearly, from 2000-2008, that manually specified multicore programming was *very hard*, *not broadly spread through the industry* and also *basically impossible to automate* |
| 21:32 | <&McMartin> | That latter is my Weird Place; I was a grad student working in compilers and program analysis research and I was seeing first hand how far you didn't get |
| 21:32 | <&McMartin> | This also meant that digging into the Genesis' distribution of labor was a huge revelation for me in like 2017~ |
| 21:32 | < catalyst> | ah right |
| 21:33 | <&McMartin> | And here in 2017, an entire decade later |
| 21:33 | <&McMartin> | We've basically made it to "we kind of have a vaguely decent idea of what kinds of problems are good matches for compute shaders; also 'compute shaders' are things that everyone knows exist" |
| 21:34 | <&McMartin> | Well, here in 2021, but we'd been there for some time by then |
| 21:34 | < catalyst> | from my pov, working on PS3 related parts of a cross platform game, there's plenty of things we could use those cores for it just fucked up the architecture of the engine enough that it wasn't really worth doing |
| 21:34 | <&McMartin> | How was its memory coherence? |
| 21:34 | <&McMartin> | Like, I get "sure, lots of asynchronous noncommunicating tasks is p. OK" but |
| 21:35 | < catalyst> | I'm afraid I wasn't involved enough to be able to think about that part but getting data on and off wasn't the main problem? |
| 21:35 | < catalyst> | there were abstractions |
| 21:35 | <&McMartin> | I was thinking more "different cores getting into fights over the same memory" |
| 21:36 | < catalyst> | oh, that didn't come up at all so far as I could tell |
| 21:36 | <&McMartin> | like, knowing nothing about it I could imagine splitting a bunch of agent AIs over all the cores, but then they all need to know about the terrain and then Woe And Mutually Exclusive Spider Town |
| 21:36 | < catalyst> | I mean, you just break down the problem small enough and set up immutable data structures? |
| 21:37 | <&McMartin> | I think the keyword I've been failing to remember is "NUMA" - Non-Uniform Memory Access |
| 22:40 | <&ToxicFrog> | IIRC the peripheral processors can't directly access system memory (although they can explicitly request DMA to/from main memory, or between the memory directly attached to other SPE) |
| 22:40 | <&ToxicFrog> | Also I keep wanting to call the coprocessors PPUs, for Peripheral Processing Units (the term used on some CDC mainframes), when the Cell does use that term, but to mean Power Processing Unit, the main core |
| 22:42 | <&ToxicFrog> | So the intended use is either to connect them in a ring, with each SPE reading data from the one before it in the ring and streaming it to the one after it (MIMD), or do a sort of scatter/gather thing where each one DMAs down a different slice of the data being worked on and DMAs the results back up (SIMD) |
| 22:43 | < catalyst> | the correct term is SPU because that's what programming them makes you want to do |
| 22:44 | <&ToxicFrog> | I remember that there's some fine distinction between SPE and SPU but I no longer remember what it is |
| 22:44 | < catalyst> | I don't care to look it up |
| 22:44 | < catalyst> | x) |
| 22:44 | < catalyst> | fortunately we can forget it every existed |
| 22:57 | | Emmy [Emmy@Nightstar-l49opt.fixed.kpn.net] has quit [Ping timeout: 121 seconds] |
| 23:15 | | himi [sjjf@Nightstar-1drtbs.anu.edu.au] has joined #code |
| 23:15 | | mode/#code [+o himi] by ChanServ |
| 23:25 | | Kindamoody is now known as Kindamoody[zZz] |
| 23:27 | | McMartin [mcmartin@Nightstar-i80eaa.ca.comcast.net] has quit [[NS] Quit: Probably not fleeing in terror] |
| 23:34 | | McMartin [mcmartin@Nightstar-i80eaa.ca.comcast.net] has joined #code |
| 23:34 | | mode/#code [+ao McMartin McMartin] by ChanServ |
| --- Log closed Mon Mar 01 00:00:48 2021 |