#qi-hardware IRC log for Sunday, 2013-03-03

kristianpaulhmm why detect silently return nothing after cable milkymist set on urjtag..00:14
kristianpaulhmm debug seems not help that much.00:31
virichellekin: I was just trolling a bit for fun. Don't take me serious on that :)08:57
apeleteGood morning #qi-hardware09:28
apeleteI am planning to solder a serial console output port on my ben nanonote by following the wiki instructions here: http://en.qi-hardware.com/wiki/Serial_console09:31
apeletebut then I stumble upon the UBB: http://en.qi-hardware.com/wiki/UBB09:31
apeletecan the UBB be used to make a serial output port more easily ?09:32
apeletelarsc: no ?09:36
larscwell, you can have UART on the UBB, but that's gpio bitbanging based uart09:36
apeletelarsc: I don't get it, what do you mean ? what I need is a serial output to debug uboot and the kernel during aerly boot09:39
apeleteis that possible using a UBB or not ?09:39
larsc(well it is, given you invest enough time to implement it, but not out of the box)09:41
apeletelarsc: ok, so txd, rxd and gnd pins accessible under the ben's battery cannot be accessed with a UBB09:42
apeletelarsc: thanks a lot. it's a pity though :)09:44
larscthe UBB pins go straigth to the MMC pins of the JZ4740. And since there is no internal pin muxer you can only either use them as GPIOs or MMC.09:48
hellekinviric :P               :P10:15
kristianpaulwpwrak: about FPGA Programming for the Masses you pointed latelly, dont you think fpgas may take advantage if the model of running apps on a operating systems change to the app be the operating system, so it can be easilly tied to a hw modelated on a fpga?15:00
whitequarkmasses can't figure out threads, much less fpgas15:04
whitequarkkristianpaul: besides, I think that it makes much more sense to use FPGA as a reconfigurable coprocessor alongside a 'proper' host CPU15:04
whitequarksuch as parallella with the ARM core powerful enough to run ubuntu and gcc, and the SIMD array. it's the same overall design15:06
whitequarkthe article correctly notes that for any particular workload, FPGA is far more power (and cost) efficient15:06
whitequarkbut what it doesn't note is that for an arbitrary workload, traditional CPU still excel15:07
kristianpaulthreads, :)15:07
larscwell, a bit simplified, but still, cpus are good for sequential stuff, fpgas for parallel15:08
whitequarklarsc: that's only a part of the story15:09
kristianpaulsure sure, looks a network sw, it have a cpu for the general stuff and ASICs for the custom sw routing fabric15:09
whitequarkwhat will you choose, a real, hardware and fast (way faster than its fpga analogue) ARM core running linux, or a hypothetical ideal-for-linux core on an FPGA?15:09
whitequarkand not all parallel workloads can work on an FPGA15:10
kristianpaullarsc: but how to agree in that parallel computing, when at the end talking with the fpga relays on a memory maped sutff not something you could ask your kernel do for you, unless you write a driver ;)15:10
larscwell these days we have stuff like OpenCL to offload tasks for example to the graphics card15:14
larscthe same can be done with reconfigurable fpgas15:14
larscthe problem though is, you can't really built bitstreams on demand these days15:14
whitequark... yet15:15
viricapelete: when I started debugging kernel things, I wrote this: http://viric.name/cgi-bin/nanonixos/doc/trunk/doc/kerneldbg/index.wiki17:13
apeleteviric: wow, thanks17:19
apeleteviric: so you soldered a serial port on the nanonote to achieve this ?17:20
viricI soldered thin wires to the pads below the battery17:23
viricso I can have the battery in17:23
apeleteviric: ok, that's what I meant by "soldered a serial port"17:24
viricthen I have a ftdi cable, for 3.3v17:25
apeleteI bought an FTDI cable too. I was wondering if the "soldering wires" option was the only one to get a serial output on the nanonote17:26
apeletebut it seems there's no other way. soldering it will be.17:27
guanucoluis1hi all, anyone used a jtag-broad based ft2232d?17:29
viricdoesn't the sheevaplug use that?17:32
guanucoluis1yes, board, sorry17:32
--- Mon Mar 4 201300:00

Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!