The top level is right now finalizing, so I should manage to push it before midnight your time and still make it

Please make sure to kickoff a pre check job on the platform after you push.

OK

So, it's still hanging in step 36 of DRC checking... I'll push it when it's done, then the pre check job should pick it up, right?

This design turned out damn ginormous, because of all the ALU and MACC stuff

ChatGPT was right, she predicted this, haha

yes! done!

aaand it's writing, then it has to upload

The GDS for Matrix Convolution and for Matrix Multiplication can already reviewed here @jeffdi https://github.com/leviathanch/kicp/tree/main/gds

@jeffdi I just tried to push, and the layout is too large for git...

I hate everything right now LOL

OK, it's in LFS now, I hope your precheck software can handle that

That

That's the chip

Just figured out, how to properly use macro blocks

YEY

I think I'll regenerate it using macro blocks, then the SRAM and the two engines will be more visible

But since I'm already over the time limit, I guess I'll settle for this implementation now... unless they wanna manufacture it on Monday, which means, I'd have another day, to clean up some stuff

Hey @jeffdi. Does this low subscription to the shuttle mean it will be delayed? (We're still having issues getting our core integrated into user_project_wrapper in a way that it passes precheck AND still will be useful and functional

OK. I made this chip design, but it doesn't seem to be includeable into caravel because of some errors from OpenLane...

OpenLane managed ONE time to place and route the core, but now it gives me congestion errors again... I guess my design is just too complex?

Also, the fact that OpenLane isn't using the available area fully is a waste of silicon material, that should really be fixed

Unused substrate... yikes

Try moving the macro up 20 microns N ?

You can change the die size for the macro integrated into user project wrapper.

Is the congestion on the user project wrapper?

No, it's when we try to rebuilt the top core we've built after having fixed some bugs we've discovered

message has been deleted

OK, after setting GRT_ALLOW_CONGESTION = true GRT_ANT_ITERS = 50 GRT_ANT_MARGIN = 12 It now passes through and is right now doing global routing

📣 Announcing the winners of the AI-Generated Open-Source Silicon Contest – Congratulations to our winners Hammond Pearce (NYU), Xinze Wang, Guohua Yin, and Yifei Zhu at Tsinghua (Berkeley Shenzhen Institute) and Asma Mohsin (Rapid Silicon) 🏆 We would like to thank all the participants for your hard work, creativity using generative AI, and incredible speed in implementing and verifying your design through the Efabless OpenLane flow. 👏 https://shorturl.at/oKNS4