| --- Log closed Fri Oct 26 06:02:28 2012 | ||
| --- Log opened Fri Oct 26 06:02:44 2012 | ||
| -!- Irssi: #openrisc: Total of 21 nicks [0 ops, 0 halfops, 0 voices, 21 normal] | 06:02 | |
| -!- Irssi: Join to #openrisc was synced in 21 secs | 06:02 | |
| stekern | wtf, snow... | 08:14 |
|---|---|---|
| juliusb | stekern: noooo! winter has officially arrived, then? | 12:40 |
| stekern | jupp | 12:44 |
| jeremybennett | I see Parallella has just a tad over $70k to go: http://kck.st/PtAZ9O | 21:32 |
| LoneTech | yep | 21:34 |
| LoneTech | nice graph: http://canhekick.it/projects/adapteva/parallella-a-supercomputer-for-everyone | 21:34 |
| LoneTech | currently moving at about $16k/h | 21:34 |
| LoneTech | I didn't get the /. article to the front page though | 21:35 |
| LoneTech | also, no serdes on the external pins. guess we can't have it all yet. | 21:36 |
| LoneTech | jeremybennett: what do you think of the processor groups (protected memory) idea? | 21:36 |
| LoneTech | I've later figured you'd probably want a "friend" level also to allow data passing from one group to another.. though that could be done by letting a group-0 node dma it | 21:38 |
| jeremybennett | LoneTech: Like the graph. simoncook has been doing something similar specifically for Parallella. Interesting graph of how much per hour he has to raise. Up until today it was steadily rising. Then it fell off a cliff and has plummeted all day. | 21:56 |
| jeremybennett | Seems to be going up at about $10k/hour at the moment. | 21:57 |
| LoneTech | jeremybennett: at current rate (per a script I've been running maybe an hour and a half) we have a bit over four hours to success. | 21:57 |
| jeremybennett | That's what I am hoping. | 21:59 |
| LoneTech | I personally believe our efforts contributed, though it's very hard to tell how much | 21:59 |
| LoneTech | the new options (case, book etc) also help, but there are nearly half as many new backers as raisers (~135/292) | 22:02 |
| LoneTech | er, sorry. nearly as many, I mean; 135 new backers in 292 increases of the total | 22:03 |
| LoneTech | we just got another $1000 backer | 22:04 |
| olofk | Just today I have seen a few articles about parallella. Both eetimes and phoronix and a few swedish news sites. I wonder why they are reporting so late | 22:05 |
| LoneTech | I spotted one article from someone who had just heard of it, but otherwise I think there was a dose of desperate backers trying to get attention | 22:08 |
| LoneTech | it's been submitted at least three times to slashdot in the last two days, for instance | 22:08 |
| Amadiro | I wonder if their IP would really have any practical use for anybody | 22:10 |
| Amadiro | In particular the ARM + parallella core thing seems a bit useless, since the ARM processor has so much power compared to the tacked-on parallella one | 22:11 |
| olofk | So if I donate $99 (+20 for shipping) I'll get a board with a 8 core adapteva + a zynq? | 22:11 |
| LoneTech | you mean the CPU RTL? not much. I guess the toughest part of it is the dual instruction issue | 22:11 |
| Amadiro | But I haven't really looked closely at it | 22:11 |
| LoneTech | Amadiro: the ARM is impressive, sure, but it does not have that much processing capability in the comparison | 22:12 |
| LoneTech | olofk: yep, if the kickstarter succeeds. | 22:12 |
| LoneTech | er, 16 core, not 8 | 22:12 |
| Amadiro | LoneTech, what dual instruction issue? | 22:12 |
| olofk | Ok, as much as I dislike Xilinx, just getting the zynq does it for me. I consider the epiphany an added bonus :) | 22:13 |
| LoneTech | each epiphany core has two operational pipes; int/int or int/float (switchable at any time). So it actually runs dual instructions. of course, it could be they're both encoded by the compiler and I just overlooked that, in which case the decoder is simple | 22:13 |
| LoneTech | olofk: then you get a zynq devel board with just 42 IO, not a bad deal anyway | 22:14 |
| olofk | Exactly. And it feels good to promote some kickstarter project that is aimed at us embedded/fpga/hardware guys | 22:15 |
| LoneTech | I agree | 22:16 |
| jeremybennett | < $60k to go. Quite addictive watching the figure tick up. | 22:16 |
| olofk | So shall I just enter $120 in the pledge amount, and they will automatically understand that I want the board? There is an option to select $99 directly, but that wouldn't cover the estimated shipping | 22:17 |
| jeremybennett | Yes - enter the $120 and tick the $99 reward. I think the extra $20 for shipping outside the US is done on trust. | 22:17 |
| LoneTech | olofk: you select the option for the base package and add on enough for shipping (and extras like the book or case) | 22:17 |
| LoneTech | in my case, I think I tacked on 75 or so (shipping, early sdk, and I like round numbers) | 22:18 |
| LoneTech | I loathe the ..99/98/95 price tags | 22:19 |
| jeremybennett | The wonders of Benford's Law: http://en.wikipedia.org/wiki/Benford%27s_law | 22:20 |
| jeremybennett | (actually it's a counter-example) | 22:21 |
| olofk | Do I need an amazon account too? This must be account-creating day for me | 22:21 |
| LoneTech | I think so. they handle the payment service | 22:22 |
| olofk | Then it's done | 22:26 |
| LoneTech | cool :) | 22:26 |
| olofk | And the first thing that happens is that four other hardware projects pops up that looks really interesting. I'm going to be broke soon | 22:27 |
| LoneTech | aww | 22:27 |
| LoneTech | just be sure to inspect them; it's not terribly unusual that wild promises are made | 22:27 |
| LoneTech | they've tried to cut down on it with rules like working prototypes, but I expect there are slipups | 22:27 |
| LoneTech | the previous project I backed is http://www.kickstarter.com/projects/1703567677/icontrolpad-2-the-open-source-controller | 22:28 |
| olofk | I have read about that one, but it didn't catch my interest | 22:34 |
| olofk | I backed Double Fine Adventure though, but that was after they already had reached their goal, so I guess it counts more as a pre-order | 22:34 |
| LoneTech | I did order a Pandora console (same makers) way back, eventually cancelled that | 22:34 |
| LoneTech | the Oculus Rift was also quite cool, but I passed on that. I think it might be a good fit together with a video input and a parallella, considering it needs an odd reprojection for perspective | 22:36 |
| LoneTech | i.e. remote controlled toy + dual fisheye wireless cameras + receivers to converter into zynq, reproject to match lenses, and hdmi out to rigt panel, for immense wide angle first person view | 22:37 |
| LoneTech | it's an odd case where mobility and low latency matter | 22:38 |
| olofk | Sounds a bit like Tobii's eye-tracking stuff. | 22:41 |
| LoneTech | well, eye tracking would be fun too, but I was just thinking remote view to start | 22:41 |
| LoneTech | mind sharing what projects interested you? | 22:41 |
| olofk | I closed the page, so I'm not sure anymore. They might not have been so interesting either if I had looked a bit closer. Mostly looked like different dev boards with some special feature | 22:44 |
| LoneTech | "smartduino" seems like another stackable module format.. might want to look at their ideas | 22:48 |
| jeremybennett | Supper calls. I hope by the time I get up tomorrow morning it's all safely funded. | 22:50 |
| LoneTech | it has slowed a bit, but I think there's a fair chance | 22:51 |
| LoneTech | $697k now | 22:51 |
| LoneTech | man, the smartduino people love awful capitalization. it's painful to read their text. | 22:53 |
| olofk | Hmm.. what do you call the collection of tools that put together fpga-specific binary from HDL sources? I'm looking for a good name to use in orpsocv3 | 23:10 |
| olofk | build_tool, toolchain, backend, par? | 23:11 |
| LoneTech | if you mean from sources, you'd include the synthesis | 23:11 |
| olofk | Yes, there will be support for exchanging the synthesis tool later on, but the default will be to use the vendor-specific synthesizer. | 23:12 |
| olofk | The name doesn't have to imply that synthesis is included in the definition | 23:12 |
| LoneTech | the tools after synthesis have had a bunch of different names.. fpgas tend to have place&route while some cplds only call it a fitter | 23:13 |
| LoneTech | it's a jumble. I think we lack a proper term for the set. | 23:13 |
| olofk | I think I'll use backend for now. That can always be changed later. As there aren't any users yet, no one will complain | 23:14 |
| LoneTech | Altera do still call the whole set fitting | 23:14 |
| LoneTech | ok | 23:14 |
Generated by irclog2html.py 2.15.2 by Marius Gedminas - find it at mg.pov.lt!