| --- Log opened Sat Jan 03 00:00:21 2015 | ||
| mithro | a long while back I came across someone who was slowly working on creating FOSS FPGA toolchain (working on bitstream generation rather than cores) - anyone seen that? | 06:38 |
|---|---|---|
| olofk | mithro: There has been a few attempts at that. Maybe you're thinking about yosys? | 06:50 |
| mithro | olofk: I remeber one in Python - but yosys and qflow seem to be further along... | 06:50 |
| olofk | sb0 has been working on this too, starting from the bitstream format iirc | 06:52 |
| mithro | It felt like something the m-lab guys might be into :) | 06:52 |
| mithro | olofk: there was a talk at c3 which I thought was about trying to do a FOSS toolchain for the spartan-6 but it turn out to just be people wanting to do it - not people who had done stuff | 06:54 |
| olofk | mithro: ah ok. I saw something about that. Thought they had done something too | 07:03 |
| mithro | olofk: I didn't watch the video, just read the slides | 07:04 |
| mithro | and looked at their github repos | 07:04 |
| olofk | Well, I guess it's good that people are still interested in this though | 07:05 |
| olofk | Personally, I'm most interested in an open-source synthesis tool to begin with | 07:05 |
| mor1kx | [mor1kx] bandvig pushed 1 new commit to withfpu: https://github.com/openrisc/mor1kx/commit/f47d4e369ed98939e7227de63961846ebc1a3f80 | 09:14 |
| mor1kx | mor1kx/withfpu f47d4e3 Andrey Bacherov: Replace digit recurrence division by Goldshmidt algorithm. The next stage is merging MUL & DIV pipes to share multiplier. | 09:14 |
| Me1234 | How to compile or1k-glibc (what source repositories to be used for gcc, linux, binutils, glibc)? | 09:48 |
| olofk | Me1234: Good question. I'm not sure about the glibc part | 10:42 |
| ysangkok | poke53282: is it necessary to specify courier as a preferred console font? on my ubuntu system, it looks horrible, and it is very hard to distinguish } and ) | 16:38 |
| poke53282 | The glibc building scripts are somewhere hidden here: https://github.com/bluecmd/or1k-debian | 21:14 |
| poke53282 | But I would suggest to use https://github.com/skristiansson/musl-cross if you are interested in a toolchain | 21:16 |
| poke53282 | ysangkok: Can you send me a image? | 21:16 |
| poke53282 | The font selection might be indeed problematic. You should use a monospaced font. | 21:17 |
| poke53282 | http://s-macke.github.io/jor1k/examples/simple.html | 21:17 |
| poke53282 | How does this one look like? | 21:17 |
| stekern | poke53282: I had to try jor1k on the new laptop that just arrived, I got around 50 MIPS in chrome | 21:33 |
| stekern | no, actually 70 | 21:35 |
| poke53282 | try firefox | 21:37 |
| poke53282 | I test the speed always with fbdemo. | 21:37 |
| poke53282 | stekern: Do you have a Haswell CPU? | 21:38 |
| stekern | 150 MIPS | 21:39 |
| stekern | in firefox | 21:39 |
| stekern | yeah, a i7 4510U | 21:39 |
| poke53282 | great | 21:39 |
| poke53282 | Doom with 30FPS :) | 21:40 |
| poke53282 | 150 is the value I expected. You can get up to 200 on a XEON cpu. | 21:44 |
| stekern | in fbdemo I get 190 | 21:44 |
| poke53282 | Yes, that's a Core-i7. :) | 21:44 |
| stekern | I wanted to be adventurous and try plasma 5, it's pretty sleek, but not so stable... | 21:46 |
| poke53282 | KDE? | 21:48 |
| stekern | yes | 21:48 |
| poke53282 | Ahh, on your Laptop, not in jor1k. :) | 21:48 |
| poke53282 | For one second I was confused | 21:48 |
| stekern | heh, plasma 5 in jor1k would have been cool though | 21:49 |
| poke53282 | Yes, like Firefox | 21:50 |
| poke53282 | My guess is, that the new Intel CPUs optimize especially for the code generated by Javascript. Either directly or indirectly (by making boundary checks a no-cost operation) | 21:57 |
| -!- janus_ is now known as ysangkok | 23:58 | |
| --- Log closed Sun Jan 04 00:00:22 2015 | ||
Generated by irclog2html.py 2.15.2 by Marius Gedminas - find it at mg.pov.lt!