<Bird|otherbox>
you think I should just suck that in for #140?
<azonenberg>
Bird|otherbox: let me see
Degi has quit [Ping timeout: 260 seconds]
<azonenberg>
It's a compatible license, 2-clause BSD
<azonenberg>
test it, if it works properly then i say go for it
<azonenberg>
it hasnt been updated since 2013 so its possible there has been some bitrot
Degi has joined #scopehal
loxodes has quit [Ping timeout: 256 seconds]
loxodes has joined #scopehal
Kliment has quit [Ping timeout: 256 seconds]
Kliment has joined #scopehal
<azonenberg>
monochroma, lain: So i'm trying to figure out how to integrate the sd card data and command bus decodes
<azonenberg>
Command bus makes sense to be separate, since you might only care about a command trace and not the actual data being accessed
<azonenberg>
Is there any reason you would want to look at only the data bus without commands?
<azonenberg>
basically, i'm thinking of having the data bus decoder take clock, dat[3:0], and a decoded command bus as inputs
<azonenberg>
So i can then display a protocol analyzer view that shows read/write commands from the command bus (ignoring anything that doesn't trigger a data bus event)
<azonenberg>
aligned with the data
<lain>
I can't think of a reason I would want data without command
<azonenberg>
does that sound like a reasonable approach to the decode, btw?
<azonenberg>
the other unsolved problem is how to present CRCs given that SD is derpy and CRC16s each of the DAT lines separately
azonenberg_work has quit [Ping timeout: 260 seconds]
<_whitenotifier-f>
[scopehal-apps] tarunik created branch fix-yaml - https://git.io/JvExD
<_whitenotifier-f>
[scopehal-apps] tarunik commented on issue #140: cmake doesn't check for presence of yaml-cpp - https://git.io/JTp7o
<_whitenotifier-f>
[scopehal-apps] tarunik opened pull request #279: Add detection for yaml-cpp, fixes #140 - https://git.io/JTp76
<Bird|otherbox>
btw: I have another commit pending for #279
electronic_eel has quit [Ping timeout: 260 seconds]
electronic_eel has joined #scopehal
<azonenberg>
Well that was nice
<_whitenotifier-f>
[scopehal] azonenberg pushed 4 commits to master [+0/-0/±6] https://git.io/JTppl
<_whitenotifier-f>
[scopehal] azonenberg 8036035 - Initial packet support for SDDataDecoder. See #169.
<azonenberg>
35% speedup of eye patterns
<_whitenotifier-f>
[scopehal] azonenberg cef868e - EyePattern: used simple LCG rather than libc rand() for antialiasing (~20% speedup). Moved some sanity checks outside inner loop
<_whitenotifier-f>
[scopehal] azonenberg ffb1b16 - EyePattern: removed redundant writes that never showed up on screen
<_whitenotifier-f>
[scopehal] azonenberg 83e4fb3 - EyePattern: eliminate more redundant memory writes in inner loop
<azonenberg>
And i'm still tweaking, i think there's room to optimize more
<azonenberg>
I'm also looking into multithreading it, but not certain yet
<monochroma>
nice!
<_whitenotifier-f>
[scopehal] azonenberg pushed 9 commits to master [+0/-0/±9] https://git.io/JThT1
<_whitenotifier-f>
[scopehal] azonenberg 90b268c - EyePattern: more aggressive early-outs to minimize wasted math. Precomputed X time scale multiplication
<_whitenotifier-f>
[scopehal] azonenberg 28b434d - EyePattern: More precomputation
<_whitenotifier-f>
[scopehal] ... and 6 more commits.
<azonenberg>
monochroma: ok i think i'm done with this round of tweaking. No multithreading yet
<azonenberg>
but it's about 2x faster than it was
<azonenberg>
Now to see if i can multithread...
<azonenberg>
Hmm, not gonna be as easy as i thought
<_whitenotifier-f>
[scopehal-apps] azonenberg closed issue #140: cmake doesn't check for presence of yaml-cpp - https://git.io/JJObb
<_whitenotifier-f>
[scopehal-apps] azonenberg pushed 3 commits to master [+2/-0/±4] https://git.io/JThsM
<_whitenotifier-f>
[scopehal-apps] azonenberg 2d2c6ee - Merge pull request #279 from azonenberg/fix-yaml Add detection for yaml-cpp, fixes #140
<azonenberg>
Bird|otherbox: finally got a chance to merge your yaml-cpp PR. What are you planning to work on next?
<Bird|otherbox>
I'll have to look at the tickets outstanding
juli966 has joined #scopehal
StM has quit [Ping timeout: 260 seconds]
StM has joined #scopehal
bvernoux has joined #scopehal
azonenberg_work has joined #scopehal
<azonenberg>
monochroma: so despite a 2x speedup, when i'm pushing enough WFM/s even on my lecroy scope, adding an eye pattern slows down glscopeclient considerably
<azonenberg>
i removed a lot of redundancy so it's a ton better now
<azonenberg>
but i feel like there's still room to improve