azonenberg changed the topic of #scopehal to: ngscopeclient, libscopehal, and libscopeprotocols development and testing | https://github.com/ngscopeclient/scopehal-apps | Logs: https://libera.irclog.whitequark.org/scopehal
sgstair_ has joined #scopehal
_sgstair has joined #scopehal
sgstair has quit [Ping timeout: 260 seconds]
sgstair_ has quit [Ping timeout: 248 seconds]
<_whitenotifier-4> [scopehal] azonenberg pushed 3 commits to master [+2/-0/±10] https://github.com/ngscopeclient/scopehal/compare/e0200ad48847...c219b4fd062f
<_whitenotifier-4> [scopehal] azonenberg a8c998c - Initial GPU acceleration of ACRMS core. See #971.
<_whitenotifier-4> [scopehal] azonenberg 05de059 - Preallocate output buffer
<_whitenotifier-4> [scopehal] azonenberg c219b4f - ACRMS filter: finished GPU path (if GPU_shader_int64 is available). Fixes #971.
<_whitenotifier-4> [scopehal] azonenberg closed issue #971: ACRMS filter execution time is high - https://github.com/ngscopeclient/scopehal/issues/971
<_whitenotifier-4> [scopehal-apps] azonenberg pushed 1 commit to master [+0/-0/±2] https://github.com/ngscopeclient/scopehal-apps/compare/d911f2aed957...254e1cfe476e
<_whitenotifier-4> [scopehal-apps] azonenberg 254e1cf - AC RMS: updated unit test to check a few more things, fixed bug in reference implementation
<_whitenotifier-4> [scopehal] azonenberg pushed 1 commit to master [+0/-0/±7] https://github.com/ngscopeclient/scopehal/compare/c219b4fd062f...b23cfb9e8602
<_whitenotifier-4> [scopehal] azonenberg b23cfb9 - Fixed a couple of off-by-ones in AC RMS and reduction-sum shaders
<_whitenotifier-4> [scopehal-apps] azonenberg pushed 1 commit to master [+0/-0/±1] https://github.com/ngscopeclient/scopehal-apps/compare/254e1cfe476e...0aed45b265ed
<_whitenotifier-4> [scopehal-apps] azonenberg 0aed45b - Updated submodules
<_whitenotifier-4> [scopehal] azonenberg pushed 1 commit to master [+0/-0/±1] https://github.com/ngscopeclient/scopehal/compare/b23cfb9e8602...74065f6a4f93
<_whitenotifier-4> [scopehal] azonenberg 74065f6 - Added environment variable for forcing vulkan device during debugging
<_whitenotifier-4> [scopehal-apps] AleksaBjelogrlic opened issue #844: Windows Portable Cannot Scale Window - https://github.com/ngscopeclient/scopehal-apps/issues/844
<_whitenotifier-4> [scopehal-apps] AleksaBjelogrlic edited issue #844: Windows Portable Cannot Scale Window - https://github.com/ngscopeclient/scopehal-apps/issues/844
<dingwat> azonenberg: minor stumbling block while following the GettingStarted for Fedora, I think libhidapi-dev should be "hidapi-devel" on Fedora
<azonenberg> dingwat: send PR to scopehal-docs?
<dingwat> azonenberg: roger. I wasn't sure if that was correct but now the build has finished so I think that is an accurate description. No other issues with the Fedora build, but pretty rough behaviour with ngscopeclient on this high-DPI laptop (Fedora 42 with GNOME), GUI is too large and the mouse position doesn't match the cursor
<dingwat> Will investigate further
<azonenberg> yeah there are various dpi bugs we need to poke at
<azonenberg> every time i think we got them all another one pops up loil
<dingwat> azonenberg: unsure where things are at with Tektronix HSI, but I have an MSO58B in the lab, so if there's anything I can test, please let me know. I tried connecting (driver: tektronix.hsi, transport: lan) and ngscopeclient crashes with a "Persistent bad IDN response, giving up". Not sure if that's my fault or the scope's fault...
<dingwat> (and I do have "High Speed Interface" enabled, on the default port of 5000)
<d1b2> <azonenberg> So the tek HSI driver needs a bridge server, i'll have to check where it is. it can't talk directly to the grpc
<d1b2> <azonenberg> we never got it fully working
<dingwat> ah ok
<d1b2> <azonenberg> the tek firmware is super unstable and HSI just gives you slightly faster waveform download
<d1b2> <azonenberg> it still uses scpi for the control plane
<azonenberg> And their SCPI is super unstable
<dingwat> :(
<azonenberg> the scope kept crashing on me and i basically lost interest
<dingwat> well, when our support guy comes by I'll complain
<azonenberg> if somebody wants to put more time into it, i'm all ears
<dingwat> I'll experiment more with the regular driver and see how unstable it is
<azonenberg> get that working first and only then think about HSI
<dingwat> roger dodger
<azonenberg> because if it's not stable - and it probably won't be - HSI won't improve things
<azonenberg> any time you trigger an error code it would lock up for several seconds, not send any response to the command you had sent
<azonenberg> then desync the request/response stream
<azonenberg> the stack is stateful
<azonenberg> if you crash or the firwmare derps at some point
<azonenberg> you can send a request and get a reply from a previous command sent by a different socket
<azonenberg> it preserves stack state across connect/disconnect cycles its nuts
<dingwat> ugh
<dingwat> well, I'm happy to complain lol
<dingwat> I already have a list
<azonenberg> lol
<azonenberg> i mean, if you want to take over maintainership of the tek driver
<dingwat> can't get remote web view to show up for some reason
<azonenberg> nobody is actively working on it right now
<azonenberg> i got sick of it, and tek wanted the demo scope back
<dingwat> I'll see if I can contribute, but I worry my coding skills aren't up to snuff