| --- Log opened Wed Jun 19 00:00:50 2013 | ||
| -!- Netsplit *.net <-> *.split quits: stekern, olofk | 03:50 | |
| -!- Netsplit over, joins: stekern, olofk | 03:52 | |
| -!- Netsplit *.net <-> *.split quits: stekern, olofk | 03:54 | |
| -!- Netsplit over, joins: stekern, olofk | 03:55 | |
| olofk | autotools is fun! | 07:01 |
|---|---|---|
| olofk | orpsocv3 is now system-installable :) | 07:02 |
| olofk | But I can't figure out how to install a whole directory tree into pkgdatadir without listing each file | 07:40 |
| olofk | Hmm.. it seems to be a bit more complicated than I was expecting | 08:36 |
| stekern | don't look at me, autotools scares me | 08:39 |
| olofk | Yeah, me too, but it lured me in with promises of trouble-free distributable tar balls | 09:00 |
| stekern | that's how it always begins, the first tar ball is always (trouble-)free | 09:01 |
| olofk | And it works great as long as you don't try to do things outside of the box...problem is the size of the box | 09:01 |
| olofk | :) | 09:01 |
| olofk | I just made a highly unofficial orpsoc 3.0 release. https://dl.dropboxusercontent.com/u/902237/orpsoc-3.0.tar.gz | 11:54 |
| stekern | nice | 11:57 |
| stekern | releases being done isn't something that happens often in this community ;) | 11:58 |
| olofk | I've come to realize the same thing | 12:05 |
| olofk | I'm thinking about getting a de0-nano. Seems like it's the most used board => More testers for orpsocv3 | 12:06 |
| stekern | I wanna get started on the sockit | 12:22 |
| stekern | I have unboxed it at least ;) | 12:22 |
| stekern | it's even plugged in at my desk home | 12:22 |
| stekern | and I've got quartus 13.0 installed | 12:22 |
| olofk | Then you're basically done. The rest was just a few clicks :) | 12:51 |
| juliusb | olofk: grats on the first ORPSoCv3 release :) | 12:52 |
| juliusb | regarding the version number change in OR1200.... maybe we should revert that? | 12:53 |
| juliusb | (de0 nano for development) yes I've found that's a great little platform | 12:55 |
| juliusb | piece of cake to program | 12:55 |
| juliusb | cheap | 12:55 |
| juliusb | can do the debugging of OpenRISC systems over the programming cable | 12:55 |
| olofk | I love my ordb2a, but having a LED wouldn't hurt | 13:07 |
| hno | if only quartus could use the jtag interface on ordb2a.. | 13:56 |
| olofk | hno: Yeah, that's a problem. No chance to use signaltap :( | 14:08 |
| olofk | Ok, megalomania is starting to hit me hard now. What's the minimal amount of peripherals I would need to boot linux? Is an UART enough? | 14:11 |
| hno | UART, and some timer should be enough (well, also MMU, interrupt controller etc). To actually do something you might need more depending on your definition of do.. | 14:17 |
| olofk | Most of those things are internal in both or1200 and mor1kx. | 14:24 |
| olofk | My definition of do would be to get through a complete boot and perhaps look at /proc/cpuinfo or something | 14:25 |
| olofk | I'm planning to do it on an evenet based simulator. Last I heard someone tried to to that, it took a few days to get through a boot. Would be fun to see if things are faster now | 14:26 |
| stekern | uart and enough mem to fit linux is all you need | 14:45 |
| juliusb | stekern: what would qualify as enough MEM? 8MB? | 15:30 |
| stekern | I think so, 16MB to be safe | 15:31 |
| stekern | and, yes, I think we should fix that version commit | 15:32 |
| echo083 | hello, is it an active channel ? | 16:02 |
| _franck_ | hi, yes it is | 16:05 |
| echo083 | _franck_ left before i was able to answer | 16:26 |
| echo083 | maybe i'll wait another few minutes :p | 16:26 |
| echo083 | please prepend my nickname if you answer me :) | 16:26 |
| hno | echo083, if there is anything you want ot ask then just ask. | 16:42 |
| echo083 | hno, a fpga can work without a soft processor like openrisc ? | 16:42 |
| echo083 | hno, ? | 16:45 |
| hno | echo083, yes ofcourse. It can implement any logic you want. | 16:49 |
| echo083 | hno, i can define my own process to be executed without loading a processor on it ? | 16:50 |
| hno | echo083, if you know how to write/express that process in verilog or vhdl then yes. | 16:51 |
| echo083 | hno, do you know the performance difference when using or not using any processor ? | 16:51 |
| echo083 | hno, i did some vhdl few years ago but i almost forgot everything | 16:52 |
| hno | echo083, "raw" logic is a lot faster than a program running on a soft core. | 16:52 |
| hno | but also much less flexible. | 16:53 |
| echo083 | hno, oh really :( do you have any number ? factor ? twice faster for example | 16:53 |
| echo083 | don't worry i just have a single question remaining ;) | 16:53 |
| hno | I don't have numbers but I would guess anywhere from 50 to 10000000 depending on the task. | 16:55 |
| echo083 | okk and my last question, how do i know how many processes the microchip can execute in parallel ? | 16:55 |
| hno | and 50 is probably very optimistic (for the soft core). | 16:56 |
| hno | echo083, you don't. It all depends on what it is you want to do. | 16:56 |
| hno | But you do know how many logic elements the FPGA have, which defines how complex logic design you can fit in the FPGA. | 16:57 |
| hno | there is not really processes in FPGA. | 16:58 |
| echo083 | if my program fits on the microchip and i declare 1000 processes to be executed in parallel if it can only manage 2 process it will just wait ressources to be available to execute remaining processes ? | 16:58 |
| echo083 | you mean it is rather thread ? | 16:58 |
| hno | It's logic and clocks. | 16:58 |
| hno | all logic executes in parallell, syncronised by clocks. | 16:59 |
| hno | If your logic design is too large then the FPGA compiler will fail, telling you that it can't fit in the selected FPGA. | 16:59 |
| echo083 | hno, the maximum size of my program is defined by the number of gates ? | 17:00 |
| echo083 | i've seen sometimes 200K gates in the microchip characteristics | 17:00 |
| hno | That's one way to see it. | 17:01 |
| echo083 | what is the main component of the microchips ? | 17:01 |
| hno | Of an FPGA? http://en.wikipedia.org/wiki/FPGA | 17:02 |
| echo083 | i tried to make the link between the gates and reality | 17:02 |
| echo083 | according to your link it is logic cell | 17:03 |
| hno | Yes. An FPGA have neither gates or reality. | 17:04 |
| echo083 | if i have a two microchips on a pcb i can reprogram the second microchip only from the first one ? | 17:05 |
| echo083 | (no link with the previous question) | 17:05 |
| hno | what is a microchip? | 17:06 |
| echo083 | or the manufacturer always provide a cable which can reprogram independantly both microchip | 17:06 |
| echo083 | microchip = fpga | 17:06 |
| echo083 | in my language :p | 17:06 |
| echo083 | maybe i'll worry about pcb with multiple fpga later ... :) | 17:07 |
| _franck_ | AFAIK, yes you can chain multiple FPGA | 17:08 |
| hno | yes. | 17:08 |
| hno | but you don't normally do it in the sense that one programs he other. | 17:09 |
| echo083 | ok | 17:09 |
| echo083 | i think i asked the main thinks i need i'll come back later with more questions for sure | 17:11 |
| echo083 | hno, thx for your help | 17:11 |
| echo083 | _franck_, thx too | 17:11 |
| -!- Netsplit *.net <-> *.split quits: jeremybennett, ams, forkG, pikeaero | 18:00 | |
| -!- Netsplit over, joins: jeremybennett | 18:07 | |
| stekern | heh, no reality in fpgas, priceless ;) | 18:11 |
| hno | stekern, well, for mostly everyone the resources in an FPGA is virtual. | 22:08 |
| --- Log closed Thu Jun 20 00:00:51 2013 | ||
Generated by irclog2html.py 2.15.2 by Marius Gedminas - find it at mg.pov.lt!