IRC logs for #openrisc Monday, 2014-03-10

--- Log opened Mon Mar 10 00:00:14 2014
blueCmdI've ordered a de0 nano and a,401,1003&Prod=PMOD-NIC10001:05
olofk_stekern: Can't remember what we said about it at the time, but is it possible to push the SoCKIT port with pregenerated HDL to avoid running qsys? Or is that too  messy?07:25
olofk_Damn! Someone beat me to it. I've had the idea for a few years. Thought that de0 nano would be a good fit
olofk_And of course they use a crappy MicroBlaze on it. That's what annoys me the most07:27
olofk_ahh... no.. wait07:28
olofk_It actually uses OpenRISC07:28
olofk_AltOR32 :)07:28
stekernthat's actually yet another or1knd implementation07:35
olofk_Who did that one? Was it Ruben Diez?07:41
stekernno, it's that guy that did the bus pirate07:44
olofk_Saw that too. Still can't find a name. Just says ultra-embedded everywhere07:45
olofk_Wow. I also want to be called Ultra-Rob07:55
olofk_It's a really nice board. I want one08:07
blueCmdolofk: seen that?08:37
olofk_blueCmd: Yep, I've seen that09:03
olofk_Hmm.. perhaps I should talk to him and see if he can sed s/orpsoc/fusesoc09:04
blueCmdolofk_: soo.. - i'm thinking of submitting it11:20
blueCmdtotally forgot about --thread, but yeah. hopefully it'll be fine12:13
olofk_blueCmd: Woohoo!! Good job12:31
blueCmda few patches seems to have got lost in the mail void, let's see if they appear13:27
gedarei'm looking for christian svensson if he might be around.. anyone can tell me his nick?18:22
stekerngedare: blueCmd18:31
gedarehi blueCmd18:31
blueCmdgedare: hello!19:01
gedaredid you get my email by now?19:01
blueCmdgedare: I did now, skimmed it - sounds good to me19:02
gedareok great. i'm one of the org admins for rtems19:02
gedareif you think of another possible openrisc mentor/co-mentor that might be nice to "pile on", but not mandatory i think.19:03
blueCmdgedare: so my background with the hardware side is limited for OpenRISC, i do have some background with HDL and FPGAs, but I'm quite strong when it comes to porting toolchains, OS internals and OpenRISC gotchas when it comes to the software19:03
gedarethat is what we needhere.19:03
gedareporting a new architecture in rtems is basically about getting a working toolchain with the rtems "triplet", and writing some basic low-level boot code, context switching, and a clock driver.19:04
gedarewe'll rely on you to help the student if he gets stuck trying to figure out some tricky bit of assembly for example.. like what registers are volatile in a function call, or how to access internal timers on an soc or some such..19:05
gedareblueCmd: i suspect this will all be in your wheel house.19:05
blueCmdgedare: yep, that sounds like my area19:06
gedaresuper. then i think the project has a good chance to get accepted, but it all depends on our overall project submissions and slot allocations from google.19:06
gedarethanks for stepping up19:06
blueCmdgedare: are there any rules for which kind of people can be mentors?19:07
gedarenice people!19:07
blueCmdor rules that I can read?19:07
blueCmdhah, I'm nice!19:07
gedarethere are... i can try to find them19:07
blueCmdjust don't ask stekern about it19:07
gedareare you from a country the USA does not export to?19:08
gedarethat's the biggest hurdle, but not impossible for a mentor actually.19:08
gedarethere it is.. if you're over 18 you're good to go, and if you can't receive exports you won't get swag19:09
gedareblueCmd: you may also like to skim this:
stekernask me what? where? did someone say my name!?19:50
stekernah, now I get it19:50
stekernblueCmd: you're nice, compared to me at least ;)19:50
blueCmdstekern: :D19:54
stekernolofk_: you work so hard changing the name and all, yet it still assumes the top module to be named orpsoc_top:
stekernseems I've got a talent to make quartus go sad panda...
olofkstekern: Yes, that hardocded orpsoc_top thing is a bit embarrasing. At least _franck_ has submitted some patches to fix it for verilator21:08
olofkstekern: Nice stack trace. File a bug report, and they might have fixed the issue by 2018... at least if you're a paying customer21:09
stekernthe worst thing is that it's workaroundable, but it requires 20x more code21:28
olofkGod, that sounds annoyin21:28
stekernit's my functions that the tools don't like again...
stekernI tried to tell it that it's no RAM, by using ram_style = "logic"21:29
stekernit didn't get my hint21:29
stekernit's 32 20-bit registers21:30
stekernthat I want to read all at the same time21:30
olofkIs it "logic" for quartus? synplify and xst uses "registers" iirc21:30
stekernat least that's what the manual says21:31
stekernstandardising that kind of things, what would the fun be in that?21:31
blueCmd*humpf* the mailserver gave me 15x "250 Queued", but only 10 mails appears to have made it. It worked when I sent to my local email21:32
olofkE-mail must be one of the greatest fucking mysteries of mankind21:33
blueCmdolofk_: not really, if you answer 250 you're supposed to not fucking drop it :P21:33
olofkI like working with computers, but I try to stay far away from printers, e-mail and perl. Those things I do not understand21:34
blueCmdif the emails haven't arrived 12:10 tomorrow I'll resend the missing ones21:35
blueCmdit's nice that the patches are away now though. hopefully it won't kill the maintainers21:35
olofk...or start killing one hostage every hour21:35
blueCmdI wonder if anyone uses,xge_mac21:36
olofkblueCmd: Haven't got a clue actually. General problem is that the smaller FPGAs that are supported by the free tools don't seem to come with high-speed transceivers21:38
olofkSo things like HDMI, PCIe and gigE are handled by external ASICs on those boards21:38
blueCmdolofk: for 1GE you have ethernet controllers that does a good work, but for 10G I haven't seen that many PHY / MAC external circuits, and the affordable ones I could find was either without PHY or with only PHY21:41
_franck_just finished a simulation (a blinker) with ghdl and a bench calling a C function. FuseSOC will support GHDL soon ;)21:43
olofkblueCmd: Haven't scanned the market for 10G phys21:43
olofkBut I realized that I lied a bit. gigE generally don't need high-speed transceivers21:43
olofk_franck_: Cool! I think there are actually some people who uses ghdl21:44
olofkTime to quit now. Hopefully my daughter will learn one day that nights are for sleeping, not jumping around in the bed21:47
stekernhuh, I thought nights were for hacking on open source RISC processors...22:00
stekernconnecting some extra dummy signals onto the registers keeps quartus happy...22:28
--- Log closed Tue Mar 11 00:00:15 2014

Generated by 2.15.2 by Marius Gedminas - find it at!