00:37.07 | IriX64 | A A' :) |
00:37.49 | IriX64 | forget how to switch them though and they say a different part of the anatomy is the first to go.:) |
00:38.46 | IriX64 | still got the book, programming for the Z80 :) |
00:45.16 | IriX64 | so lets see, byte 0 , byte 1 ... byte 512 thats sector 1 then immediatly followed by sector 2 ... to 1024 sectors, then and so on to 256 megs and ive got a portable floppy disk commonly called a jump drive. right? |
00:45.42 | IriX64 | left out track but it's implied. |
00:46.07 | IriX64 | fat is appropriate. |
00:46.20 | IriX64 | fat32 can be supported. |
00:46.30 | IriX64 | somebody check me. |
01:06.26 | IriX64 | where do i find #devicedriverwriters :) |
01:16.58 | Maloeran | Efnet's #kernel for Linux driver development, I'm sure there's something on freenode too |
01:18.11 | brlcad | same channel |
01:27.10 | brlcad | heh |
01:27.40 | Twingy | sean, I'm gonna make a bike that generates energy back into the grid |
01:28.09 | Twingy | it'll generate like 2 cents an hour :} |
01:29.11 | brlcad | heh, would make for a nice business venture for health clubs |
01:29.18 | Twingy | hehe |
01:30.03 | Twingy | heh |
01:30.15 | Twingy | only $10k in equipment |
01:30.25 | Maloeran | I'll only need 250 hours at 200 watt to risk exceeding that 5$ gift limit the next time I drop by |
01:30.34 | Twingy | haha |
01:31.09 | Twingy | I think I will need to make it an energy capture machine |
01:31.26 | Twingy | because at < 175W it doesn't do anything |
01:31.34 | Twingy | rerun |
01:36.14 | brlcad | er, don't think this is a rerun -- at least i haven't seen it |
01:36.44 | brlcad | and it just referred to the new episode from two weeks about about takara getting blown up |
01:37.35 | Twingy | ah |
01:37.41 | Twingy | well, I'll watch it in 2 hours |
01:37.47 | Twingy | comes on at 11 again |
01:42.16 | brlcad | the allucian alliance thanks you |
01:49.34 | *** join/#brlcad dan_falck (n=danfalck@pool-71-111-76-8.ptldor.dsl-w.verizon.net) |
02:04.59 | CIA-9 | BRL-CAD: 03lbutler * 10brlcad/src/librt/ (150 files): Doxygen updates |
02:24.54 | IriX64 | you picked a fyne tyme to leave me lucille ;) |
02:26.51 | IriX64 | sigh , I can hear the flapping of wings, but the song eludes me. :) |
03:11.18 | Maloeran | Erik, question : would it be of any value for ray-tracing nodes to be able to join and depart the work network at any time? |
03:12.33 | ``Erik | I think so |
03:13.11 | Maloeran | Okay, that's what I initially had in mind. Still planning chunks of distributed preparation and synchronisation |
03:17.49 | Maloeran | To translate fully built graphs from pointers back into indices for newly joined nodes is rather akward |
05:33.42 | CIA-9 | BRL-CAD: 03lbutler * 10brlcad/misc/doxygen_structure: Doxygen updates for librt |
07:29.52 | *** join/#brlcad clock_ (i=clock@84-72-88-219.dclient.hispeed.ch) |
08:00.47 | CIA-9 | BRL-CAD: 03brlcad * 10brlcad/misc/Makefile.defs: rework the noprod logic for issue encountered on netbsd on vax where the shell is invoking with a different precedence on &&/||, so be more explicit |
11:23.15 | *** join/#brlcad SWPadnos_ (n=Me@dsl245.esjtvtli.sover.net) |
14:59.13 | ``Erik | see ya in utah, dude |
15:07.12 | brlcad | wooh hooo! got benchmark working under vax finally |
15:07.33 | brlcad | stupid libm bug |
15:07.45 | brlcad | yeah, see ya Maloeran .. happy trails |
15:31.38 | ``Erik | nifty |
15:31.45 | ``Erik | and how many vgr's is a vgr? |
15:47.29 | brlcad | eh, don't know that yet.. still running the benchmark :) |
15:48.08 | brlcad | it's on 3 of 6 and been going for about an hour |
15:48.59 | brlcad | it's down to just a few hundred rays per sec though, so it should be interesting |
15:49.57 | brlcad | this isn't an optimized build yet either, so technically it could be even slower than 1 vgr if the originals were optimized (which I don't know if they were or not) |
15:50.45 | brlcad | though it certainly seems to feel like it's in the 0 to 10 vgr range |
15:52.21 | brlcad | spent *hours* last night debugging trying to figure out why it was crashing, stack would hose after calling atan2() and gdb would infinite loop up on certain prints of floats |
15:53.15 | brlcad | then spent hours relinking with a static libm just to make testing a little easier, and the problem was solved -- had to hack up libbn a little for vax floating point, but it's all good now |
16:22.36 | *** join/#brlcad clock_ (i=clock@84-72-60-33.dclient.hispeed.ch) |
17:48.03 | ``Erik | O.o |
17:53.47 | brlcad | Summary: |
17:53.47 | brlcad | Abs unknown 1198.23 432.80 371.64 346.31 639.24 642.15 605.06 Fri Sep 15 09:06:11 EDT 2006 |
17:53.50 | brlcad | *vgr unknown 8.74 6.45 6.62 6.49 9.04 .04 6.23 |
17:54.35 | brlcad | not to shabby.. i think it is throttled, just perhaps throttled to a faster vax than vgr used to be (which I can probably tweak via simh settings) |
17:56.34 | brlcad | also exposes a problem in the m35 reference image I think.. it still had the usual 74 off by 1 (though also 3 off by many) |
17:57.31 | brlcad | have to check the logs but i'm guessing someone uploaded a new ref image, keeping the same performance log, causing the continued off-by-1 numbers everywhere |
18:26.02 | ``Erik | <-- not so good at reading vgr accounts, mostly looks for the stuff in the -=-=-= |
18:30.24 | ``Erik | maybe you should run a really old version (like, the version our vgr's are based off of) to tune the vm, changes in the code may have improved efficiency... |
18:30.24 | ``Erik | :) |
18:43.27 | brlcad | vgr is the last number on the vgr line, 6.23 |
18:43.44 | brlcad | it's an average of the individual runs, which preceed it |
18:43.57 | brlcad | line before it is the same, but with rtfm counts |
18:44.45 | brlcad | i was thinking the same regarding improvements, that could very well explain the variations too if there were, say optimizations to the sphere shot routine |
18:45.00 | brlcad | checking to see if kermit can pull up some old binaries even |
18:45.17 | brlcad | though i really dont' want to install 4.3 bsd.. |
18:51.44 | ``Erik | um, compile old source on it? |
18:51.56 | ``Erik | maybe hack the simh config to make it like a 1ghz vax for the compile? hehehe |
18:52.50 | brlcad | still trying to find out if simh is clock synced |
18:52.58 | brlcad | or even settable |
19:20.24 | ``Erik | it's written in C... you have the source code... it's all settable. :D |
19:28.45 | brlcad | heh |
19:28.51 | brlcad | actually i've gotten some more info |
19:29.27 | brlcad | it's simulating a microvax II (1985), not a vax 11/780 (1979) |
19:29.43 | brlcad | i'm having trouble booting up the 11/780 simulator, but giving it a go |
19:30.25 | brlcad | someone who seems to be in the know in classiccmp seems to think it's not throttled, just happens to run about that slow |
19:41.20 | ``Erik | eck |
19:41.55 | ``Erik | what's cpu load when it's chunking? |
19:42.00 | brlcad | not sure i believe him though -- it's too close |
19:42.10 | brlcad | it's full cpu, single threaded |
19:42.26 | ``Erik | huh, if it's burning a full core, maybe it just IS that slow :/ |
19:42.38 | brlcad | it's full cpu even if it's idle |
19:42.50 | ``Erik | busywait :/ |
19:43.30 | brlcad | heh, http://www.openbsd.org/images/vax.jpg |
19:43.51 | ``Erik | hahahaa |
19:49.07 | ``Erik | amusing, I look for a pic of an 11/780 and it takes me to ftp.arl.mil |
19:50.52 | IriX64 | thats a hell of a hot vax. :) |
19:53.18 | IriX64 | simh eh? lets try. |
19:58.15 | *** join/#brlcad IriX64_ (n=IriX64@bas3-sudbury98-1168056792.dsl.bell.ca) |
19:58.44 | IriX64_ | was asking, what are you benchmarking the simulated machine with brlcad? |
20:02.37 | IriX64 | :) |