| - S0rin (QUIT: Ping timeout: 240 seconds) (~S0rin@user/s0rin) | 00:26 | |
| + S0rin (~S0rin@user/s0rin) | 00:30 | |
| minute | Chain-Q: if it's not too much of an ask, and you could bring an a4000 there, we could run a test on it... but no stress | 00:30 |
|---|---|---|
| - S0rin (QUIT: Ping timeout: 248 seconds) (~S0rin@user/s0rin) | 01:14 | |
| + S0rin (~S0rin@user/s0rin) | 01:18 | |
| Chain-Q | minute: *thumbsup*. I don't promise 100% I'll manage that, but I'll try. I'll drop you a note on Saturday when I leave home if I could take the 4K with me. | 05:43 |
| + Nowee (~NoweeEewo@lfbn-ncy-1-4-192.w83-194.abo.wanadoo.fr) | 09:20 | |
| - xet7 (QUIT: Ping timeout: 240 seconds) (~xet7@user/xet7) | 10:46 | |
| - Nowee (QUIT: Read error: Connection reset by peer) (~NoweeEewo@lfbn-ncy-1-4-192.w83-194.abo.wanadoo.fr) | 12:14 | |
| + Nowee (~NoweeEewo@lfbn-ncy-1-4-192.w83-194.abo.wanadoo.fr) | 12:14 | |
| minute | Chain-Q: very nice, thanks | 13:35 |
| Chain-Q | minute: turns out, another user will be there too (99%) with another 4000 + CSMKII :) so it'll surely work out | 13:55 |
| minute | awesome | 13:55 |
| minute | ok so i can reproduce the jitter now | 14:17 |
| minute | by changing the interrupt pulse length. if it is too short, amiga will miss some interrupts | 14:17 |
| minute | i have a register now for setting the exact pulse length | 14:17 |
| minute | forcie: could you try this firmware? http://dump.mntmn.com/BOOT.bin-z3-20220428 | 14:36 |
| minute | this has a much longer irq pulse | 14:36 |
| minute | also the pulse is now configurable with register 0x1024 | 14:36 |
| minute | rtg is still mega slow for me (no idea why) | 14:37 |
| minute | i can watch individual lines being drawn in blanker | 14:38 |
| minute | oh wow it's because of 68030 cache stuff >:0 | 14:41 |
| minute | disabling 68030 datacache makes it fast again | 14:42 |
| _Bnu | Ah, aren't you supposed to keep pulling the interrupt until it's serviced? | 14:47 |
| forcie | minute: sure, i'll be home late today but i am able to try it out around 21:00 berlin time | 14:47 |
| _Bnu | At least that was the behavior I observed with the uhhh, what's it called. | 14:48 |
| _Bnu | X-Surf or whatever. | 14:48 |
| _Bnu | It would keep tugging at IRQ6 until the driver told it to sit the hell down. | 14:49 |
| forcie | ah, the X-Surf. thats a selection of ISA NIC:s and special Jens:y Mach logic | 14:52 |
| Nowee | mh. semaphores ain't an option? | 14:53 |
| _Bnu | I don't know what semaphore is, so I can't answer that... | 14:53 |
| _Bnu | Sempahore Goes to Hollywood [DivX].avi | 14:54 |
| Nowee | well, instead of "tugging" (polling), you set a flag somewhere. "pending" "delivered" | 14:54 |
| minute | _Bnu: hmm, in my experience if i held the irq too long the amiga would just freeze, so idk | 14:55 |
| minute | how you described it was my first approach back in the day (during ethernet stuff) | 14:55 |
| Nowee | Semaphores where coined by Dijkstra in 1965. I'd suggest reading a lil about it | 14:57 |
| Nowee | heh | 14:57 |
| Nowee | have a nice day ;) | 14:58 |
| - Nowee (PART: !!unknown attribute: msg!!) (~NoweeEewo@lfbn-ncy-1-4-192.w83-194.abo.wanadoo.fr) | 14:58 | |
| _Bnu | I have no idea how you'd set a semaphore with an IPL pin, so uhh. | 14:58 |
| _Bnu | I suppose it might freeze if it somehow stays up after you get out of the IRQ server code and it isn't serviced? | 14:59 |
| minute | idk! | 15:02 |
| _Bnu | Why not... ask the scientists to research it. | 15:19 |
| minute | you must construct additional pylons | 15:20 |
| Chain-Q | yeah, the "interrupt pulse" idea is defnitely not good, TBH... it should stay up until it's serviced and ACK'd in some register on the card. but I noticed a lot of weirdness regarding this, I mean the ethernet driver had trouble servicing frames, and there was different behavior on int2 and int6... but I'm kinda sure the "interrupt pulse" approach that fixed the "Amiga freezing" is just hiding some even | 15:52 |
| Chain-Q | nastier bug, maybe | 15:52 |
| Chain-Q | at least it sounds like it | 15:52 |
| minute | well, i can try again to hold int6 for longer, just as i remember it it kinda pauses the system... | 16:37 |
| _Bnu | Weird... the interrupt should get served very quickly. | 16:58 |
| _Bnu | Are you starting to hold it before setting the flag that there's an interrupt from the card, so the interrupt server thinks there's nothing? | 16:59 |
| _Bnu | (AmigaOS doesn't like that very much.) | 16:59 |
| minute | maybe in the past i did! | 17:34 |
| minute | i'm currently bulding a firmware which can do the hold-until-cleared behavior | 17:35 |
| minute | lets see how it does | 17:35 |
| + xet7 (~xet7@user/xet7) | 17:53 | |
| minute | ok that works, at least for audio | 17:54 |
| - xet7 (QUIT: Ping timeout: 240 seconds) (~xet7@user/xet7) | 18:18 | |
| + xet7 (~xet7@user/xet7) | 18:31 | |
| - xet7 (QUIT: Remote host closed the connection) (~xet7@user/xet7) | 18:32 | |
| + xet7 (~xet7@user/xet7) | 18:32 | |
| _Bnu | Smol audio. | 19:04 |
| _Bnu | Big audo. | 19:04 |
| _Bnu | It's audo when big, because the i doesn't fit. | 19:04 |
| minute | :3 | 19:05 |
| minute | ok, now the card can't autoconfig anymore arrrghh wtf vivado | 19:11 |
| - xet7 (QUIT: Quit: Leaving) (~xet7@user/xet7) | 19:29 | |
| apolkosnik[m] | They secretly replaced "viva" with "f*ck", so it's like a tornado | 19:32 |
| + xet7 (~xet7@user/xet7) | 19:32 | |
| apolkosnik[m] | *"viv" | 19:33 |
| minute | tw reports in a1k that the longer irq time fixed the skips for him! | 20:30 |
| - xet7 (QUIT: Remote host closed the connection) (~xet7@user/xet7) | 20:55 | |
| forcie | testing, just making sure i drop the micro-sd card inside the closed case at least once first | 20:59 |
| forcie | minute: it is practically working now, with almost no skips! | 21:05 |
| forcie | although i can notice some skips still, like in the sweep waves | 21:05 |
| minute | forcie: thank you!! tomorrow i'll have another version then. | 21:06 |
| forcie | trying axtest now | 21:07 |
| forcie | some misses still but looks better | 21:08 |
| forcie | https://pastebin.com/fGahfqeE | 21:09 |
| forcie | nevermind the 20220427 in the title, supposed to be 0428 | 21:09 |
| minute | much better yes | 21:24 |
| minute | do you have mon? and do you know the card address (showconfig shows it) | 21:24 |
| minute | you could try `pokew 40001024 f0` (if 40000000 is the card address, otherwise adjust) | 21:25 |
| forcie | looks mapped at $40000000 | 21:33 |
| forcie | poked it, what is supposed to change? | 21:33 |
| minute | forcie: any less skips? or still the same? | 21:46 |
| forcie | still a rythmic skipping in sweep24.wav, but only noticable in the higher frequencies | 21:50 |
| forcie | but i couldnt hear any change after f0 was written | 21:51 |
| forcie | also, I take it that the offset $1024 is write-only? | 21:52 |
| minute | ok, thanks! | 21:52 |
| minute | write only, yes | 21:52 |
| forcie | alright, explains why viewing it showed no change | 21:52 |
| minute | yeah | 21:53 |
| apolkosnik[m] | or maybe snekado | 22:38 |
| Chain-Q | minute: btw, one thing I would be very interested to hack on is either mpega.library or some sort of similar API hackery for the ZZ9000AX. i think there was also mpeg.device and maybe a few others. not sure if you had plans already or anyone signed up for that, I don't want to tip on anyone's toes. | 22:51 |
| Chain-Q | I mean there's even some GPLv2 code for Delfina mpeg.device on Aminet. | 22:53 |
| Chain-Q | so you just need some chunk of code on the ZZ9000 side that can decode mpeg frames, and wire the two sides together (I know it's more complicated in practice, but still) | 22:54 |
| apolkosnik[m] | Is that in the fw posted yesterday? | 22:56 |
| minute | Chain-Q: mpeg as in video? we do have an mp3 decoder in zz an a simple player for the amiga side, but no api level integration yeah | 23:19 |
| Chain-Q | mpeg as in mpeg audio :) the delfina definitely can't decode video. but theoretically mpeg video is also an option | 23:21 |
| Chain-Q | https://aminet.net/package/mus/play/dmdev <- stuff like this | 23:21 |
| Chain-Q | for example | 23:21 |
| Chain-Q | but as I said, I could hack on this, if no one felt like it yet | 23:21 |
| Chain-Q | I think the mpeg.device was originally conceived for the CD32 MPEG module by C=, but this is 50% fact and 50% assumption from my part, so please someone correct me if I'm way off | 23:23 |
| minute | there's also MHI, have you looked at that? | 23:23 |
| Chain-Q | I'm aware of it :) but no, not in detail | 23:23 |
| minute | same | 23:24 |
| minute | anyway, nobody's working on this yet! | 23:24 |
| Chain-Q | oki. lets chat about maybe on saturday. :) shortly. | 23:25 |
| minute | sure | 23:25 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!