| - S0rin (QUIT: Ping timeout: 244 seconds) (~S0rin@user/s0rin) | 00:36 | |
| + S0rin (~S0rin@user/s0rin) | 00:45 | |
| - S0rin (QUIT: Ping timeout: 260 seconds) (~S0rin@user/s0rin) | 01:19 | |
| + S0rin (~S0rin@user/s0rin) | 01:22 | |
| - RobDangerous (QUIT: Ping timeout: 252 seconds) (~Thunderbi@p200300ec8f03e000807883f7a61c70ca.dip0.t-ipconnect.de) | 01:23 | |
| - shanshe_ (QUIT: Read error: Connection reset by peer) (uid398662@id-398662.lymington.irccloud.com) | 04:22 | |
| + shanshe_ (uid398662@lymington.irccloud.com) | 04:24 | |
| - xet7 (QUIT: Remote host closed the connection) (~xet7@user/xet7) | 05:02 | |
| + xet7 (~xet7@user/xet7) | 05:03 | |
| + demolition (~demolitio@xd93dd958.cust.hiper.dk) | 06:52 | |
| - tryphoncosinus (QUIT: Ping timeout: 265 seconds) (~lug@modemcable134.206-37-24.static.videotron.ca) | 07:03 | |
| - demolition (QUIT: Read error: No route to host) (~demolitio@xd93dd958.cust.hiper.dk) | 07:05 | |
| - S0rin (QUIT: Ping timeout: 244 seconds) (~S0rin@user/s0rin) | 08:08 | |
| + mnemotron (~wayne@194.135.47.239) | 08:08 | |
| - mnemotron (QUIT: Client Quit) (~wayne@194.135.47.239) | 08:08 | |
| + S0rin (~S0rin@user/s0rin) | 08:08 | |
| + mnemotron (~wayne@194.135.47.239) | 08:09 | |
| + RobDangerous (~Thunderbi@p200300ec8f03e00015c7397c97387cc6.dip0.t-ipconnect.de) | 09:52 | |
| + tryphoncosinus (~lug@modemcable134.206-37-24.static.videotron.ca) | 11:01 | |
| mnemotron | hey, what exactly is the best way to report problems with the zz9000? there is quite a large facebook group dedicated to the card and several people have problems in various configurations, but i think just crying on facebook won't help anyone :-D maybe an admin of the group could pin a post with a pointer to the right place. i presume that would be https://community.mnt.re/c/zz9000/5 ? | 11:05 |
|---|---|---|
| mntmn | mnemotron: yep | 11:09 |
| mntmn | mnt community is the best place | 11:09 |
| mntmn | i am not on facebook | 11:10 |
| mnemotron | so i figured ;) | 11:39 |
| mnemotron | question from myself: my A3000 needs some seconds to boot up - is this related to the usb boot function of the zz9000? should this get faster if i use the no-boot firmware? | 11:40 |
| mnemotron | i.e. Denkminute when switching it on ;) | 11:41 |
| mntmn | well, try the no-boot firmware! | 11:41 |
| mnemotron | k | 11:42 |
| mntmn | the autoboot stuff normally doesn't take several seconds, but YMMV | 11:42 |
| mntmn | i'm now debugging the speed of the Z2 firmware | 11:42 |
| mntmn | oop bustest at 0x210000 yields 0.9MB/s | 11:48 |
| mntmn | for reads. but writes are 1.8MB/s | 11:49 |
| _Bnu | I'm not sure if the cycle stuff was ever adjusted for Z2... | 11:57 |
| mnemotron | switched to the non-boot firmware, but i should've timed the other version first... i'll just use this fw, anyway. | 11:59 |
| mntmn | _Bnu: i just found a weird change that i made where i commented out the dtack timeout stuff, testing now | 12:01 |
| mntmn | mnemotron: the noboot one is a bit more stable if you don't need the autoboot feature | 12:02 |
| mnemotron | sounds good | 12:02 |
| mnemotron | and do you have any idea why the 256 mb won't show up on first boot? | 12:02 |
| mntmn | nope, unknown | 12:02 |
| mnemotron | wäh | 12:03 |
| mntmn | i mean, i know why, because that's a workaround | 12:04 |
| mntmn | when i enable the ram on cold boot, the board shows up as "defective" | 12:04 |
| mntmn | so far i couldn't figure out why, and why this happens only from 1.9+ | 12:04 |
| mnemotron | strange | 12:06 |
| mnemotron | probably some timing problems on initialization | 12:07 |
| mntmn | possible | 12:11 |
| mntmn | what is "normal" Z2 bandwidth/speed? | 12:13 |
| mntmn | on a3000 | 12:13 |
| _Bnu | Absolute maximum Zorro II speed is somewhere around 2.8/2.9MB/sec. | 12:13 |
| mntmn | ok thanks | 12:14 |
| mntmn | we never ever used axi dma for z2 huh | 12:14 |
| _Bnu | Nope. | 12:14 |
| _Bnu | I think because it was close enough to what it could feasibly do anyway? | 12:15 |
| _Bnu | While Z3 was really suffering from getting reads and writes passed through the ARM, haha. | 12:15 |
| mntmn | yeah | 12:17 |
| mnemotron | is there any advantage to using the z2 firmware in an a3000? | 12:18 |
| _Bnu | If you run some Kickstart that doesn't do Z3 autoconf, I guess. | 12:19 |
| mntmn | mnemotron: i'm just testing zorro 2 in general. | 12:20 |
| mnemotron | alrighty | 12:20 |
| _Bnu | Or if you want to do Z2 testing, like what mntmn is currently doing. | 12:20 |
| mnemotron | i have 3.2 roms in my a3000, no 1.4 anymore. i'll stick to z3 then. | 12:21 |
| mntmn | _Bnu: ok when commenting out the communication from FPGA->ARM->FPGA in the z2 read stuff, i get 2.1MB/s | 12:28 |
| mntmn | (dummy reads) | 12:28 |
| mntmn | cycle time 930ns vs 4000 or something insane | 12:28 |
| mntmn | so i slowed down something on the arm side >:| | 12:28 |
| mntmn | ok, it's not the housekeeping work in the isr0 | 12:46 |
| mntmn | interesting, increasing dtack_timeout makes it a bit faster | 13:20 |
| mntmn | _Bnu: ah, we _do_ use DMA for z2 _writes_ already | 13:34 |
| mntmn | trying to implement axi dma for z2 reads now so that we don't have to worry about arm read perf | 13:43 |
| mntmn | ok with dma the reads go up to 2.4MB/s but i also need one more wait state | 14:11 |
| mntmn | also, i get garbage data ;) | 14:12 |
| _Bnu | I like garbage data... | 14:41 |
| _Bnu | Well, I guess I prefer garbege datas. | 14:42 |
| _Bnu | Otherwise I wouldn't have used those for the PiStorm buptest. | 14:42 |
| mntmn | ok stuff is starting to work | 14:43 |
| mntmn | garbege!! | 14:43 |
| mntmn | i poke deadbeef and i get BEEFDEAD | 14:43 |
| mntmn | that's almost good! | 14:43 |
| _Bnu | Now you just got to get 32-bit burst writes/reads in there... | 14:44 |
| lastebil | wait, we have someone in here named mnemotron, is this a spreadpoint reference? | 14:44 |
| mntmn | _Bnu: 32 birt | 14:46 |
| - mnemotron (QUIT: Read error: No route to host) (~wayne@194.135.47.239) | 14:47 | |
| _Bnu | 32 berts and ernies... | 14:48 |
| mntmn | according to my code i need a different dtack length/time for tf536 than for other machines | 14:49 |
| mntmn | but this shorter dtack time makes the z2 reads+writes slower for other machines | 14:50 |
| mntmn | >:| | 14:50 |
| - Fate (QUIT: Quit: Poof!) (~fate@bnc.nordkommission.de) | 14:55 | |
| + Fate (~fate@bnc.nordkommission.de) | 14:56 | |
| mntmn | ok with axidma, z2 p96 is much faster | 15:01 |
| + mnemotron (~wayne@194.135.47.239) | 15:45 | |
| mnemotron | re | 15:45 |
| mntmn | wb | 15:46 |
| mntmn | grrrr when i open memacs now it doesn't switch to it, instead it changes the palette of workbench | 16:09 |
| mntmn | when toggling with amiga-m only the palette is toggled, not the pan offset | 16:10 |
| mntmn | split screen works for some reason | 16:10 |
| mntmn | oh man this whole panning code is so messy | 16:15 |
| _Bnu | 8D | 16:20 |
| _Bnu | It is fortunate that it was very well documented. | 16:20 |
| _Bnu | Maybe there is actually documentation for it in the archive now, I haven't checked recently. | 16:20 |
| _Bnu | Though. Did you accidentally delete any of my panning code stuff? | 16:21 |
| _Bnu | It was very problematic originally, because the pan offset would get overwritten by the scandoubler thing. | 16:21 |
| _Bnu | And then the scandoubler thing would sometimes not update to overwrite the RTG pan offset thing... | 16:22 |
| mntmn | lol | 16:26 |
| mntmn | i just put in some printf debugging | 16:26 |
| mntmn | now it works | 16:26 |
| mntmn | w t f lol | 16:26 |
| mntmn | ah but i also commented out a few things | 16:26 |
| mntmn | there are some calculations in case REG_ZZ_PAN_LO: that add the amiga screen pan offset, i commented out those for better testing, and then it worked again | 16:28 |
| mntmn | retrying with that code active... | 16:29 |
| mntmn | omg no i know what went wrong | 16:29 |
| mntmn | i had switched to Release mode and the BIF file references the Debug .elf | 16:30 |
| mntmn | :B :B :B | 16:30 |
| mntmn | it is slightly unfortunate that the pan stuff reuses the rect_x1 etc and blitter_colormode registers | 16:31 |
| mntmn | that way, it is very hard (impossible?) to debug with pokes on an rtg screen because p96 constantly clobbers these registers | 16:32 |
| _Bnu | Yeah, I mean, you could always rework it to use some dedicated ones, I think there's still a lot of free register space available. | 16:35 |
| mntmn | yeah, i think that's a good TODO | 16:36 |
| mntmn | now just have to fix regread... | 16:37 |
| _Bnu | IIRC it just uses one of the address entries or something with DMA RTG. | 16:37 |
| mntmn | currently i get 256 degrees core temperature | 16:37 |
| mntmn | _Bnu: ah ok! | 16:37 |
| _Bnu | (Rather than the X/Y values.) | 16:37 |
| _Bnu | But the DMA RTG stuff is way harder to get things mixed up with. | 16:38 |
| _Bnu | Unfortunately it's kind of unusable with Z2 due to address space limitations. | 16:38 |
| _Bnu | Though that could be worked around with something like the ZZ3660, haha. Just intercept the address range. | 16:40 |
| _Bnu | No address lines, no problem . . . | 16:41 |
| mntmn | well, the address could just be moved somewhere else | 16:41 |
| _Bnu | Yeah, though it's still somewhat problematic with only 8MB available. | 16:42 |
| _Bnu | Would have to skip out on things like P2C/P2D and such. | 16:42 |
| mntmn | so far there isn't even an 8mb firmware! | 16:42 |
| _Bnu | *a maximum of 8MB | 16:42 |
| mntmn | only 4 and 2... | 16:42 |
| _Bnu | Yeah, most often people have other crap installed that wants space there. | 16:43 |
| mntmn | :D | 16:43 |
| mntmn | i wonder how fast memory in z2 space normally is... | 16:43 |
| _Bnu | Max ~2.9MB/sec. | 16:44 |
| _Bnu | That's all it can really do. | 16:44 |
| mntmn | i mean some memory is _mapped_ into z2 space but it is not actually on the z2 bus | 16:44 |
| mntmn | on some accelerators | 16:44 |
| _Bnu | Yeah, because those accelerators suck. | 16:45 |
| _Bnu | They should be mapping it to Z3 in Z2 autoconf space on anything but Kick 1.3. | 16:45 |
| _Bnu | And on Kick 1.3, they should have a boot ROM mapping 32-bit Fast. | 16:45 |
| mntmn | mhm mhmmhmmhmhm | 16:46 |
| mntmn | ok z2 regreads don't arrive at the arm anymore | 16:47 |
| _Bnu | Good... they're not needed... | 16:47 |
| mntmn | almost not... | 16:47 |
| _Bnu | But yeah, any Kickstart 2.0 and up allows you to autoconf "Z3" Fast RAM over the Z2 autoconf range. | 16:49 |
| _Bnu | So if your accelerator handles the local Fast address decoding for the assigned range, you never ever need to take up Z2 space for it. | 16:50 |
| mntmn | 10 minutes before the next meeting... will i make the regreads work before... | 16:50 |
| _Bnu | And if you have a boot ROM you can just check if the memory has been assigned yet and if it hasn't you can just AddMemList it in the boot ROM and handle a hard coded address range for it. | 16:51 |
| _Bnu | Kick 1.3 curiously supports up to 2GB of Fast RAM, haha. | 16:51 |
| mntmn | ah lol regread is fpga read i'm stuipd | 16:53 |
| mntmn | s t u i p d | 16:54 |
| mntmn | ah cool i deleted important code | 16:54 |
| mntmn | code yes, test no | 17:02 |
| - RobDangerous (QUIT: Ping timeout: 260 seconds) (~Thunderbi@p200300ec8f03e00015c7397c97387cc6.dip0.t-ipconnect.de) | 17:03 | |
| - mnemotron (QUIT: Quit: He Who Must Not Be Named) (~wayne@194.135.47.239) | 17:09 | |
| mntmn | ok ok | 17:42 |
| mntmn | program failed | 17:45 |
| mntmn | illegal instruction | 17:46 |
| _Bnu | Important cord. | 17:47 |
| mntmn | don't pull on that one, you never know what it might be connected to | 17:54 |
| mntmn | ok regreads fixed | 17:56 |
| mntmn | *** who wants to test z2 firmware? *** | 17:57 |
| mntmn | http://dump.mntmn.com/zz9000-firmware-r-1_9_2-rc1-zorro2-noautoboot.zip | 18:05 |
| _Bnu | I told Mark Sealey about it on Discord... | 18:27 |
| _Bnu | I removed the Z-500 from my A500+ a little while ago, and I currently can't reach it... | 18:27 |
| _Bnu | Or I probably could, but it would be painful. | 18:28 |
| _Bnu | Grabbing the A4000 would be worse though, since it's on the floor over by the other table, and lifting it would be a terrible idea, haha. | 18:29 |
| mntmn | thanks @ mark | 18:49 |
| mntmn | oh yeah don't strain yourself! i was more thinking about other peeps here who use A2000 | 18:49 |
| mntmn | i will try on A500 next | 18:49 |
| mntmn | unclear what's up here https://community.mnt.re/t/bugs-in-1-9-1-update/549 | 19:18 |
| _Bnu | Uhhh, no idea. If it's the newer P96, Thomas Richter suddenly felt the need to change how the SetSprite thing works. | 19:36 |
| _Bnu | For the multi monitor whatever. | 19:36 |
| tryphoncosinus | I do not know exactly what you changed in this rc1 but native amiga screen issue I have in my A2000 NTSC changed a bit. After warming up, image is scrolling down and no more black screen. Only when I power on the Amiga after a long rest, image is jerky or alternatively black. | 19:41 |
| _Bnu | Do you have anything like an OSSC to hook up your A2000 to? | 19:45 |
| _Bnu | Check if the vsync is actually stable, or if it's bouncing around between two values constantly. | 19:45 |
| mntmn | yeah that would be interesting :3 | 19:57 |
| mntmn | tryphoncosinus: i didn't change anything about the video capture, but every fw has very slightly different timing so maybe you are exactly on the edge | 19:57 |
| mntmn | a500+ firmwares (for denise adapter): | 19:57 |
| mntmn | http://dump.mntmn.com/zz9000-firmware-r-1_9_2-rc1-a500plus-2mb.zip | 19:57 |
| mntmn | http://dump.mntmn.com/zz9000-firmware-r-1_9_2-rc1-a500plus-2mb-dtack2.zip (might work better with TF536) | 19:57 |
| _Bnu | Codename Mango Chutney 3.0. | 20:00 |
| mntmn | hehe | 20:03 |
| tryphoncosinus | I do not have any OSSC. I do not have access to vsync on the monitor, this part seems to be automatic. Looking at the screen for a longer time, the monitor tries to catch 800x600 or 720x480p modes. The latter looks like a misaligned two half images scrolling down. May be another monitor I have could let me control manually the vsync... | 20:14 |
| mntmn | maybe it toggles quickly between ntsc/pal or interlace? | 20:19 |
| mntmn | do you have a USB UART adapter/cable or could you get one? | 20:19 |
| mntmn | the firmware logs over uart if it is switching modes and ntsc, interlace | 20:20 |
| tryphoncosinus | This is not an interlace looking. This is a misalignment of pair lines and odd lines. I do not believe there is NTSC/PAL switching. I don't have any USB UART and frankly, I do not know what it is. | 20:24 |
| tryphoncosinus | I remember I used once a USB FTDI cable. | 20:33 |
| mntmn | that's kinda similar | 20:35 |
| mntmn | so a UART is just a very simple serial port | 20:35 |
| mntmn | with only RX/TX/GND lines | 20:35 |
| mntmn | there are some pins for this on the zz9000 | 20:35 |
| mntmn | tryphoncosinus: something like this https://www.amazon.com/IZOKEE-CP2102-Converter-Adapter-Downloader/dp/B07D6LLX19 | 20:36 |
| mntmn | of course we could also make some debug overlay in zz9000 itself but there is not much time atm :| | 20:37 |
| tryphoncosinus | Sure, I will be happy to help and make the debugging easier. Please, confirm it is the same item here (https://www.ebay.ca/itm/254459459793?hash=item3b3ef738d1:g:cuoAAOSwPsFfFfjF) and I will buy it locally. | 20:43 |
| mntmn | tryphoncosinus: should be fine. also you should get a usb 2.0 extension cable, or you will have a hard time connecting it | 20:51 |
| tryphoncosinus | This is done. Yes I have cables available (at least 2 male-female ones). Thank you. | 20:59 |
| mntmn | cool | 21:00 |
| + gdonner (~gdonner@c-98-253-93-116.hsd1.in.comcast.net) | 21:52 | |
| gdonner | so much cool new Amiga hardware in progress; just amazing | 21:53 |
| gdonner | hopefully the new Taiwan chip plant will soon be built in Arizona to mitigate the FPGA (and other chips) shortage | 21:54 |
| mntmn | ah tsmc in arizona? | 22:12 |
| mntmn | ah yeah that's cool. will go into production only in 2024 though, so we have to be patient ;) | 22:12 |
| gdonner | oy vey; yeah, a bit of a wait :-| | 23:10 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!