IRC logs for #openrisc Wednesday, 2013-08-14

--- Log opened Wed Aug 14 00:00:10 2013
amsolofk: yes, ld01:00
amsjuliusb: an archive is a library in any sense of the word01:01
amsolofk: .so and plain old executable are actually several object files combined into one object; the format (ELF) of .so, .o and an executable is the same.01:05
amsa .a file is more or less exactly like a .tar file01:05
stekernI think it's this ps/2 core that's borken...02:23
stekernI have yet to find a *good* description of the physical protocol03:07
stekernnone clearly states if the device reads data on rising edge or falling edge of the clock03:08
stekernfinally, the mouse is recognised in Linux05:58
stekernI bet my changes to the ps2 core broke the keyboard05:58
stekernbecause I had tweaked it to work with that, and now I basically removed those tweaks...05:58
stekernso, hopefully *tonight* I might be able to play some06:01
juliusbams: ah! interesting, thanks.08:27
juliusbstekern: slots for presentations at ORCONF filling up fast, you'd better get your submission in soon ;)08:32
juliusbactually, we have 2 so far! which is cool08:32
juliusbwe have basically confirmed the venue, too, the Cambridge Computer labs08:33
juliusbThe William Gates building, I believe it's called08:33
juliusbnot sure why they didn't call it the Bill Gates building08:33
stekernoh oh, what should I present?08:34
juliusbno, up to you08:34
stekernfamily photo slides?08:34
juliusbmaybe some intepretive dance?08:34
stekernyeah, that sounds good08:34
stekernwhat do we have so far?08:35
juliusbAnyway, I'll do an email later with more details08:35
stekernI think I'd like to do a "what have happened in mor1kx cappuccino since last orconf" at least08:37
juliusbStefan Wallentowitz emailed me just now to ask to present his OpTiMSoC stuff, and there's another student who has built their own SOPC-builder type thing using the OR1200 as the central piece of IP and he wants to come and demo and talk about that08:38
juliusbyeah for sure08:38
stekerncould have the side effect of updating the documentation a bit too ;)08:38
juliusboh... no.... don't do that!08:38
juliusbpeople might find it easier to use!08:38
stekernoh, those sounds very interesting08:38
juliusbyeah I really want to hear from Stefan08:39
juliusbi want to finish my little bit of work with the TCM08:40
juliusbit's working with all the GCC tests now08:40
juliusbthere's one problem though in that you could technically execute from outside the TCM address range, but I think my fetcher doesn't work for that08:40
juliusbso if it doesn't have single-cycle fetch then it's busted, but I will try to sort that out08:40
juliusbbut i need a test for that08:44
juliusband not sure how to split the software between to address ranges08:44
juliusband load the software in properly, too 08:44
juliusbi may just release it with the restriction that all code must be in the tcm08:44
stekernthat's why I'm thinking about adding the support in the current fetcher, so it still works for people who like to mix08:44
stekernbut there's a lot other things that are in flux in the fetcher and lsu right now08:45
stekernI want to move the whole refill and write logic out of the caches and into the fetch and lsu08:45
stekernbecause now there is duplicated bus access logic in the caches and fetch/lsu08:46
stekernso the cache interface would basically be (from the cache side) refill_req_o, refill_start_addr_o, refill_done_i08:47
stekernand then just a "write port" into the cache RAMs exposed to the fetch/lsu08:48
stekernand that makes sense for the store buffer logic too, because then that could be residing in the lsu and not in the cache08:49
stekern(release with restriction) do that08:59
stekernI'm of the opinion that it's better to make things available that works OK, but have a known set of restrictions and drawbacks, than to keep them unpublic until everything is 'perfect'09:02
stekernas long as it doesn't break anything or imposes restrictions and drawbacks on already existing functionality09:05
stekernaah! I'm finally able to play!13:29
stekernmouse works perfectly in dott, and it13:30
stekern's actually really playable13:30
stekerni.e. things moves around smoothly13:30
stekernwell, relatively at least =P13:36
olofkstekern: That's great news. Was it all in the ps2 rtl you had to change things?14:01
stekernto get the mouse working14:05
olofkIs the keyboard still working?14:05
stekernI guess I just bandaided over the problems when I got the keyboard working last time I looked at that core14:06
olofkIt's good then that there is support to apply patches to cores in orpsocv3 :)14:06
stekernit's actually just good that it runs a bit slower than on fast machines, gives it that authentic feel when you have to wait for scenes to change14:08
olofkJust like downloading porn on a modem connection. Always felt a little more like a strip show when the pictures slowly appeared from top to bottom14:09
stekerngood analogy14:09
stekernand I'm happy that I've got the keyboard working, using the shortcuts is the way to play those games14:10
olofkThe only time I had to use short cuts was in the Virtual Reality scene in Sam'n'Max. Couldn't click fast enough to pull out the sword and what else had to be done14:12
olofkBut I couldn't live without using . of course. They talk so slow in all Lucasarts games14:13
stekernI've never been a huge fan of sam'n'max14:13
stekernand the version I had had a bug that crashed the game when you are looking through a pair of binoculars, so I've only played it to the end in scummvm14:14
olofkThings like that tend to take the fun out of games14:14
stekernI haven't tested running on or1200 yet, I'm building an orpsoc right now to see how it compares14:22
olofkWhat's missing in mor1kx right now that or1200 has? FPU?14:22
stekernyeah, that's about it I think14:24
stekernand MAC instructions14:24
olofkI've been thinking about switching to mor1kx as a basis for all my orpsoc experiments. Doesn't really matter much which CPU I use14:25
stekernof course you should14:54
olofkI'm getting close to having a nice structure for the verilator stuff, so I might do that next14:55
stekernfull throttle is pretty much playable too:
stekerncurse of mi, not so much :(15:12
stekernit works, but too slow and jerky15:20
stekerni'd like to swi15:33
stekerntch over to orpsocv3 too15:33
stekerneasier to concentrate on one unified soc, instead of two15:35
stekernand then there's a bunch of orpsocv2 stuff that I should push to orpsocv2 in svn, but feels like that effort would be better spent shuffling stuff over on orpsocv315:37
stekerndott is playable on or1200, but noticably jerkier and slower than on mor1kx16:43
stekernfull throttle isn't really playable on it16:44
olofkVerilator is running in orpsocv3 now! It's an ugly as hell hack but I got the basics worked out at least21:32
olofk11 of the orpsocv2 tests makes an assert fail. Any clues?21:33
--- Log closed Thu Aug 15 00:00:12 2013

Generated by 2.15.2 by Marius Gedminas - find it at!