Friday, 2021-10-08

Las[m]Also: What process should I use for upstreaming my work?00:10
Las[m]Another question: Is there a guide for running the tests?12:30
lkclLas[m], that's up to Jean-Paul as part of his decision and responsibility to maintain and develop the coriolis2 VLSI suite.  i _believe_ he said "long-term, yes".12:53
lkclrunning the benchmark test suite (alliance-check-toolkit) is a matter of typing "bin/", so... ah.... not really :)12:54
lkclif you're modifying things for libre-soc you've already agreed to our Charter so just ask for the repo write access, explain in advance what you intend to do, and go for it12:55
lkclif you mean alliance/coriolis2 modifications then  you should join the alliance-users mailing list and ask there
Las[m]lkcl: I mean Libre-SOC test suite, and upstream to the soc repository13:43
Las[m]Specifically, the work in
Las[m](check pr branch)13:43
Guest55"What is that?!"13:53
Guest55libre-soc on that ?13:54
Las[m]BTW I've already packaged Coriolis and it does work, so that's why I'm asking about the tests for Libre-SOC itself. That way we can have13:55
lkclLas[m], ah you're back14:34
lkclyeah looks perfectly reasonable14:34
lkcli git a pull-push of the branch, which allowed me to do a git diff14:34
lkclin future just drop nix-related and CI-related stuff directly into the master branch14:35
lkclbecause it's unrelated to actual development14:35
Las[m]Yeah I was gone for a bit unfortunately.14:37
Las[m]So do you want to review it, or should I just push?14:38
lkcldone already.14:38
lkclas it was (mostly) a completely separate directory (nix/*) i'm quite happy for you to add that directly14:39
lkcldid you update / change the submodule?  src/soc/litex/florent     |   2 +-14:39
lkclalso i liked the use of an env-var in pinouts.py14:40
* lkcl thinks...14:40
lkclthat should actually be added as a default to the PSpec14:40
Las[m]I assume I have commit access then, thanks!14:41
lkcl(which is a random-config-collecting-of-options class for "stuff specified from the command-line")14:41
lkclyou do now14:41
lkclmy thoughts therefore would be that it would be good to allow the location of the JSON file to be specified via the command-line14:42
Las[m]I did.14:45
Las[m]I'm pretty sure I pushed all of my changes into upstream libresoc-litex though, so you can likely revert that submodule update.14:45
lkclcan you please deal with it?14:51
lkcli have to do a video for a course today14:51
lkclgenerally submodule updates should be done as a single update.15:36
lkclwe learned quite quickly that if you include them in with other code-commits, git can't properly handle it because there's an exceptional non-standard code-path treating the ".gitmodules" file completely differently from standard "index" files15:37
lkcl(*.py, *.nix etc.)15:38
klyslkcl when should you be getting chips back from imec?15:55
klysoh I see it here 06 dec 2021, carry on:
Chips4Makers[m]klys: I expected to see the chips already but I guess they got delay due to chip squeeze. I will see somebody of imec next week Wednesday.16:53
klyschips4makers[m], would you happen to now the quantity being produced?17:07
klysoh thanks.  sounds like they will be moderately priced then, as 5000 EUR / 160 = 36.15 USD17:13
Chips4Makers[m]Where do you get the 5000 EUR from ? The cost is much more.17:15
klysoh, I don't know I just saw it on
Chips4Makers[m]The 5000 EUR mentioned was for possible PCB for testing and doing the test of the chip. It was not about production cost of the chip.17:18
klyswell if it's 10M then that would yield chips at 62,500.00 ea.17:21
Chips4Makers[m]We used TSMC MPW (multi-project wafer) prototype service in order to test if a design works. Such a MPW has much lower setup cost than a normal product run as mask costs are shared but a much higher per unit cost. When going for volume you order a mask set for your own and then have much lower per unit costs. The latter only if you have thousands of products to sell though.18:00
Chips4Makers[m]Actual price for this run including packaging was somewhat over 150€/piece18:16

Generated by 2.17.1 by Marius Gedminas - find it at!