azonenberg changed the topic of #scopehal to: libscopehal, libscopeprotocols, and glscopeclient development and testing | https://github.com/glscopeclient/scopehal-apps | Logs: https://libera.irclog.whitequark.org/scopehal
Degi_ has joined #scopehal
Degi has quit [Ping timeout: 252 seconds]
Degi_ is now known as Degi
<_whitenotifier-9> [scopehal-apps] azonenberg pushed 1 commit to master [+4/-0/±8] https://github.com/glscopeclient/scopehal-apps/compare/a4a261f8c932...3ded92d60e3f
<_whitenotifier-9> [scopehal-apps] azonenberg 3ded92d - Initial trigger integration with filter graph editor
<d1b2> <azonenberg> Playing with trigger configuration in the filter graph editor
<_whitenotifier-9> [scopehal-apps] azonenberg pushed 1 commit to master [+0/-0/±2] https://github.com/glscopeclient/scopehal-apps/compare/3ded92d60e3f...28371994e7d9
<_whitenotifier-9> [scopehal-apps] azonenberg 2837199 - Filter graph editor now allows dragging from a block's input to empty space, enabling new channels in the process. See #161.
Yamakaja has quit [*.net *.split]
florolf has quit [*.net *.split]
elms has quit [*.net *.split]
mxshift has quit [*.net *.split]
elms has joined #scopehal
mxshift has joined #scopehal
Yamakaja has joined #scopehal
Yamakaja has quit [Changing host]
Yamakaja has joined #scopehal
florolf has joined #scopehal
<t4nk_freenode> johnsel, are you still around?
<t4nk_freenode> meanwhile I've gotten myself an fx3 dev board ;)
<t4nk_freenode> had quite a bit of chagrin, it doesn't play nice with my usb ports, but somehow it's working stably at the moment
<t4nk_freenode> and I like it a lot, throughput is 'awesome' from what I was used to
<d1b2> <johnsel> I sure am, but I have an important health call in 45mins. We can chat after though
<t4nk_freenode> k man, I think I'll be around for a couple of hours yet, don't let me keep you from important stuff
kbeckman1 has joined #scopehal
gurki_ has quit [*.net *.split]
tnt has quit [*.net *.split]
azonenberg has quit [*.net *.split]
kbeckmann has quit [*.net *.split]
Ekho has quit [*.net *.split]
gurki_ has joined #scopehal
tnt has joined #scopehal
Ekho- has joined #scopehal
azonenberg has joined #scopehal
<d1b2> <johnsel> Back
<d1b2> <johnsel> t4nk_freenode: if you’re still here
<t4nk_freenode> hi, yeah, I am ;)
<t4nk_freenode> I was just trying to make sense of crc calculation :|
<t4nk_freenode> I have this recording device that has an sd card slot
<t4nk_freenode> I've set things up so I can attach a logic analyzer
<t4nk_freenode> I'd like to emulate an sd card so that I can store songs straight to my hd via usb ;)
<t4nk_freenode> weren't you the one who mentioned the rpi pico to me too? I've also got a couple of those now
<t4nk_freenode> there's only about 150k of data on a freshly formatted sd for the device with an empty song, so I was hoping to be able to get that working from a pico alone
<t4nk_freenode> and afterwards routing it to my pc via an fx2 board
<d1b2> <johnsel> I very well might have, I probably said it has a similar feature to the TI AM335x that made the beaglelogic possible, the PRU
<t4nk_freenode> yeah
<t4nk_freenode> I found that very interesting ;)
<d1b2> <johnsel> it is quite interesting as it gives the little bugger some quite high speed capability compared to it's cores and certainly compared to the costs
<t4nk_freenode> been using sig_rok to capture some data. while it's nice to get somewhat of an overview, I find it very frustrating to do something useful with it I must say
<d1b2> <johnsel> yes sigrok is quite annoying to work with
<d1b2> <johnsel> that's why I ended up here in first place 🙂
<t4nk_freenode> yeah, stm32f411 has 100Mhz i/o, but very little flash and mem
<t4nk_freenode> the protocol decoder gives a good idea of the commands being sent, but the data is completely invisible as is
<t4nk_freenode> and it seems impossible to save it, or save some selections
bvernoux has joined #scopehal
<d1b2> <johnsel> you've picked yourself quite a difficult task then
<t4nk_freenode> heh, and I don't know if it's because I'm capturing a lot of data... but everything is slow like mad
<t4nk_freenode> sd decoder is bad, but adding timing makes it unusable really
<t4nk_freenode> yeah, I don't know how things will pan out, but I'm pretty hopeful about it
<t4nk_freenode> lol, but I ran into this crc7-thing which I don't yet fully understand :)
<t4nk_freenode> still reading up though
<d1b2> <johnsel> why not use scopehal instead?
<d1b2> <johnsel> it has SD data decoders
<t4nk_freenode> well, all I have to work with is some fx2 boards, and the fx3, I don't think any of that has yet been implemented
<d1b2> <johnsel> is it not doable for yourself?
<t4nk_freenode> signal clock runs @16Mhz I think, so fx2 isn't really comfortable to use
<t4nk_freenode> that would be a whole nother project I think, lol, and I'm facing limited mental capabilities as is :))
<d1b2> <johnsel> hmm, well do you know the clock? A quick google says it can be up to 25mhz
<t4nk_freenode> yeah, I think it runs @16Mhz, and the fx2 is supposed to go up to 24MHz, but I wouldn't count on that flying entirely
<t4nk_freenode> and you'd want some more overhead for comfort too
<t4nk_freenode> the fx la I have as a product is really badly constructed, it will crash to the touch
<t4nk_freenode> I have some dev boards I could connect, they're a bit better
<t4nk_freenode> but I don't even see a way to easily compare captured data from fx2 and fx3 with SR
<d1b2> <20goto10> libscopeexports.{so,dll} is missing from the GitHub Actions export on Windows and Linux (haven't tried OSX). I can add an issue later and a PR if I get time
<azonenberg> t4nk_freenode: we have two decodes, one for SD command and another for data
<azonenberg> command only looks at the command line, data looks at the whole parallel bus
<azonenberg> (in scopehal)
<azonenberg> 20goto10: This is only a short term issue as scopeexports is a component of glscopeclient which is scheduled to be discontinued
<azonenberg> that same functionality is going to get folded into ngscopeclient in the near future
<azonenberg> Limited dev resources have made this take longer than planned
<azonenberg> There's also a few... not exactly bugs, but awkward UI quirks that can easily lead to incorrect exports... in the CSV exporter now so i wanted to rewrite it anyway
<t4nk_freenode> ([ 59%] Building CXX object lib/scopeprotocols/CMakeFiles/scopeprotocols.dir/SDCmdDecoder.cpp.o)
<t4nk_freenode> :b
<t4nk_freenode> ngscopeclient just crashed my pc ;)
<t4nk_freenode> after rebooting, ngscopeclient will segfault immediately; it seems there's probable still the old issues with VulkanWindow.cpp, see if I can find them
<azonenberg> t4nk_freenode: if your whole system crashed before it even got to the point of displaying a waveform, we can rule out any bugs in the rendering shaders
<azonenberg> i'd suspect gpu driver bugs at that point
<azonenberg> are you running recent drivers?
<t4nk_freenode> I guess so, amdgpu, kernel 6.1.11
<t4nk_freenode> (but there may have been some usb stuff causing the crash)
<azonenberg> ah ok
<t4nk_freenode> after the reboot there's a segfault immediately
<azonenberg> well, let me know what you find. I strongly recommend having the vulkan validation layers active when tracking down possible vulkan issues
<azonenberg> it's extremely helpful
<t4nk_freenode> azonenberg, apart from that.... should I get it to run.. could I import a csv file that's been exported from pulseview?
<t4nk_freenode> I'm talking 3G in size
<azonenberg> You can import CSVs, it expects a single header row w/ channel names names
<azonenberg> then leftmost column as time in seconds
<azonenberg> then values in subsequent columns
<azonenberg> i've imported gigabyte sized CSVs in the past without issue, although it's slow
<t4nk_freenode> if only I could edit the capture in PV, then I could cut out a whole bunch of unnecessary 'data'
<azonenberg> the hard cap right now is 4GB per memory allocation in Vulkan, which means max waveform size is 1G points for uniformly sampled float32 or 4G for uniformly sampled digital, or 512M points for sparse sampling
<t4nk_freenode> heh, yeah, the exported csv has
<t4nk_freenode> logic,logic,logic,logic,logic,logic
<t4nk_freenode> 0,0,0,0,0,0
<t4nk_freenode> so no time index, clock signal might also change frequency at some point
<azonenberg> yeah you might have to preprocess it in python or somethign to generate a time column then
<azonenberg> good news is, if you're doing that it's trivial to skip the first N lines or something to shrink the output file
<t4nk_freenode> yeah, I was already thinking of doing that indeed
<t4nk_freenode> but over my cold dead body if it had to be python :)
<azonenberg> Lol
<azonenberg> you can use your language of choice
<azonenberg> do it in awk for all i care
<_whitenotifier-9> [scopehal] azonenberg opened issue #752: CSV import: allow data-only files with no timestamp column - https://github.com/glscopeclient/scopehal/issues/752
<_whitenotifier-9> [scopehal] azonenberg labeled issue #752: CSV import: allow data-only files with no timestamp column - https://github.com/glscopeclient/scopehal/issues/752
<azonenberg> t4nk_freenode: see above wrt future imporvement, but i'm not working on this right away
<t4nk_freenode> yeah, nice, can't do everything all at once though ;)
<t4nk_freenode> need to get some sleep myself first, see about things later
<t4nk_freenode> thnx
<t4nk_freenode> (was getting "ngscopeclient: /usr/include/vulkan/vulkan_raii.hpp:3356: const vk::raii::DeviceDispatcher* vk::raii::Device::getDispatcher() const: Assertion `m_dispatcher->getVkHeaderVersion() == VK_HEADER_VERSION' failed. Aborted" btw)
<t4nk_freenode> my vulkan headers are 1.3.236
<azonenberg> So that error seems to suggest the vulkan headers and library source files are mismatched
<azonenberg> which sounds like a broken SDK install?
<t4nk_freenode> is this runtime or compiletime?
<azonenberg> Compile time
<azonenberg> basically part of the code is using one vulkan header version and another is using a different one and it's complaining
<azonenberg> maybe try a make clean and rebuild in case you have cached files from an older sdk or something? not sure
<t4nk_freenode> mmm, I started with a new clone
<t4nk_freenode> however, I'm on gentoo and I didn't have to do much about getting the sdk I believe
<t4nk_freenode> it worked last time I built it
<t4nk_freenode> a decent while back now
<azonenberg> I'm not sure then. but the vulkan raii layer etc is all just compile time header wrappers around the c api
<azonenberg> so if it's choking that means part of the library was compiled with an older version than the rest
<t4nk_freenode> argh, crashed again, so the other remarks were rubbish
<t4nk_freenode> bedtime.
Ekho- is now known as Ekho
<_whitenotifier-9> [scopehal-apps] azonenberg pushed 1 commit to master [+0/-0/±1] https://github.com/glscopeclient/scopehal-apps/compare/28371994e7d9...4d0cd050e566
<_whitenotifier-9> [scopehal-apps] azonenberg 4d0cd05 - FilterGraphEditor: fixed crash if one input is null
bvernoux has quit [Quit: Leaving]