ah, intended to go but forgot all about it with various other stuff going on. I'm currently doing som design work among other things for Zeropoint, quite fun stuff
The first Göteborg @risc_v meetup is a wrap: Rivos, Inc @gaisler_space @AssuredAB @ZeroPointTechno @Kryptoblog Thanks all! It was a blast!
Oh, and if you're interested in giving a talk -- ping me!
Interesting to hear about using qemu together with verilator outputs, a possibility that's been brought up at work but didn't even know AMD had official work on it. Last serious technical stuff for the day I think, time to find a chimay
Pretty sure I just saw my living room (image from unpacking of Subservient) in Philipp Wagners talk at FOSDEM. Good with a surprise to wake you up properly after a night of beers
Watching the first session for the ASYNC summer school. Big brained stuff. As usual my first thought is "how can I build a synthesizer with this?"
Day one of RISC-V Week done with then. I should try Clash. I should also fire up a Linux-ready RV core on one of my boards and do some more serious project rather than messing around bare metal. Right now I feel more like trying a beer though
got a weird mood yesterday and started cleaning up the audio synthesizer I originally intended for openmpw. Lot of things going on as usual so we'll see when I actually make repos public but man cocotb is nice, having scipy directly available in DSP testbenches and such
So I'm not only going to show up at the RISC-V Week next month, I'll actually speak for a bit. Now, I'm not all that worried given some prior stage experience, but I suspect the audience here will be somewhat different from the crust punks I screamed in front of in my late teens
A-well-a everybody's heard about the SERV!
Don't you know about the SERV? Come to the @risc_v days in Paris May 3-5 and check out @knordmark89 and mine presentation How much score could a CoreScore score if a CoreScore could score cores
open-src-soc.org/2022-05/pro…
got slightly panicky since I didn't start until right before christmas (and had other full time work in the days after), but seems like I might get a revised Subservient submitted for mpw4... waiting for the tapeout job results, precheck looks fine
Gotten comfortable at my new assignment (hacking away at some QNX stuff within the Geely sphere), and rested up with some vacation. I wonder what the natural next step would be to better support openlane with fusesoc, excluding the major restructuring to be done at some point
closest thing to a "tech project" during my weeks off was extending a friends old TR-505 deum machine with individual outputs. really minimal hack but I always enjoy poking on musical gear
Nordmark retweeted
And if someone wants to do a master thesis or other research projects around SERV I got plenty of things like this that I would love to investigate (e.g. 2 and 4-bit versions, debug interface, share decoder between several cores) so feel free to reach out if you're interested
Nordmark retweeted
I have an idea around a potential optimization for SERV and I wonder if there are any existing algorithms or methods to help me here. So I would love to get some pointer from people who are better at logic optimization than I am.
The idea goes like this...
Nordmark retweeted
ok, so watching gds files in 3d turned out to be very easy and also ridiculously cool. Here's the subservient SoC.
Forget waveforms. From now on I will only do my chip debugging by flying around between metal1 and metal2 looking for anomalies in the connectivity
github.com/klasnordmark/cara… and here we have the fork of the caravel_user_project repo then, changed to work with Subservient, including a small test toggling a GPIO.
github.com/klasnordmark/subs… Subservient together with a bit of RAM, unfortunately not a larger OpenRAM block due to issues - might be changed before submission. Tests to be added as well, a fork of caravel_user_project with this and full-chip tests added coming soon
This core is used to provide a hardened macro, no use trying to build complete GDS of a chip with the way the open MPW submissions work now.












