<azonenberg>
So i'm working on building a scopehal-testdata repo (not yet logging commits here) with various test waveforms for use in developing decodes, UI stuff, etc
<azonenberg>
they're kept small to fit within github size limits so generally i only have 1-2 waveforms of history while the original multi-gigabyte datasets on my cluster have tens or even hundreds of waveforms
<kc8apf>
Spell out IOActive on about me and IOA and glscopeclient slides
<azonenberg>
Ah that was a leftover from an earlier draft
<azonenberg>
i was trying to avoid a line break
<kc8apf>
copyright jumps around in position and is missing on some slides
<azonenberg>
That's part of the template, weird
<kc8apf>
I'm working on my DEFCON slides so I'm in the nitpicking mindset
<azonenberg>
Anyway, good catch. Copyrights fixed
<kc8apf>
release timeline having "now" as the first version feels awkward
<kc8apf>
maybe "prerelease" or "devel" or "git-main"?
<azonenberg>
yeah prerelease sounds better
<bvernoux>
azonenberg, slide 50 is fun
<bvernoux>
you have received contribution from Pico, Digilent Siglent and IOACtive ;)
<bvernoux>
oups no IOActive have not provided scope ;)
<bvernoux>
it is a joke because it is like it is a vendor
<azonenberg>
Yeah no scope from them. But they're letting me take time away from work to work on the project
<azonenberg>
And i use it on projects for them pretty often
<bvernoux>
yes of course
<bvernoux>
one strange things is when you call them IOA
<bvernoux>
page 5
<bvernoux>
not IOA product
<azonenberg>
It's a pretty common abbreviation internally and in the industry, I spelled it out fully in the slides already
<bvernoux>
will be more clear to say not IOActive product
<azonenberg>
just to be extra clear
<azonenberg>
you're looking at a draft more than 20 minutes old :p
<bvernoux>
ha ok it is already fixed ?
<azonenberg>
Yeah just not reuploaded
<bvernoux>
else it is very good
<bvernoux>
do you plan to show a live demo also ?
<azonenberg>
Yes
<bvernoux>
ha great
<azonenberg>
Overall format of the talk is this slide deck in the first half hour (52 slides but several are titles, and i plan to skip over the list of supported hardware pretty quickly)
<bvernoux>
ha yes slide 2 you say you are doing 30 minutes demo
<azonenberg>
then the second half of the talk will be a quick live streaming demo followed by an interactive workshop going through some pre-recorded data files while the audience follows along
<azonenberg>
The organizers booked me as the last slot of the day
<azonenberg>
So we can run as late as people want to keep seeing stuff
<bvernoux>
IOACtive are very nice to let you speak about this project on your work time
<azonenberg>
the official cutoff is after 30 mins
<azonenberg>
It's a good PR opportunity for them
<bvernoux>
Yes I imagine it is good for them too
<azonenberg>
And they're generally supportive of me working on the project, like i said i use it internally on work projects
<bvernoux>
yes it is a win-win game
<azonenberg>
i wrote several decodes on company time when i encountered a protocol we didn't have any tooling for
<azonenberg>
then upstreamed them
<bvernoux>
I was not aware IOActive work on DDR3 stuff ...
<azonenberg>
I didn't write that for work
<bvernoux>
I was imagining they are more focused on SW stuff
<azonenberg>
But the SPI flash decoder i wrote for analyzing a secure boot implementation
<azonenberg>
looking at exactly what addresses it reads when
<bvernoux>
ha ok
<azonenberg>
Most of the high speed stuff, all of the ethernet and pcie, etc i wrote on my own
<azonenberg>
i wrote the spectrogram decoder at work too, it was very helpful for some power analysis stuff
<bvernoux>
You can add for MSO5000 that R&D plan to release a new firmware to break the O(1) ;)
<azonenberg>
Being able to correlate frequency domain activity in a power trace to time domain bus activity
<bvernoux>
That will put a bit more pressure on Rigol ;)
<azonenberg>
so for example if you see a burst of SPI traffic then some activity in the power trace right after
<azonenberg>
that's probably decryption or signature checking
<azonenberg>
i have some really awesome recordings of exactly that but they're from client hardware and i can't share
<azonenberg>
unfortunately i dont have any demos of that from open or non-confidential sources right now
<azonenberg>
but i'll talk about it in general terms
<bvernoux>
in latest discussion with the support they told me they will deliver something for end of July ;)
<bvernoux>
Yes so far we have no any proof they are working on it except some Email exchanged with their support
<azonenberg>
And yeah if i remember i'll mention that in the performance discussion
<kc8apf>
fwiw: I think QEMU/KVM w/ virtio-gpu-gl can do OpenGL 4.3 w/ compute shaders
<bvernoux>
I will ask some news to know if they can provide a beta version to check if the performance are better
<azonenberg>
kc8apf: Interesting. Would be interested in seeing test results from that
<azonenberg>
to date, i have not heard any reports of anyone getting glscopeclient to run correctly under a hypervisor
<azonenberg>
so if we can get a known working config that would be nice
<bvernoux>
kc8apf, ha yes will be really interesting
<azonenberg>
bvernoux: I tweeted a link to the con invite earlier today and tagged a whole bunch of scope vendors
<kc8apf>
virtio-gpu-gl is a paravirtualized GPU build on Mesa
<azonenberg>
Pico has already told me they plan to attend my talk
<bvernoux>
azonenberg, Yes I have seen that and retweeted it too very nice
<bvernoux>
azonenberg, I hope vendors will react ...
<bvernoux>
azonenberg, Do you have some contact with Keysight ?
<bvernoux>
They could provide you a big scope for free
<azonenberg>
Dan bogdanoff liked the tweet and we've talked about the project a bunch
<azonenberg>
He's been trying to get me a loaner scope for dev and testing but it hasnt happened yet
<azonenberg>
I don't know if he plans to watch the talk or not
<bvernoux>
azonenberg, maybe that will change after your presentation
<azonenberg>
I get the impression he personally likes the project but it's been an uphill battle to get management to be supportive too
<bvernoux>
Also Tek could provide you a MSO6 to finish your tests ...
<bvernoux>
It was planned IIRC
<kc8apf>
for Pico, what does 75% complete mean?
<azonenberg>
Yes we've already talked about me testing some new firmware from them with a loaner scope
<bvernoux>
kc8apf, it means it basically work ;)
<azonenberg>
kc8apf: status of how finished the driver is. Lots of stuff like MSO channels are still missing last i checked
<azonenberg>
but it's somewhat usable
<kc8apf>
kinda confusing as you explicitly list what does/doesn't work for other devices
<azonenberg>
then xzc... can never spell his name right
<azonenberg>
is working on 5000 series but hasnt merged anything yet
<azonenberg>
And yeah i might reformat that a bit
<bvernoux>
azonenberg, are you planing to do the CTS ?
<bvernoux>
it is very new things very special ;)
<bvernoux>
mainly doing demodulation on unknown signal
<azonenberg>
No, I'm busy on billable stuff. Didnt get around to asking for time off to attend the whole con
<bvernoux>
it is only related to digital demodulation anyway
<azonenberg>
so i'm taking an hour or two out of my day to speak then back
<bvernoux>
it is virtual
<bvernoux>
with signal streamed over network
<bvernoux>
the hard part is to use old gnu radio 3.7
<bvernoux>
to retrieve the signals
<bvernoux>
I have built SigDigger v0.2 developer version to check if lot of things have improved ;)
<miek>
i was just looking at that as i can't be bothered installing gr3.7 again :p
<bvernoux>
miek, I'm using their VM that I have customized
<miek>
it looks like you just need to connect a ZMQ SUB source to the server IP, and the port is the frequency in MHz
<bvernoux>
yes
<bvernoux>
the idea is to capture the data over network to a file then use more modern tool on the iq file ;)
<bvernoux>
miek, I was training with Inspectrum but I do not find how to configure the thresholds
<bvernoux>
miek, to saturate amplitude ...
<bvernoux>
I have clearly missed something to the last step as the export of binary corresponding to threshold does not provide what I see
<miek>
you can adjust the amplitude plot with the power max slider
<bvernoux>
miek, ha it is the hint I'm searching ;)
<bvernoux>
ha yes the power max was the hint thanks
<bvernoux>
I was not thinking the threshold was linked to that
someone--else has joined #scopehal
<azonenberg>
bvernoux: so any other suggestions on the slides or you think it's good?
<bvernoux>
azonenberg, for me all is good
<electronic_eel>
azonenberg: i also had a peek over the slides. the "IOA" abbreviation was uncommon to me, but others have already commented on it
<electronic_eel>
if i understood it correctly, you will add some live demo. then i think it will work well
<azonenberg>
Yeah i fixed it. There will be 30ish mins of this slide deck
<azonenberg>
followed by 30 mins of me demoing stuff while the audience follows along
<azonenberg>
i linked the repo and the test data repo in the conference discord already
<electronic_eel>
"audience follows along" meaning they look at your screencast or do they install it themselves on their systems and play with it themselves?
<azonenberg>
The latter
<azonenberg>
although they can of course watch as i demo too
<electronic_eel>
using it themselves will be much better than just the screencast
<electronic_eel>
do you have a page with links to the packages for debian by GyrosGeier and my Fedora / RHEL ones?
<electronic_eel>
so that the users can easily install it
<azonenberg>
Not currently
<electronic_eel>
or how do you plan for them to run it?
<azonenberg>
I just linked the repo
<electronic_eel>
virtual machines isn't easy because of the opengl
<azonenberg>
I dont think GyrosGeier's testing packages are fully up to date
<azonenberg>
i think he was just testing on one snapshot
<electronic_eel>
i can update the fedora packages if you think it helps. they are about 2-3 weeks old
<azonenberg>
Sure, go for it
<electronic_eel>
my experience with classroom setups is that if you don't provide preconfiguried gear for everyone, you'll waste a lot of time just getting the basics running for everyone
<azonenberg>
Yeah thats why i'm linking the repo etc today
<azonenberg>
and the talk isnt until tomorrow
<electronic_eel>
and even if you provide preconfigured & tested gear, always someone will manage to mess it up
<Degi>
Hm, maybe give precompiled binaries
<azonenberg>
GyrosGeier: can you update your debian packages?
<electronic_eel>
maybe the people attending my training sessions aren't that skilled, but having taken part in others' training sessions i don't think i'm an outlier there
<azonenberg>
no i agree
<azonenberg>
I had *hoped* to have a full release by this point
<azonenberg>
But didnt quite reach that point
<electronic_eel>
no problem. i just think not everyone will be into compiling this themselves.
<_whitenotifier-a>
[scopehal-docs] doegox opened issue #33: Add install prefix to ffts build command - https://git.io/Jc5JW
<electronic_eel>
seems like some course participants are already compiling