#qi-hardware IRC log for Thursday, 2012-12-13

wolfspraula small fpgatools update - I think my next target after blinking_led will be the j1 mini-core02:24
wolfspraulhopefully I can finish the blinking_led in about 1 more week02:24
wolfspraulthat's all :-)02:25
wpwrakwhee ! the blinking LED is the creation of life in a new universe. a core is merely evolution :)02:34
wolfspraulbtw this is what the .c looks like now after your model->rc cleanup... https://github.com/Wolfgang-Spraul/fpgatools/blob/master/blinking_led.c02:57
wolfspraulthat is the final version of blinking_led, only that the engine underneath does not yet generate a working bitstream...02:58
wolfspraul(and at the top I added a little verilog equivalent for comparison and documentation)02:58
wolfsprauloh actually, I just notice it's not the final .c yet. the clock network is there, but some other networks are missing...03:01
wolfspraulmore work in fnet_autoroute() :-)03:01
wpwrakhmm, quite a bit of redundancy03:10
wpwrakmay look better if you use static initialization for logic_cfg03:11
wpwrakoh, and why y before x ? if they're coordinates, that seems rather confusing03:12
wolfspraulyes I know, but there's a lot of fpga literature and sources that do that, although in general coordinates are confusing in the fpga world03:13
wolfspraulat the beginning I was thinking about this. x/y in the bottom left, x extending to the right and y upwards03:14
wpwrakah, i see. historical convention then.03:14
wolfspraulbut then I followed what looks like at least the largest 'convention'03:14
wolfspraulwhich is y going downwards (0 at the top), and y before x03:14
wolfspraulbut then, you find everything, because i guess everybody stumbles upon this and then partially goes this or that way03:15
wpwrakheh :)03:15
wolfsprauloftentimes in the same sources you find some coordinates where y extends downwards, and others where it extends upwards03:15
wolfspraulyes, let's just say 'convention', although I am just trying to find a convention where maybe there is no convention03:16
wpwrakmaybe you should just switch to polar coordinates :)03:16
wpwrakthere you are ;-)03:16
wolfsprauly before x, y increasing downwards, x increasing rightwards03:16
wolfspraulat least fpgatools has only 1 coordinate system, some other popular fpga tools have >10 :-)03:16
wolfspraulstatic init for logic_cfg, sure03:17
wolfspraulI am not focusing on that 'higher' layer right now that much03:17
wpwrak(10 dimensions) sounds like string theory :)03:17
wolfspraulwell the different types/categories of things each have their own coordinates03:17
wolfspraulso X2 'logic col' is one right of X1 'logic col'03:18
wolfspraulbut there may be a X0 bram col in between :-)03:18
wpwrak(static init) actually, i see that it's more complicated since you carry over a lot of information from step to step. you'd probably need some sort of preprocessor.03:18
wolfspraulno need right now, I focus on the engine underneath03:18
wolfspraulthat .c is just the tip of the iceberg03:18
wolfspraullater on maybe I find a way to write a bison parser or so, or we plug fpgatools in as an iverilog backend...03:19
wolfspraulit does look a lot cleaner now though03:19
wolfspraulwithout the if(rc) checks03:19
wpwrakyeah, it's pretty smooth reading03:20
wolfspraulsome more networks missing, the connections between the luts, and the outgoing led wire...03:21
wolfspraulmaybe 1 week is too optimistic, he03:21
wolfspraulfirst get the led to blink...03:21
wolfspraulthanks for looking over it...03:22
wpwraki see that you often have (y, x, idx) groups. maybe they could go into a struct, to reduce the number of arguments ?03:22
wolfspraulthe temptation of C++ looms over it, but we shall resist...03:24
wpwrakall the good things C++ tempts you with you can also do in C :)03:25
wolfspraulwhen it's time to bury the project and declare it a failure, I will introduce C++ templates and operator overloading one month in advance...03:25
wolfspraulthen we can blame C++03:26
wolfspraulthat's the secret plan B here...03:26
wpwrakmaybe that's why C++ is so popular03:27
wpwrakwell, its star seems to be sinking a bit. maybe developers have gotten better and don't have quite so many failed projects to explain03:27
mat146asus or acer?11:34
mat146http://allegro.pl/ShowItem2.php?item=2851480171 http://www.x-kom.pl/p/121329-notebook-laptop-15,6-acer-tm5744-i3-380m-4gb-500-dvd-rw.html11:34
--- Fri Dec 14 201200:00

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