whitequark changed the topic of #glasgow to: digital interface explorer · code https://github.com/GlasgowEmbedded/glasgow · logs https://libera.irclog.whitequark.org/glasgow · discord https://1bitsquared.com/pages/chat · production https://www.crowdsupply.com/1bitsquared/glasgow (FUNDED)
geekcowboy_ has quit [Ping timeout: 248 seconds]
geekcowboy_ has joined #glasgow
geekcowboy_ has quit [Ping timeout: 258 seconds]
geekcowboy_ has joined #glasgow
GNUmoon2 has joined #glasgow
egg|anbo|egg_ has quit [Remote host closed the connection]
geekcowboy_ has quit [Ping timeout: 248 seconds]
geekcowboy_ has joined #glasgow
geekcowboy_ has quit [Ping timeout: 268 seconds]
geekcowboy has joined #glasgow
geekcowboy has quit [Ping timeout: 258 seconds]
geekcowboy has joined #glasgow
GNUmoon2 has quit [Ping timeout: 244 seconds]
jstein has joined #glasgow
geekcowboy has quit [Ping timeout: 248 seconds]
GNUmoon2 has joined #glasgow
geekcowboy has joined #glasgow
geekcowboy_ has joined #glasgow
geekcowboy has quit [Ping timeout: 248 seconds]
jstein has quit [Ping timeout: 240 seconds]
geekcowboy_ has quit [Ping timeout: 248 seconds]
jstein has joined #glasgow
geekcowboy_ has joined #glasgow
egg|anbo|egg_ has joined #glasgow
geekcowboy_ has quit [Ping timeout: 268 seconds]
geekcowboy_ has joined #glasgow
jstein has quit [Remote host closed the connection]
egg|anbo|egg_ has quit [Remote host closed the connection]
geekcowboy_ has quit [Ping timeout: 258 seconds]
geekcowboy_ has joined #glasgow
geekcowboy_ has quit [Ping timeout: 240 seconds]
geekcowboy_ has joined #glasgow
egg|anbo|egg_ has joined #glasgow
_whitenotifier-d has joined #glasgow
<_whitenotifier-d> [glasgow] mndza synchronize pull request #265: Add bitarray implementation to eliminate external dependency - https://git.io/JtZfr
egg|anbo|egg_ has quit [Remote host closed the connection]
egg|anbo|egg_ has joined #glasgow
egg|anbo|egg__ has joined #glasgow
egg|anbo|egg has quit [Ping timeout: 268 seconds]
GNUmoon2 has quit [Remote host closed the connection]
GNUmoon2 has joined #glasgow
ali_as has quit [Remote host closed the connection]
ali_as has joined #glasgow
egg|anbo|egg_ has quit [Remote host closed the connection]
egg|anbo|egg has joined #glasgow
redstarcomrade has joined #glasgow
geekcowboy has joined #glasgow
geekcowboy_ has quit [Ping timeout: 245 seconds]
geekcowboy_ has joined #glasgow
geekcowboy has quit [Ping timeout: 240 seconds]
geekcowboy has joined #glasgow
geekcowboy__ has joined #glasgow
geekcowboy_ has quit [Ping timeout: 252 seconds]
geekcowboy_ has joined #glasgow
geekcowboy has quit [Ping timeout: 240 seconds]
geekcowboy__ has quit [Ping timeout: 252 seconds]
GNUmoon2 has quit [Remote host closed the connection]
GNUmoon2 has joined #glasgow
geekcowboy has joined #glasgow
geekcowboy__ has joined #glasgow
geekcowboy_ has quit [Ping timeout: 252 seconds]
geekcowboy_ has joined #glasgow
geekcowboy has quit [Ping timeout: 252 seconds]
geekcowboy has joined #glasgow
geekcowboy__ has quit [Read error: Connection reset by peer]
geekcowboy_ has quit [Ping timeout: 252 seconds]
geekcowboy_ has joined #glasgow
geekcowboy has quit [Ping timeout: 240 seconds]
geekcowboy__ has joined #glasgow
geekcowboy_ has quit [Ping timeout: 252 seconds]
egg|anbo|egg has quit [Remote host closed the connection]
geekcowboy__ has quit [Ping timeout: 252 seconds]
geekcowboy__ has joined #glasgow
geekcowboy__ has quit [Ping timeout: 252 seconds]
geekcowboy__ has joined #glasgow
jstein has joined #glasgow
redstarcomrade has quit [Quit: Connection closed for inactivity]
geekcowboy__ has quit [Ping timeout: 252 seconds]
geekcowboy__ has joined #glasgow
GNUmoon2 has quit [Ping timeout: 244 seconds]
geekcowboy__ has quit [Ping timeout: 252 seconds]
<kbeckmann> is there a way to lower the sample rate when running `glasgow run analyzer`? I keep getting FIFO overrun even when only sampling one pin. I'm on an old RevB.
geekcowboy__ has joined #glasgow
<kbeckmann> `glasgow run benchmark` gives me around 28 MB/s on all ports of my PC. is that good or bad? I have a vague memory of getting a bit faster transfers than that.
<d1b2> <Attie> do you have anything else on the bus? (e.g: webcam) and do you have a root port available?
<d1b2> <Attie> i'm not sure about revB, but revC certainly should exceed ~40MiB/s
<electronic_eel> kbeckmann: 28 MB/s is a bit low, see for example here for my results: https://github.com/GlasgowEmbedded/glasgow/issues/220#issuecomment-752655318
<d1b2> <Attie> (... and I'm not aware of a reason for that to be different)
<kbeckmann> i connected it directly to the mainboard so that should be fine. i also tried a bunch of different ports.
<kbeckmann> oh yeah that is definitely faster than my results.
<d1b2> <Attie> those ports can still share bandwidth with others
<electronic_eel> could also be a python or libusb issue
<kbeckmann> i ran it in a loop and disconnected everything including my keyboard. still getting 28 MB/s
<kbeckmann> yeah perhaps. i'm on archlinux, upgraded recently. running latest main branch of glasgow.
<d1b2> <Attie> hmm... that's frustrating - thanks for trying though
<kbeckmann> could be the cable i suppose
<kbeckmann> let me try with a few different ones
<electronic_eel> maybe try booting the system with a live linux on usb memory stick and use a different distro. to rule out kernel, python, libusb and so on
<electronic_eel> but i'd test the cables first...
<whitequark> kbeckmann: no, because the analyzer does not stream samples
<whitequark> it uses compression
<kbeckmann> ah, i see, thanks.
<kbeckmann> i've now tried 3 different cables in various ports, still getting 28 MB/s. i guess i should try other host machines.
<whitequark> kbeckmann: why do you believe this is a bandwidth issue?
<whitequark> it's much more likely a latency spike issue
<whitequark> Glasgow does not have a lot of on-board FIFO memory, so any latency spikes tend to crash the analyzer
<whitequark> wait
<whitequark> please disregard everything i said
<whitequark> you're using `glasgow run analyzer`, not `glasgow run --trace`
<kbeckmann> yes that's right
<whitequark> i'm sorry; i misremembered
<whitequark> the analyzer applet does stream samples of course
<kbeckmann> (my reasoning was mostly more MB/s is better)
<whitequark> or, hm
<whitequark> i forgot more about how it works than i thought i have
<whitequark> what do you have connected, electrically?
<kbeckmann> Nintendo 64 audio dac data
<whitequark> are any of the pins floating?
<kbeckmann> so it shouldn't be that much bandwidth
<whitequark> try `glasgow run analyzer --pull-downs`
<kbeckmann> yes but not the ones i chose to sample
<kbeckmann> i am running `glasgow run analyzer --port A -V 3.3 --pins-i 0 dump.vcd`
<whitequark> oh, okay
<whitequark> so just one pin, and it's driven by the DUT
<kbeckmann> i am on a RevB which doesn't have pulls
<electronic_eel> is the electrical connection sound or could it be that the cables wiggle and sometimes disconnect?
<whitequark> can you show me the partial trace it writes before overflow?
<kbeckmann> i used a saleae logic with the same cables and got good data
<kbeckmann> sure one second
<kbeckmann> whitequark: https://allg.one/QxlO.vcd
<whitequark> wtf
<kbeckmann> here are four pins being sampled https://allg.one/OKLE.vcd
<whitequark> something's broken in the analyzer
<kbeckmann> is there a way for me to 100% confirm that i am running the correct firmware?
<kbeckmann> i am on an old RevB just so there is no confusion.
<whitequark> there is a check that the firmware is compatible, but you can also do `glasgow flash --remove-firmware`
<whitequark> and re-plug the device
<whitequark> revB should work fine for this
<kbeckmann> thanks, i have now done this
<kbeckmann> it behaves the same still
<whitequark> as expected
<kbeckmann> yeah, i just remember last time i was here asking about weirdness it was because i was stuck on an old firmware.
<electronic_eel> i once had the issue that my glasgow git tree was on an old abandoned branch and it built buggy gateware
<whitequark> kbeckmann uses main, so shouldn't be that
<kbeckmann> i will make a fresh clone just to be sure.
<whitequark> oh wait
<whitequark> okay the analyzer applet is really broken
<whitequark> this could be a workaround
<kbeckmann> oh, thanks! will try
<kbeckmann> i got 60ms of data this time, but still got the fifo overrun
<kbeckmann> it's very consistent. every time i run it, i get the same amount of data
<whitequark> okay, i'm far too exhausted to dig out the memory of how that applet worked and fix it
<whitequark> i suggest you write a new one for your use case specifically, sampling at a low rate
<kbeckmann> no worries at all!
<whitequark> something more analyzer-like and not... whatever this applet is doing
<kbeckmann> thanks for the help!
redstarcomrade has joined #glasgow