First off, before everything else, today is Liam's birthday. I doubt words can truly express how grateful I am to count him as a friend. I admire him and aspire to be more like him every day. He is also literally, figuratively, objectively, and metaphysically awesome. If you are reading this blog the day it's released you should go message him Happy Birthday, or even better fill his inbox with so many birthday messages that it fills his MIT allotted cloud storage.
Now somehow, Liam's birthday seems to be one of the few predictable events for me in the last month.
A Summary
The last month feels like a series of random events strung together in an almost nonsensical way. In what feels like a blitz, I met a slightly ludicrous number of people, had people offer to fly me out to NYC and SF, run into a bunch of VCs, accidentally applied for a job, visited mansions with living rooms bigger than my home, cracked Jane Street's ASIC reverse engineering challenge, continued CPU work, began work on a GPU, and accepted a job to build a TPU chiplet. Oh and I won the Eigenprize.
How it happened
At the start of August, I received an email from the Eigenprize asking to talk. A few months ago, I had applied for the prize after Liam recommended I give it a shot. The application was short: a few sentences about who you are and what you do. Over the following few weeks, I had a chance to meet Spencer, Mark, and Steven from Orbit. Meeting them was amazing and learning how they think was enlightening. I don't think I've met a group of people quite like them ever before. I hope to become as dedicated, curious, and determined to achieve the impossible as them. A little while later, I received an email telling me I won second place on the Eigenprize and $30k. I am incredibly grateful for the prize. The majority of the funds will be used to fund tapeout of our chip designs.
After winning the Eigenprize, I visited my family. During this time, a large number of people reached out to me. Some were visiting from seeing my blog post on the top ten of Hacker News and others from seeing the Eigenprize. This led to me having a chance to meet a great number of awesome people. Another benefit of being back in the Bay Area was that I could meet Peter Schmidt-Nielson at Saturn Data.
Earlier in the year, I asked Peter about his FPGA startup. After speaking for a little, he offered me his company's old prototypes. For a broke college student like me, these were a massive boon and are the only way Liam and I can run our CPU currently. While in the Bay, I visited his office and got to see his most current work. To be honest, his server designs were mindblowing. I'm really looking forward to his future designs. While speaking to him, I mentioned our plans to tapeout on GF180 to Peter. He recommended we shoot higher: for 28nm, 16 or even single digits. I believed that such advanced processes were out of reach, but he convinced me that with a little more funding we could actually do it.
After returning to MIT, I met with a 16 year old hedge fund manager. He was actually super cool. Honestly, if you told me he was thirty I would totally believe you. His world model was really interesting and gave me some interesting insight. I personally wish my world model was a percent as developed as his. He heard I was interested in chips and directed me to his friend Rob. After speaking to Rob a little, he told me I should speak to one of his friends at Etched. A little while after this, I realized Rob was one of the founders of Etched. It also turns out that speaking to Rob's friend was also just taking a technical interview for a role at Etched. On the positive note, the technical interview was pretty fun and I got a chance to speak to the author of the FP4 multiplier challenge from before. What's next? I suppose we'll see.
Soon after that, I was approached by the CEO of Zero ASIC, a chiplet based ASIC design company on if I wanted to intern there over the semester. He was looking for an intern to work part time on an open source TPU. At the end of the project we should be ready to take out a 16mm^2 TPU on 12nm silicon which is pretty exciting. Along with the TPU, Liam and I will now be working on a GPU for the FLAME lab over the next semester with the eventual goal of tapeout as well. Who knows perhaps we can port Quake 2 by the end of the semester. To be honest we were probably going to build both eventually but its nice to have even greater goals to aspire to.
Now between all of these i met with more people than I can reasonably name in this blog, plus it feels like an invasion of privacy for me to drop the names of everyone that's reached out to talk with me recently without their explicit permission.
The day before this blog post Liam and I finally cracked Jane Street's ASIC reverse engineering challenge. We actually started when it released, before most of the events described in this blog. However, after solving most of it we hit a roadblock and didn't pick it back up until right before the due date. To be honest I wish I had devoted more time to it. I did a lot of work on the netlist extraction, verilog setup, and testbench generation, early on, but the aforementioned events took up a ton of time in the middle. Liam carried during those times, and I certainly wouldn't have been able to do it without him. If you would like more details, you should read our writeup on his blog.
Finally, that brings us to September 4th, the day this blog is released, and more importantly Liam's birthday. If you haven't wished him, you should wish him now.