| --- Log opened Tue Nov 19 00:00:31 2013 | ||
| hikenboot | hello poke53281 you around? | 02:20 |
|---|---|---|
| djbclark | hikenboot: No, the armv7 (version 7 I presume) processor - e.g. the thing that's in the beaglebone etc. | 02:59 |
| hikenboot | ah your wondering if the emulator can out perform armv7 my guess and its just a guess is no | 03:04 |
| hikenboot | but I was actually wondering if the vm might work if it had a host cpu driver in similar to vmware to emulate on a real 64 bit processor instead of the emulated processor. It would require a drive be installed on the host system which is probably why they didnt opt to do it...but I still wonder | 03:05 |
| hikenboot | be right back | 03:09 |
| hikenboot | back again hope I didnt miss anything! | 03:15 |
| hikenboot | of course it wouldnt actually be openrisc if it was done that way either! | 03:15 |
| hikenboot | gee wiz if I have to reboot windows one more time! | 03:40 |
| stekern | hikenboot: to me it sounded he wondered which was faster, emulate or1k or x86 by using the armv7 as the host | 04:08 |
| stekern | theoretically, or1k would probably be able to be faster, since it's a closer match as an architecture | 04:10 |
| stekern | but a lot more time has probably been spent on optimising the emulators for x86 than for or1k (although poke53281 have done a hell of a job improving the qemu or1k port in that regard) | 04:12 |
| stekern | djbclark: apart from being fun, why do you want to run an emulated or1k (or x86) on armv7? | 04:14 |
| stekern | I'll be completely satisfied with the answer "just for fun" ;) | 04:15 |
| hansfbaier | Hello, I have four sessions of 1:30h teaching at a university here, and I want to use the time to introduce OpenRisc to the students. | 06:11 |
| hansfbaier | I'm just a bit confused how to boil all of that down to something palatable for that short time. | 06:11 |
| hansfbaier | Also many of the students run Windows, so I need to bake a Live CD with everything on it. | 06:12 |
| hansfbaier | I'd like to simulate the GPIO core and show the linux driver of it. | 06:12 |
| hansfbaier | Which simulator would be best to use, or1ksim? | 06:12 |
| hansfbaier | (I'd probably use verilator to simulate the GPIO) | 06:13 |
| olofk_ | hansfbaier: Cool. Sounds like a great opportunity to show off OpenRISC and get students interested in computer architecture | 06:42 |
| olofk_ | verilator would be the fastest. You just need to write some C++ stuff that connects to the GPIO pins to indicate something happens when you run the code | 06:43 |
| olofk_ | Are you planning to use ORPSoC to build the system? | 06:43 |
| hansfbaier | olofk_: Would be nice, what do you think would be best? | 06:53 |
| olofk_ | I think you should extend or1200-generic with the GPIO core and use that. If you want to use GDB to upload programs however, you wont be able to use verilator right now | 06:55 |
| olofk_ | But you can still preload the RAM with a program | 06:56 |
| hansfbaier | olofk_: For gdb I'll have to use or1ksim? | 06:56 |
| hansfbaier | Would that work with GPIO? | 06:56 |
| olofk_ | hansfbaier: You can connect GDB to an Icarus or Modelsim simulation as well | 06:57 |
| olofk_ | I'm not sure if or1ksim has any support for GPIO | 06:57 |
| hansfbaier | I'm a total n00b concerning simulation, have used only FPGAs so far.... | 06:59 |
| hansfbaier | concerning OpenRISC | 07:00 |
| stekern | hansfbaier: I think julius did something like that (verilator with gpio) in mor1kx-devenv, I'd take a peek there if I'd be you | 11:44 |
| hansfbaier | stekern: I'm not sure If I'll do openrisc though, now thinking over it, I only have 2 weeks for preparing (alongside paid work) | 11:45 |
| hansfbaier | and also I need to whip up a live distro and everything | 11:45 |
| hansfbaier | and it's probably pretty much overkill for 4 sessions a 1:30h | 11:45 |
| hansfbaier | And also I don't have terribly much experience with simulators and openrisc which could provide for unpleasant situations | 11:45 |
| hansfbaier | Although it would be really nice. I think OpenRISC is the ideal learning environment for IT students. | 11:46 |
| hansfbaier | Or maybe dedicate one session as a demo showing the de0_nano and the SocKit | 11:51 |
| hansfbaier | stekern: will take a look at it though, If I could get it running quickly it would definitely be worth considering. | 11:54 |
| hansfbaier | stekern: Do you think verilator could run satisfactory on a netbook? | 11:57 |
| ysangkok | poke53281: hi | 12:14 |
| -!- ysionnea1 is now known as ysionneau | 17:18 | |
| poke53281 | hikenboot: Sorry, was not around the last two days | 17:38 |
| poke53281 | ysangkok: Hi | 17:38 |
| hikenboot | hi no problem | 17:38 |
| hikenboot | poke53281, without the command make fetchtoolchain the next command make checktoolchain fails with make: *** No rule to make target 'checktoolchain'. Stop? | 17:40 |
| hikenboot | have to run out, but just leave me a message if you have a hint to why it's doing that...it only happens when I skip the step you suggested I should skip | 17:46 |
| poke53281 | ignore checktoolchain. Continue with the next command. | 17:49 |
| poke53281 | ignore also fetchtoolchain. Just download the files I provided in the previus link | 17:50 |
| hikenboot | still happens unless I run the fetchtoolchain gives the make error...perhaps its something wrong with my ubuntu build machine...what distro are you using to build it? | 19:02 |
| poke53281 | I use Ubuntu too. | 19:11 |
| poke53281 | Ahh, sorry. I though you mean fetchprogs | 19:13 |
| poke53281 | Of course, fetchtoolchain should work. | 19:14 |
| poke53281 | fetchtoolchain will take extremely long. Approx. 3-10 hours. | 19:14 |
| poke53281 | And if it fails somewhere in between, you should restart the fetching of the toolchain. | 21:47 |
| poke53281 | Then it makes also sense to execute the "git clone" commands in the Makefile script separately. | 21:47 |
| hikenboot | it wont I did it once and it didnt fail...however I couldnt get anywhere without the make fetchtoolchain...I deleted it figuring it would fail later on as you said | 21:49 |
| hikenboot | thanks for the pointers..I appreciate it. | 21:49 |
| hikenboot | Question for you do you know about physical to virtual cpu mapping in qemu or xen or plex86? | 21:49 |
| poke53281 | The best way by building the toolchain is to follow the makefile scripts and understand what they are doing. Most of them are harmless. I didn't put a lot of checks in the script. This is the big problem. You need some experience in using these scripts. | 23:05 |
| poke53281 | hikenboot: Yes, in principle I know about these. | 23:05 |
| poke53281 | Everything is part of the memory management unit. | 23:06 |
| poke53281 | If you talk about virtualization and not emulation then you have to know that they use special hardware support of x86 processors. | 23:43 |
| --- Log closed Wed Nov 20 00:00:33 2013 | ||
Generated by irclog2html.py 2.15.2 by Marius Gedminas - find it at mg.pov.lt!