I wrote a list of startups I would like here: https://github.com/samsquire/startups
isn't 25 more-or-less a startup incubator a la YC?
Ask HN: Startup ideas that you'll never do?
51–60 of 69 posts
Re: Ask HN: Startup ideas that you'll never do?
#52IMDB for books
goodreads?
Re: Ask HN: Startup ideas that you'll never do?
#53Bitgrid - A new computing architecture, which seems absurd at first, but might be the one that democratizes access to Petaflops/second.
Take an FPGA, rip out all of the routing hardware, and basically have a sea of LUTs in a grid, clock them in two phases like the color on a chess board.
It's a huge waste of transistors, but every single bit of compute is pipelined, so you could run an LLM with one output per cycle, for example. I'd estimate a 8192 * 8192 array could fit in a chip the same size as my i7 processor.
(Assume an FP16 multiply fits in 1024 cells, there could be 65,536 of them, each giving an answer each clock cycle, at 2Ghz --> 131 TFlop FP16) If you manage to optimize the multiply to fit in a smaller area, you can immediately get more TFlops out, something that an FPGA can't do.
I'll be glad to help anyone who wants to make this thing go.
Re: Ask HN: Startup ideas that you'll never do?
#54Re: Ask HN: Startup ideas that you'll never do?
#55Earlier quoted context omitted.
isn't 25 more-or-less a startup incubator a la YC?
Does YC provide tech and other business functions to its startups?
I don't know if they do, per se ... but a lot of incubator/coworking spaces provide levels of technical support beyond just power & internet
Re: Ask HN: Startup ideas that you'll never do?
#56Re: Ask HN: Startup ideas that you'll never do?
#57I wrote a list of startups I would like here: https://github.com/samsquire/startups
17 can be done by tools like Splunk
I am looking for visualisations that model things to the point that bottlenecks and inefficiency are obvious.
I think visualisations of business processes can be similar to video games.
Here's a visualisation I'm playing with:
Re: Ask HN: Startup ideas that you'll never do?
#58A way to crowdsource patronage - everything/everybody seems to be asking for subscriptional support (Patreon or the like) I would love to support a lot of the creators I know more personally ... but if I were to add-up even the $1/mo they each ask for, I'd be into the $100s a month easily ... so I live with the ads they have Something a la Brave Browser's tokens, though, would be great - pitch, say, $50 into a "patro…
Do you want to talk more about this idea? This is what I want too!
I am thinking of sponsoring 12 really high quality blog posts for $12 a year.
Re: Ask HN: Startup ideas that you'll never do?
#59Earlier quoted context omitted.
17 can be done by tools like Splunk
I am looking for visualisations that model things to the point that bottlenecks and inefficiency are obvious. I think visualisations of business processes can be similar to video games. Here's a visualisation I'm playing with: https://processes2.chronological.repl.co/
there are definitely tools (monitoring, observability, etc) out that that do that (or most of that) currently :)
Re: Ask HN: Startup ideas that you'll never do?
#60Earlier quoted context omitted.
I am looking for visualisations that model things to the point that bottlenecks and inefficiency are obvious. I think visualisations of business processes can be similar to video games. Here's a visualisation I'm playing with: https://processes2.chronological.repl.co/
there are definitely tools (monitoring, observability, etc) out that that do that (or most of that) currently :)
I've used grafana, but what I want are visualizations that represents trains or cars or particles and queuing theory.