amirhirsch

6 ms·

31.
by amirhirsch 4mo ago
I am working on hardware debugging now. There are no good dataset for EE root-cause-analysis. Would be interested in hearing more from your experience: my first name at my HN handle .com
32.
by amirhirsch 5mo ago
The thing many people miss is that the challenge itself isn't the primary signal. The challenge creates an opportunity to observe user activity. You're browser is also fingerprinted.
33.
by amirhirsch 5mo ago
Shout out to all my SF 5am crew checking if their overnight prs passed CI. Real 597 “member of technical staff” energy. I guess we should expect this, it is a Tuesday!
34.
by amirhirsch 5mo ago
I like this too. I have been intentionally -maxxingmaxxing to get the meme out there. It's a good canary to sort out who gets the spicy takes from the pedestrians who probably still copy-paste into the ChatGPT web app like a psychopath
35.
by amirhirsch 5mo ago
I would love to sign up for antigravity cli but when I click on Get Plan it says: “This account isn't eligible for Google Antigravity benefits included with Google AI plans Google Antigravity benefits included with Google AI plans aren
36.
by amirhirsch 5mo ago
ROTFLMAO made my day!
37.
by amirhirsch 5mo ago
Let’s reduce this to absurdity: I think you only wanted clarification of CTF (Capture the Flag) and not AI (Artificial Intelligence) and not GPT-4 (Generative Pre-Trained Transformer version 4) and not CLI (Command Line Interface) and not
38.
by amirhirsch 5mo ago
MIT should spin up MIT.ai for-profit, raise 100B @ 1T, then go buy big computer.
39.
by amirhirsch 5mo ago
At least in my experience (MIT ’06) many of the people most comfortable gaming academics ended up in finance. I've always felt that it was these kind of folks that caused the 2008 financial crisis
40.
by amirhirsch 6mo ago
Peter Turchin’s theory of “elite overproduction” suggests this is a cause for social instability and revolutions
41.
by amirhirsch 6mo ago
Please have an endpoint for up-to-minute billing with hooks for notifications and setting limits, importantly have it correlate-with/be the endpoint that shows usage analytics. Previous-co could never get argo billing to match argo ana
42.
by amirhirsch 6mo ago
Dude is on hacker news on a Sunday. half the GDP of the world is competing with him. What metrics would you like to see?
43.
by amirhirsch 6mo ago
Gp was referring to AN/TPY2 which is the THAAD radar. Iron dome has nothing to do with that systems.
44.
by amirhirsch 7mo ago
Any sufficiently advanced incompetence is indistinguishable from malice. If you are the editor of Gray’s Anatomy, incompetence is malice.
45.
by amirhirsch 7mo ago
Critically, Amdahl’s Law remains a conjecture until we have the maths to disjoin complexity class P from NC.
46.
by amirhirsch 7mo ago
Not sure about the conclusion regarding NVidia value capture. I imagine the context for many applications will come from a physical simulation environment running in dramatically more GPUs than the AI part.
47.
by amirhirsch 8mo ago
Beat me to it. Now I have to delete my reply.
48.
by amirhirsch 8mo ago
> S&S Deli in Cambridge Good lunch spot for a nudnik
49.
by amirhirsch 9mo ago
Here’s some other hints: combine hash stages 2 and 3, it can be two muladds and a XOR For the first several rounds (when every tree value is in use) Combine the stage 5 XOR with the subsequent round’s tree XORs. You can determine even/
50.
by amirhirsch 9mo ago
I think I can hit #1 (current #1 is 1000). sub 900 not possible though. Let me put down my thought process: You have to start to think of designing a 6-slot x8-len vector pipeline doing 48 hashes in parallel first which needs at least 10 st
51.
by amirhirsch 9mo ago
====================================================================== BROADCAST LOAD SCHEDULE ====================================================================== Round | Unique | Load Strategy ------|--------|---------------------------
52.
by amirhirsch 9mo ago
take advantage of index collisions, optimizing round 0 and 11, speculative pre-loading, and the early branch predictor (which now I am doing looking at bits output at stage 3)
53.
by amirhirsch 9mo ago
I'm at 1137 with one hour with opus now... Pipelined vectorized hash, speculation, static code for each stage, epilogues and prologues for each stage-to-stage... I think I'm going to get sub 900 since i just realized i can in-para
54.
by amirhirsch 9mo ago
Very Nice. There is an issue with panning on the million point demo -- it currently does not redraw until the dragging velocity is below some threshold, but it should seem like the points are just panned into frame. It is probably enough to
55.
by amirhirsch 9mo ago
I gave your reply the most generous interpretation and read it in the ironic way as you point out in the edit
56.
by amirhirsch 9mo ago
this won't cost the city too much, there's only like a hundred kids under 6 in this city and 3% of them are mine.
57.
by amirhirsch 10mo ago
I did write this 20 years ago https://fpgacomputing.blogspot.com/2006/05/methods-for-recon... The vendor tools are still a barrier to the high-end FPGA's hardened IP
58.
by amirhirsch 10mo ago
python won because of enforced whitespace. It solved a social problem that other languages punted to linters, baking readability into the spec
59.
by amirhirsch 10mo ago
Probably also end of ZIRP and some “AI washing” to give the illusion of progress
60.
by amirhirsch 10mo ago
The effect of these tools is people losing their software jobs (down 35% since 2020). Unemployed devs aren’t clamoring to go use AI on OSS.

More ›