#milkymist IRC log for Saturday, 2012-02-11

fpgaminerkristianpaul: Glad you got it working :) I didn't do the port with serial communication on it :P But yeah, I vaguely remember having to tweak the clock on that port00:14
kristianpaulBtw how this thing of00:16
kristianpaulBtw how is the math related to the relation of hashes per second and megahertz00:18
kristianpaulfor example i set your minner loop log of 5 but lets said i can achieve 200Mhz clock00:18
kristianpaulWill be the same if i got a loop log2 at 3 but 100Mhz  clock?00:19
kristianpaulsame performance i mean00:19
fpgaminerloop_log2=0 will give 1 hash per cycle00:38
fpgaminerloop_log2=1 will give 1 hash every two cycles00:39
fpgaminerloop_log2=2 will give 1 hash every four cycles00:39
kristianpaulah ok00:44
rohfpgaminer: are you using a m1 to mine?01:03
kristianpaulfpgaminer: had you ever used serial comunication with your minner core?01:04
kristianpaulI would like to know if what i'm tryin here worked for someone else already :)01:04
wpwrak_bah, where's the pioneering spirit in today's youth ?01:13
kristianpauli'm trying !!!01:20
Action: kristianpaul already bringed the scope..01:21
kristianpaulbtw i havet xst asumes GNS a when a non defined wire is assigned to a instanciate module :-|01:21
kristianpauldoest work i doest work as i would expect.. one of two i need to confirm ;)01:22
wpwrak_i already realized that it's mandatory to study the logs very carefully. xst won't just stop and tell you if something bad happens.01:42
kristianpaulat least until the point you know what belons to what :)01:43
fpgaminerkristianpaul: I haven't personally used serial comms with my mining core, but I know other people have.03:08
GitHub72[migen] sbourdeauducq pushed 1 new commit to master: https://github.com/milkymist/migen/commit/e62ac1d3a1a2855ed241ac9f29d9103afdc0d2a010:57
GitHub72[migen/master] corelogic/misc: contiguous split - Sebastien Bourdeauducq10:57
lekernelwpwrak_: what's the next step for MIDI? a preprocessor so you #include the device descriptions?11:03
wpwrak_hehe :)13:20
wpwrak_naw, i think i'll split the database: one part goes into a separate file where only device descriptions are allowed, and maybe leave the option open to have declarations in a patch as well13:21
wpwrak_what's more important now will be the ability to detect what's actually connected13:22
wpwrak_and to associate events with the source13:22
lekernelshall we push out webupdate images as soon as this is done?13:23
lekernelthen for the next release (1) USB bugfixing (2) support for short movies13:24
wpwrak_it'll still be a while. my plan for now it to finish the existing feature set (i have a few small changes and i'll also rename a few identifiers, e.g., split "range" for devices into "fader" and "pot"), document it, and then let people have a look at it.13:25
wpwrak_then add multi-images, the other "major" language extension13:26
lekernelcan we release an update before image sequences?13:26
wpwrak_image sequences should take only a little time. probably less than a day13:27
lekernelmaybe, but release soon, release often13:28
lekernelMIDI-USB is hot13:28
wpwrak_sure. we should make more use of daily build, too13:28
wpwrak_hehe, true :)13:28
lekernelhow? have an option in the web update to install the daily build, with a "use at your own risk" warning?13:29
lekerneland btw daily builds should have the last commit id in the version numbers13:30
wpwrak_something like that. maybe just have the means to select from a list of install options. and that could include the daily builds. maybe just make the URL for a master list configurable.13:31
wpwrak_yes, we should track versions at a lot more places. build of libs, rtems, fn, soc, bios13:32
lekernelok. further reasons to get that MIDI-USB enabled update out soon *g*13:33
lekernelthen better version tracking + daily build support + USB bugfix + image sequences in the next13:34
lekerneland hopefully some RTEMS cleanup13:34
wpwrak_image sequence would be next. then the language should be more or less complete. then it needs documenting :)13:35
lekernelI'll submit the idea of adding the last git commit in their version string13:35
wpwrak_linux ? ;-)13:35
wpwrak_the ultimate rtems cleanup ;-)13:35
lekernelatm all builds from HEAD have "4.10.99" which doesn't makes much sense13:35
lekernelI think Linux will take ages13:36
lekernelespecially if we run into M1-crashing heisenbugs and performance issues, and we probably will13:37
wpwrak_i'd be worried a bit about performance issues, yes13:38
wpwrak_also the MMU may need some tuning before it works well13:38
GitHub192[migen] sbourdeauducq pushed 4 new commits to master: https://github.com/milkymist/migen/compare/e62ac1d...244bf1720:02
GitHub192[migen/master] corelogic/record: add to_signal convenience function - Sebastien Bourdeauducq20:02
GitHub192[migen/master] bus/asmibus: fix typo - Sebastien Bourdeauducq20:02
GitHub192[migen/master] corelogic/misc/multimux: less confusing variable name - Sebastien Bourdeauducq20:02
kristianpaulwhat about a mmu-less linux is that severe lacks mmu that it will break most of the software and drivers currently Flicker Noise need?21:43
Fallenouon mmu-less linux like uclinux any program can crash another program (or the kernel) AFAIK21:47
Fallenousince there is no memory protection21:47
Fallenou=> no segfault, you can write anywhere and crash everything21:47
Fallenouit's really a problem if you want to have a stable system21:47
kristianpaulyes i see now..21:50
Fallenougoing to sleep, will get up early tomorrow to work on the mmu :) would be nice to make memory *stores* to work too !22:19
Fallenougn8 !22:20
Fallenoufor those who want to have a look at the mmu project there is a ready-in-a-few-seconds simulation project here : https://github.com/fallen/milkymist-mmu-simulation22:20
Fallenouif you already have ISE installed you can simulate the lm32-mmu in less than a minute22:21
Fallenouwill document this a little bit more tomorrow :)22:21
dvdkFallenou: just some 15minutes ago I googled for your code (and found it)22:27
dvdkso nobody around who can share some insights on the MMU driver?23:02
dvdki'm just wondering what 'wcsr tlbvaddr,%0' refers to.  tlbvaddr?  that csr is not in the lm32 manual afaik23:02
dvdkdid you guys extend gnu as?23:03
kristianpauli remenber Fallenou modified toolchain23:08
dvdkso unnccessary... just use hexcodes.23:09
dvdkok, cloning his git repo, and looking at the verilog code to see which new CSR numbers he's using23:10
dvdkOMG, he's really done it: kristianpaul: do yo uknow e23:11
dvdkuh, copy-paste reror23:11
--- Sun Feb 12 201200:00

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