US - Fable 5, GPT5.6, Grok 4.5, and much more coming
China - open source leader, DeepSeek V4, GLM 5.2, 3T models coming soon
EU (self proclaimed bastion of democracy) - we are now spying on you, and we did it in the most undemocratic way possible!
I feel shame at my nationality
> Be Federico Faggin
> Born in Vicenza, Italy, in 1941.
> Study physics. Become obsessed with semiconductors.
> Move to Silicon Valley in your 20s.
> Get hired at Intel.
> Design the first commercial microprocessor by hand.
> The Intel 4004. The chip that started everything.
> Help lay the foundation for every computer, smartphone, and AI system that follows.
> Receive the National Medal of Technology and Innovation from the President of the United States.
> Barely anyone in Italy knows your name.
> Meanwhile, Italy names airports after politicians.
> Be one of the reasons the modern world computes.
> Be Italian.
Remember that with the web there was the same story. The WWW was invented in Europe. The open source stack that created a true revolution of startups was massively based in Europe. Yet all the value was mostly captured in the US. Now it is worse than in the web case.
Modern AI resulted from research made also by many non-US scientists (Hinton, the French folks, Linnainmaa, many others). The pre-training corpus was produced worldwide with massive code contribution from Europe OSS. What is happening with frontier LLMs is unacceptable.
I would like to thank Zed A. Shaw @lzsthw for opening my eyes on the fact that programming at large is viewed exclusively as an activity subservient to corporations. This explains all the weird criticism I've been getting about doing programming recreationally. Thank you Mr Shaw!
@ali_alimoff Followed.
Found you through that old reddit post about some folks deleting content from their platform in a not cool nor informing way ;)
Thanks for trying to build something nice yourself
The Linux kernel AI guidelines are the first sane that I read. It is not a coincidence. Where high level work is done, high level work is pretended, regardless of the tools. https://t.co/yMqJZtOBU1
Modern DRAM is based on a brilliant design from IBM.
But, we're still paying for a latency penalty that's existed since the 60s!
In this video, I'm introducing my research project (Tailslayer) that immensely reduces p99.99 latency on traditional RAM!
By implementing a hedged read strategy taking advantage of (undocumented!) channel scrambling offsets, I've gotten as much as 15x reductions in tail latency.
The technique works across Intel, AMD, Graviton, DDR4, DDR5, x86, ARM, you name it.
Check out the C++ lib I wrote, watch the video, and try it yourself!
@_trish_xD It seems to me that most web applications are really simple CRUD apps, with _maybe_ some complex data to handle, and vanilla js is more than enough for that kind of stuff.