| *** tpb has joined #timvideos | 00:00 | |
| mithro | https://www.youtube.com/watch?v=axCBA3VD5dQ | 00:15 |
|---|---|---|
| *** samsagaz has quit IRC | 00:53 | |
| mithro | tannewt: evening | 01:18 |
| mithro | tannewt: I'm currently struggling with a good way to store the buffers | 01:18 |
| tannewt | mithro what buffers? | 01:21 |
| mithro | tannewt: endpoint buffers | 01:22 |
| tannewt | its usually dma or direct ram access | 01:22 |
| tannewt | so you just tell the peripheral where it is | 01:23 |
| cr1901_modern | misoc/litex systems don't have a DMA controller I don't think (wouldn't be that difficult to add one I imagine) | 01:23 |
| mithro | tannewt: Sure, I'm just trying to figure out the best way to store the buffers in the ram.... | 01:24 |
| tannewt | what are your options? | 01:25 |
| mithro | tannewt: Pretty much anything :-P | 01:27 |
| tannewt | I'd leave it up to the code, not the peripheral | 01:27 |
| tannewt | its something tinyusb already manages | 01:28 |
| mithro | tannewt: Well, having a one 32bit pointer for each endpoint is very expensive | 01:28 |
| *** rohitksingh has quit IRC | 01:29 | |
| tannewt | ah, so you want an offset? | 01:29 |
| tannewt | samd actually stores them in ram in "sram registers" | 01:29 |
| tannewt | and the peripheral has one pointer to that | 01:30 |
| mithro | tannewt: So, I could have it so that as DATA OUT transfers happen, it just fills up a slot in buffers and each EP has a value which tells you which slot to look in to find the data... | 01:31 |
| mithro | Or I could just have each EP have a single buffer | 01:32 |
| tannewt | I think its good to have the location configurable because you save another copy to put it into something bigger | 01:34 |
| tannewt | its common for mass storage for the buffer to be longer than a single data transaction | 01:35 |
| mithro | high speed devices allow a packet size of 8, 16, 32 or 64 bytes | 01:35 |
| tannewt | but it may chain more than one | 01:36 |
| tannewt | all the msc data transfers will be at least 512 bytes | 01:38 |
| tannewt | so if you force a specific 64 byte buffer you'll incur a copy to the larger buffer later | 01:38 |
| xobs | Often they're just fifos. | 01:39 |
| xobs | So no need for dma. | 01:39 |
| mithro | xobs: That would be another option | 01:40 |
| mithro | I'm also trying to figure out if I should use separate buffer space for in and out endpoints or use a combined buffer space... | 01:45 |
| xobs | Usually I've seen it separated. | 01:47 |
| cr1901_modern | mithro: For litesdcard, when I last experimented, I ended up using a combined _double buffer_ space (buffer0-in, buffer1-in, buffer0-out, buffer1-out). Unfortunately I don't remember the rationale for this | 01:48 |
| cr1901_modern | in and out were separated by an offset, but they were both part of the same contiguous dual port memory chunk | 01:49 |
| *** swalladge has quit IRC | 02:09 | |
| *** swalladge has joined #timvideos | 02:19 | |
| xobs | Okay, on an ICE40, what would cause the VCCPLL line to draw so much current? | 05:15 |
| xobs | I don't have exact numbers yet, but I have VCCPLL on its own regulator, and that regulator gets warm as long as the FPGA is out of reset. This is regardless of whether the bitstream uses a PLL at all. | 05:15 |
| xobs | But I know if I spray cold at it, the VCCPLL regulator defrosts very quickly. And the other regulators not so much. | 05:16 |
| *** futarisIRCcloud has joined #timvideos | 05:40 | |
| *** rohitksingh_work has joined #timvideos | 05:58 | |
| *** rohitksingh_wor1 has joined #timvideos | 06:44 | |
| *** rohitksingh_work has quit IRC | 06:44 | |
| tannewt | xobs: do you know of an svd to c converter we could run during a build process? | 06:50 |
| xobs | tannewt: Just the svdconv.exe that cmsis generates. | 06:52 |
| tannewt | is it cross platform? | 06:54 |
| xobs | Not that I can see right now. Which makes it rather useless. | 06:54 |
| tannewt | ya, I agree | 06:54 |
| tannewt | I was thinking svd + converter would be a good foundation for cross-mcu libraries | 06:55 |
| xobs | svd is easy enough to parse in Rust, mostly because there's a crate for it already. I started writing a utility to generate documentation, but I'm sure it could easily be modified to generate C or H files. | 06:58 |
| *** rohitksingh_work has joined #timvideos | 07:00 | |
| tannewt | nice! | 07:01 |
| tannewt | will have to keep that in mind | 07:01 |
| *** rohitksingh_wor1 has quit IRC | 07:03 | |
| *** futarisIRCcloud has quit IRC | 10:49 | |
| *** mauz555 has joined #timvideos | 11:18 | |
| *** rohitksingh_work has quit IRC | 12:07 | |
| *** rohitksingh has joined #timvideos | 13:02 | |
| *** rohitksingh has quit IRC | 13:28 | |
| *** rohitksingh has joined #timvideos | 13:45 | |
| *** ZipCPU has quit IRC | 13:45 | |
| *** ZipCPU has joined #timvideos | 13:54 | |
| *** rohitksingh has quit IRC | 14:10 | |
| *** puck_ has quit IRC | 14:18 | |
| *** puck_ has joined #timvideos | 14:20 | |
| *** rohitksingh has joined #timvideos | 15:22 | |
| *** rohitksingh has quit IRC | 15:53 | |
| *** mauz555 has quit IRC | 16:10 | |
| *** mauz555 has joined #timvideos | 16:12 | |
| *** rohitksingh has joined #timvideos | 16:42 | |
| *** rohitksingh has quit IRC | 17:01 | |
| *** mauz555 has quit IRC | 17:13 | |
| *** rohitksingh has joined #timvideos | 17:14 | |
| *** mauz555 has joined #timvideos | 18:41 | |
| *** mauz555 has quit IRC | 19:23 | |
| *** rohitksingh has quit IRC | 19:42 | |
| *** rohitksingh has joined #timvideos | 19:55 | |
| *** mauz555 has joined #timvideos | 20:22 | |
| *** rohitksingh has quit IRC | 20:34 | |
| *** mauz555 has quit IRC | 20:41 | |
| *** mauz555 has joined #timvideos | 20:42 | |
| *** mauz555 has quit IRC | 21:04 | |
| *** mauz555 has joined #timvideos | 21:29 | |
| *** mauz555 has quit IRC | 21:40 | |
| *** mauz555 has joined #timvideos | 21:53 | |
| *** mauz555 has quit IRC | 22:02 | |
| *** mauz555 has joined #timvideos | 22:07 | |
| *** mauz555 has quit IRC | 22:22 | |
| *** Kripton has quit IRC | 22:39 | |
| *** futarisIRCcloud has joined #timvideos | 22:41 | |
| *** Kripton has joined #timvideos | 22:45 | |
| futarisIRCcloud | https://twitter.com/TinyFPGA/status/1065634543162544136 | 22:47 |
| froztbyte | mithro: heeeeeey you're doing a ccc talk | 23:31 |
| froztbyte | neat. | 23:31 |
| froztbyte | I'll be there :) | 23:31 |
| froztbyte | (and hopefully timing lines up to meet y'all) | 23:32 |
Generated by irclog2html.py 2.13.1 by Marius Gedminas - find it at mg.pov.lt!