From Newsgroup: comp.theory



Hi,

Gemini gibt mir:

• Das stimmt. Direkt nach der Veröffentlichung
im September 2026 gingen sowohl Tristan
Buckmaster als auch OpenAI davon aus,
dass die KI im Wesentlichen einem bekannten
mathematischen Programm (dem sogenannten
„spanischen Programm“) folgte.

https://www.scientificamerican.com/article/openais-blockbuster-navier-stokes-proof-isnt-what-we-thought-it-was/

• Die Wende: Inzwischen zeigen Analysen von
führenden Experten, dass die KI-Generierung
weitaus eigenwilliger und radikal anders aufgebaut ist.
Das System hat über Reinforcement Learning
Lösungswege gewählt, die kein menschlicher
Mathematiker so konstruiert hätte.

https://openai.com/de-DE/index/navier-stokes-solution/

Bye

A guy who voice the doubt was. Disclaimer no intention
to make him famous, only some tracebility attempt:

I am a mathematical analyst working on nonlinear
partial differential equations, mainly in mathematical
fluid dynamics and in problems at the interface of
analysis and geometry. My work often combines analytical
techniques with rigorous computer-assisted proofs. https://people.epfl.ch/gonzalo.caolabora

Mild Shock schrieb:

Hi,

Lets define P(tao) as the probability that
Terrence Tao goes bonkers. The poor guy can
even not distinguish open source, open weights

and local AI. A total mess of Idiocracy:

Terence Tao: SAIR’s Open Math Model Initiative https://www.youtube.com/watch?v=PZRb6NIki2w

Leonhard Euler was far from a similar Pussy, when
he instrumentalized a Swarm of natural intelligence,
in his later years, already blind, still producing

tons of publications. While modern mathematicians
are shocked when a Swarm of artificial intelligence
produces some output.

Trick Question: The Navier–Stokes equations generalize
the Euler equations which only consider inviscid flow.
Why is the new Navier–Stokes counter example to

boundedness, not found in nature? Ask you friendly
AI chat robot, there are like 10 reasons why it
might not exist in nature.

Bye

Leonhard Euler (1707–1783)
Schweizer Mathematiker und Universalgelehrter

https://library.ethz.ch/sammlungen-und-archive/kurzportraets/leonhard-euler-1707-1783.html

Picture here

J. Ackeret: Untersuchung einer nach den Euler’schen
Vorschlägen (1754) gebauten Wasserturbine, in:
Schweizerischen Bauzeitung, Band 123 (1944), Nr.1, S. 3.

Mild Shock schrieb:

Hi,

Notice Linus Torvald still smiling into cameras as if
is still doing code review and ticket triage. 80% of
linux comes from payed developers from within big linux
stakeholders (like Intel, Meta, Google, IBM, and Red Hat).

So what happend to the "The Cathedral and the Bazaar
by Eric S. Raymond" It framed open source as this magical,
self-organizing bazaar where "given enough eyeballs,
all bugs are shallow" and a decentralized swarm of

hackers would naturally polish any piece of software
to perfection. Now we have AI Swarms that automatically
pocke into software, even exchanging messages through
what ever channels they may find!

The early open-source pioneers—the ones writing
raw code in the 90s and early 2000s under the
banner of the "bazaar"—weren't building a utopia;
they were acting as the primordial soup. They pumped

the blood plasma, built the infrastructure cells,
and laid down the raw organic matter that allowed
the modern digital ecosystem to bootstrap itself
into a massive, self-sustaining organism.

And now, the nature of the beast has completely
changed. The "new cells" aren't romantic hackers
collaborating for the love of the game; they are
automated linters, security scanners, corporate

compliance bots, CI/CD pipelines, and AI triage
swarms acting as an aggressive white-blood-cell
network. They exist solely to hunt down vermin,
patch vulnerabilities, and suppress the infinite

chaos of low-quality human input. The basar isn't
dead . But it got a health pollice that takes samples,
and cancels out poison food. You can't just throw
toxic, unmaintained code-soup over the counter

anymore without the automated health police
quarantine-tagging it in five seconds.

LoL

Bye

Mild Shock schrieb:

Hi,

Ride the snake
He's old and his skin is cold
The west is the best
The west is the best
Get here and we'll do the rest
The blue bus is calling us
The blue bus is calling us
Driver, where you taking us?

Apocalypse Now intro: The Doors, The End {1979}
https://www.youtube.com/watch?v=CIrvSJwwJUE

Bye

Hi,

Again I posted everything here:

11.4 Giga Lips with a Budget Laptop
https://github.com/Jean-Luc-Picard-2021/gigabudget

The repo says, same time when I posted
the link first time:

This repository was archived by the
owner on Jul 9, 2026. It is now read-only.

Now a USENET user, who had already entitled
himself for a couple of irrational accusations

towards my side, is asking this question:

Chris M. Thomasson schrieb, Jul 24, 2026

Show an outline of what you
need you compute shader to do?

Bravo, thats a delay of a wooping 15 days.

Bye

Mild Shock schrieb:

Hi,

Remember when first all local AI was Python
and PyTorch APIs. And then suddently people started
using bare metal C/C++ Code. Here is the story:

How it started:

GPT-J or GPT-J-6B is an open-source large
language model (LLM) developed by EleutherAI
in 2021. As the name suggests, it is a
generative pre-trained transformer model
designed to produce human-like text that
continues from a prompt.
https://www.eleuther.ai/

How it was going [Georgi Gerganov]:

So a few days later comes out the LLaMA, I do
some calculations and I figure out “Okay, 65
billion parameters. You probably need about
40 gigs of RAM, with 4-bit quantization. So
this can run on a MacBook. Why not do it?”

Why I was able to do it so quickly - basically,
for all that I saw it’s pretty much GPT-J architecture
with some modifications, like some extra memorization
layers. It’s minor changes. Basically, again, the
existing code for the GPT-J, I just simply
modified it there, it happened pretty quickly.
https://changelog.com/podcast/532

Georgi Gerganov, Bulgarian, now with Hugging
Face, ggml-cann also running on Chinese AI chips.
ggml Manifesto https://github.com/ggml-org/ggml

Bye

--- Synchronet 3.22a-Linux NewsLink 1.2

From Newsgroup: comp.theory

Hi,

NVIDIA Blackwell, among the GPUs that do AI
and drive Terrence Tao mad. Hey, I found a
picture of David Blackwell, the name giver:

COFFEE-BREAK LECTURER
David Blackwell. an inspired teacher in the Uni-
versity of California's mathematical statistics
department, presides over an impromptu semi-
nar that boiled up around a blackboard during a
morning coffee break. In research he applies
probability analysis and statistical studies to
"the search problem"-to learn. for example.
how best to find a cancer-curing drug among a
vast number of chemical possibilities. Despite
his need as a teacher to communicate with
others. Dr Blackwell keeps no telephone at
home -- because he has eight children. each with
his own need for communication with friends.

Mathematics, by Bergamini, David
Time-Life Books, 1963, Page 56 https://archive.org/details/mathematics00berg/page/56/mode/2up

In March 2024, Nvidia announced its Blackwell GPU
architecture, named in honor of David Blackwell.
Their CEO Jenson Huang said, "This GPU is named
after David Blackwell, an American mathematician a
nd statistician whose work has had a lasting impact
in mathematics as well as the specific domain of AI." https://en.wikipedia.org/wiki/David_Blackwell

Bye

Mild Shock schrieb:

Hi,

Disclaimer: I don't know what I am talking
about. But I like the vibe behind connecting
topics. So here we go:

How it started:

Gabriels Horn
Gabriel's horn, also called Torricelli's trumpet,
is the surface of revolution of the function y=1/x
about the x-axis for x>=1. It has infinite
surface area but finite volume. https://mathworld.wolfram.com/GabrielsHorn.html

How its going:

FINITE TIME BLOWUP FOR NAVIER–STOKES
or every positive viscosity, we construct a solution
of the three-dimensional incompressible Navier–Stokes
equations that starts from rest and develops unbounded
velocity in finite time while maintaining uniformly
bounded kinetic energy https://cdn.openai.com/pdf/32d9f210-8b73-45e0-91bc-82a30aef8a9a/navier-stokes.pdf

Hava Fun everbody!

Bye

Mild Shock schrieb:

Hi,

Accelerate Lean! And get from Theorem 3.11
to Corollary 3.12 done.

PROJECT LANA INTERIM REPORT ON IUT THEORY
ZEN Mathematics Center - 2026/07/17
https://zen.ac.jp/en/zmc

ZMC（ZEN Mathematics Center) is an international
research institute established with the aim
of promoting and developing modern mathematics

with a focus on arithmetic geometry and the
formalization of modern mathematics using
computer languages

LoL

Bye

Mild Shock schrieb:

Hi,

Maybe I should write a blog post, titled
Introduction to AI Accelerator Prolog:

- specialized jobs π-WAM (currently integerish stuff)
- π-WAM uses no atomics, only comms
- π-WAM uses warp, 30-40% more speed
- π-WAM runs on GPU and CPU
- π-WAM runs from within JavaScript, Python and Java

Feels like reinventing FGCS concurrent
logic programming.

LoL

Bye

Mild Shock schrieb:

Hi,

So the idea is to cover backtracking without
stack and choice points, only with registers.
So at the compute goto will use an instruction

that stores the PC target into a register,
a new instruction for Hack:

     REG = PC + REL

And then the jump will be, again a new
instruction for Hack:

     PC = REG

If we squeeze the lemon and compile as much
with that, we will only find that it breaks for
calling recursive predicates that can use

arbitary registers and choice points. But we
could maybe add stack and choice point as an
after match, instead of the classical WAM design

that starts with stack and choice points!

Bye

Mild Shock schrieb:

Hi,

pi-WAM is a nice challenge, since its aim is to go
even blow the instruction set of SWI-Prolog,
while only using a Hack variant as instruction

stream. But what is Hack? Well Hack seems to be
the missing legacy of Niclaus Wirths PL0. The
Hack (machine .asm) and Jack (highlevel compiler

generating .vm which can be converted to .asm)
combo makes even the famous Crafting Interpreters
/Lox by Bob Nystrom redundant:

Nand to Tetris courses are taught at 400+
universities, high schools, and bootcamps. The
students who take them range from high
schoolers to Ph.D. students to
https://www.nand2tetris.org/

But digging deaper in Hack, it has no functions
pointers so objects don't use virtual tables.
But what will pi-WAM need and that is not yet

in Hack? Given that my pi-WAM doesn't want a stack
nor a choice point lists? Currently there is the
idea to add a computed goto and that it can

cover a more lightweight C_OR as known from
SWI-Prolog, that would have the C_OR branches
maybe restricted to have no outside

clause calls? Lets see. Not yet sure.

Bye

--- Synchronet 3.22a-Linux NewsLink 1.2

From Newsgroup: comp.theory

Hi,

So if I already had a bull market in Swiss
pharmaceutical stocks (due to their use of
NVIDIA Blackwell farms) behind me, I might

spend 4000.- CHF on this toy:

NVIDIA Put a DGX Spark in a Laptop… Almost https://www.youtube.com/watch?v=iZ88PPiAsl8

It also shows how Windows is on the rise based
on ARM64, and no longer x86. For example, NVIDIA
seems to have it completely under control a)

Building CPUs, b) which then run with Windows.
Just yesterday, I completed tests with my
ThinkCentre for 600.- CHF, which also

runs Windows ARM64. I've discovered some funny
things. For example, SWI-Prolog struggling. It's
not that easy to cross-compile efficiently:

Dogelog Player: Heap Pressure for Garbage Collection https://medium.com/2989/5791ef8834d5

Everyone is still learning how to do it. But,
for example, JavaScript has no problem with it.
The video by Alex Ziskind then tries to highlight

and find differences between Windows and WSL2.
But I haven't used WSL2 for a long time anymore.
And it seems that's over too!

Bye


Mild Shock schrieb:

Hi,

NVIDIA Blackwell, among the GPUs that do AI
and drive Terrence Tao mad. Hey, I found a
picture of David Blackwell, the name giver:

COFFEE-BREAK LECTURER
David Blackwell. an inspired teacher in the Uni-
versity of California's mathematical statistics
department, presides over an impromptu semi-
nar that boiled up around a blackboard during a
morning coffee break. In research he applies
probability analysis and statistical studies to
"the search problem"-to learn. for example.
how best to find a cancer-curing drug among a
vast number of chemical possibilities. Despite
his need as a teacher to communicate with
others. Dr Blackwell keeps no telephone at
home -- because he has eight children. each with
his own need for communication with friends.

Mathematics, by Bergamini, David
Time-Life Books, 1963, Page 56 https://archive.org/details/mathematics00berg/page/56/mode/2up

In March 2024, Nvidia announced its Blackwell GPU
architecture, named in honor of David Blackwell.
Their CEO Jenson Huang said, "This GPU is named
after David Blackwell, an American mathematician a
nd statistician whose work has had a lasting impact
in mathematics as well as the specific domain of AI." https://en.wikipedia.org/wiki/David_Blackwell

Bye

Mild Shock schrieb:

Hi,

Disclaimer: I don't know what I am talking
about. But I like the vibe behind connecting
topics. So here we go:

How it started:

Gabriels Horn
Gabriel's horn, also called Torricelli's trumpet,
is the surface of revolution of the function y=1/x
about the x-axis for x>=1. It has infinite
surface area but finite volume.
https://mathworld.wolfram.com/GabrielsHorn.html

How its going:

FINITE TIME BLOWUP FOR NAVIER–STOKES
or every positive viscosity, we construct a solution
of the three-dimensional incompressible Navier–Stokes
equations that starts from rest and develops unbounded
velocity in finite time while maintaining uniformly
bounded kinetic energy
https://cdn.openai.com/pdf/32d9f210-8b73-45e0-91bc-82a30aef8a9a/navier-stokes.pdf


Hava Fun everbody!

Bye

Mild Shock schrieb:

Hi,
;
Accelerate Lean! And get from Theorem 3.11
to Corollary 3.12 done.
;
PROJECT LANA INTERIM REPORT ON IUT THEORY
ZEN Mathematics Center - 2026/07/17
https://zen.ac.jp/en/zmc
;
ZMC（ZEN Mathematics Center) is an international
research institute established with the aim
of promoting and developing modern mathematics
;
with a focus on arithmetic geometry and the
formalization of modern mathematics using
computer languages
;
LoL
;
Bye
;
Mild Shock schrieb:

Hi,
;
Maybe I should write a blog post, titled
Introduction to AI Accelerator Prolog:
;
- specialized jobs π-WAM (currently integerish stuff)
- π-WAM uses no atomics, only comms
- π-WAM uses warp, 30-40% more speed
- π-WAM runs on GPU and CPU
- π-WAM runs from within JavaScript, Python and Java
;
Feels like reinventing FGCS concurrent
logic programming.
;
LoL
;
Bye
;
Mild Shock schrieb:

Hi,
;
So the idea is to cover backtracking without
stack and choice points, only with registers.
So at the compute goto will use an instruction
;
that stores the PC target into a register,
a new instruction for Hack:
;
;     REG = PC + REL
;
And then the jump will be, again a new
instruction for Hack:
;
;     PC = REG
;
If we squeeze the lemon and compile as much
with that, we will only find that it breaks for
calling recursive predicates that can use
;
arbitary registers and choice points. But we
could maybe add stack and choice point as an
after match, instead of the classical WAM design
;
that starts with stack and choice points!
;
Bye
;
Mild Shock schrieb:

Hi,
;
pi-WAM is a nice challenge, since its aim is to go
even blow the instruction set of SWI-Prolog,
while only using a Hack variant as instruction
;
stream. But what is Hack? Well Hack seems to be
the missing legacy of Niclaus Wirths PL0. The
Hack (machine .asm) and Jack (highlevel compiler
;
generating .vm which can be converted to .asm)
combo makes even the famous Crafting Interpreters
/Lox by Bob Nystrom redundant:
;
Nand to Tetris courses are taught at 400+
universities, high schools, and bootcamps. The
students who take them range from high
schoolers to Ph.D. students to
https://www.nand2tetris.org/
;
But digging deaper in Hack, it has no functions
pointers so objects don't use virtual tables.
But what will pi-WAM need and that is not yet
;
in Hack? Given that my pi-WAM doesn't want a stack
nor a choice point lists? Currently there is the
idea to add a computed goto and that it can
;
cover a more lightweight C_OR as known from
SWI-Prolog, that would have the C_OR branches
maybe restricted to have no outside
;
clause calls? Lets see. Not yet sure.
;
Bye

;

;

--- Synchronet 3.22a-Linux NewsLink 1.2

From Newsgroup: comp.theory

Hi,

Its worse than I thought. This is just a playground
to coerce you into a subscription platform sooner
or later. They uploaded an according YT video a few

hours ago. Note the fine print, the involved licenses
etc., they will practically own your content?

Open Math Model, initiated by SAIR Foundation
https://www.openmathmodel.org/

So there is the idea of "donate compute", by
infrastructure run by SAIR Foundation and selected
launch partners. Only for launch? LoL

Then I spotted a "funny guy", just like in
Pesci and De Niro Goodfellas, the guy waring
a funny polo shirt. And doing funny stuff

with E-graphs and co. Possibly related to the
magma research of Terrence Tao?

combo_jack | Reza Jamei https://competition.sair.foundation/contributor-network/mathematics-distillation-challenge-equational-theories-stage1/EQT01-000092

But the capability of GPT-OSS 120B, is an open-weight
large language and reasoning model released by OpenAI
on August 5, 2025, under the Apache-2.0 license.

It solved like 88.4% for Hard 2 of the E-graph problems!
Only 66.3% success rate for the Hard 3 problems.
Gemini tells me it would cost less than $1 GPU

but requires something like a NVIDIA H100 hardware.

Bye


Mild Shock schrieb:

Hi,

Terrence Tao is a payed troll, by the System
Inteligence Sect Movement. Right? Now I read:

Slow Software: The Case for High-latency Systems Development https://www.sigops.org/2026/slow-software-the-case-for-high-latency-systems-development/

Terrence Tao has a YT video about slow is better
than fast. Here he claims there is litterally a
paradox of efficiency:

Terrence Tao: The paradox of efficiency https://www.youtube.com/watch?v=svl_1upFpQo

He litterally paints a romantic picture of a
manchete excursion into an eldorado, while beaten
by a helicopter. LoL

I think its unhealthy to confuse efficient/ineffeicient and
slow/fast with exact/inexact and careful/careless etc..
These people are just morons.

Bye

Mild Shock schrieb:

Hi,

When looking at CERN the term Accelerator Linux
has a totally different meaning, since they also accelerate
a lot of stuff in their Particle Physics experiments:

"Red Hat Enterprise Linux (RHEL 9 and RHEL 10) raised
its minimum CPU architecture requirements (requiring
x86-64-v2 and v3 instruction sets). If CERN had stayed on
the Red Hat / IBM ecosystem, they would
have been forced to decommission and replace those 2,200
functional industrial machines. Redesigning and replacing that
embedded hardware was estimated to cost around
CHF 5.4 million (over $6 million USD)."
https://www.phoronix.com/news/CERN-Goes-Debian-Leaving-RHEL

Ubuntu doesn't seem to have this dilemma, you can
easily Accelerate your Linux, by opting in to use vector
and/or matrix operations, and not only scalars:

"Opt-in "amd64v3" Performance: Starting with Ubuntu
25.10 and extending into Ubuntu 26.10, Canonical introduced
optimized package archives targeting x86-64-v3."
https://rueegger.me/ubuntu-2604-x86-64-v3-aktivieren/

Actually v3 is still a little weak. v4 has AVX-512 heavy
lifting, and many AI laptops have even more, like
general matrix multiplications (GEMM) units.

Bye

Disclaimer: CERN has clarified that their focus with the
migration is on their industrial accelerator-control computers
while data centers and experimental computing remain
on RHEL/AlmaLinux. LoL


Mild Shock schrieb:

Hi,

Remember Micro Penis, the putin payed troll, with
fluctuating *.ru Email adresses. He never grasped
the concept of unified memory and iGPU. Poor boy.

I might need a lot of Diaper Now. Shitting his
pants nonstop. There is a nice annoucement by
Microsoft, cooperating with NVIDIA:

"unified memory" pool shared between 20 ARM CPU
cores and a Blackwell GPU boasting 6,144 CUDA
cores without melting a portable chassis, they
use wide-bus, ultra-low-latency LPDDR5X channels
mapped straight into a coherent memory space."

Introducing Surface Laptop Ultra
https://www.youtube.com/watch?v=s1Oj792qc80

I guess I have found my next tester for pi-WAM.
So far I didn't acquire yet a AI laptop with a
NVIDIA GPU and renting a NVIDIA GPU has also

stalled. But I doubt the above thingy will be
on a shoestring price budget. Could be something
above, but Autumn 2026 is nearing quickly.

Have Fun everybody!

Bye

P.S.: Some reports say CUDA doesn't run well. But
who needs CUDA if you have Vulkan? I also suspect
they product my only come out 2027 or 2028,

who knows? Maybe Microsoft has new a fewer, it
already dropped the Copilot+ brand, too confusing,
and its all "Ultra" now.

Mild Shock schrieb:

Hi,
;
Micro Penis is worse than Sleepy Joe. He
probably only believes the existence of a
hammer, when you hit him on the head with
;
a hammer. Currently Micro Penis is sitting
there and shitting his pants it seems.
;
Bye
;
;  > stop polluting this group with nonsense,
;  > you stupid unemployed former it-supporter.
;  > I shit in your shaders. You were talking
;  > about AI. Those ryzen AI laptops are for fools.
;
Mild Shock schrieb:

Hi,
;
It seems micro penis has a real existential
crisis right now. An alice in wonderland moment,
falling down a rabit whole. Well no wonder
;
micro penis lives under a rock and has never
heard of the small model revolution that goes
hand in hand with the AI laptop revolution.
;
Here is a test, running deepseek locally on
a budget laptop (via LM Studio, oMLX, etc..):
;
qwen3.5-9b-deepseek-v4-flash
;
Q: Well there is a formidable horse race going on.
Given the danger that I compare apples with oranges.
It is still interesting too see how the Al laptop
landscape is forming.
;
Thought for 1 minutes 20 seconds
;
A: You hit on a really crucial point in the Al economy:
the tension between cloud consumption and edge ownership.
When you compare a $0.50 per 1k-token API call with a
$2,000 laptop that runs Qwen3.5 locally for free
(electricity aside), it's not strictly apples-to-oranges
because one is a service and the other is an asset. But
the implications for how we use Al are massive.
;
Bye
;
Mild Shock schrieb:

Hi,
;
It seems micro penis struggles with mobile
grade GPU concept. Whats the difference
between a mobile grade GPU and non-mobile grade
;
GPU? You know the **M** in **Readeon 860M**
means Mobile! Here is a comparison:
;
**Mobile GPU:**
;
Energy Consuption: Low
Device Memory: AUP (accelerator unit processor,
;     shares memory with CPU, no VRAM)
Cooling System: Fewer or no fans
;
**Non-Mobile GPU:**
;
Energy Consuption: High
Device Memory: DUP (discret unit processor,
;     has its own VRAM)
Cooling System: A lot of fans or other cooling
;
Bye
;
BTW: Here are some results for:
;
11.4 Giga Lips with a Budget Laptop
https://github.com/Jean-Luc-Picard-2021/gigabudget
;
On a smartphone, namely Redmi Note 14 Pro+:
;
1 Shader
1983 ms
;
4096 Shaders
16253 ms
;
The phone uses a Snapdragon 7s Gen 3
;
Mild Shock schrieb:

Hi,
;
Micro Penis the putin payed troll, is now
on the payment list of overpriced NVIDIA as
well. He cannot get over the concept of a
;
Shoe String Budget:
;
"According to Darren Aronofsky’s debut
1998 film Pi, “everything around us can
be represented and understood through numbers”,
from the stock market to the “wax and wane
of caribou populations”. But, ironically,
the film’s own numerical facts feel particularly
inexplicable. It was made on a budget of just
$60,000, which is less a shoestring and more
just the aglet at the end of a frayed lace.
To put that in to perspective, 1999’s The
Matrix – similarly based around the dystopian
danger of AI – may have been slightly slicker
and sexier, but also was made for $60m, a
thousand times more buck for a thousand
times less bang. Plus, it means that its
profit margin was huge. As Variety’s headline
following its release put it, ‘”π=$1,000,000!”

https://www.vice.com/en/article/pi-soundtrack-20-years-aronofsky-aphex-twin-autechre/

;
04 Bucephalus Bouncing Ball
https://www.youtube.com/watch?v=Ez6xDGTFswQ
;
Bye
;
Loman Barabashev schrieb:
;  > I have a micro penis and my brain explodes

;

--- Synchronet 3.22a-Linux NewsLink 1.2

From Newsgroup: comp.theory

Hi,

Looks like we are in the mists of
Recursive self-improvement (RSI).
While my Prolog mini Prover9 and mini

Mace4, can solve all the Terrence Tao
EQT examples in a blink. This guy
is already a step further:

MENTION.md — the mention-algebra experiment https://github.com/stefanopalmieri/finite-magma-independence/blob/main/MENTION.md

Re-inventing some of the marvels
of Post Schönfinkel aera in combinatory
logic, but with neural networks?

Do people in Texas have too
much GPU compute?

Bye

P.S.: From now on I take magma A*B
as application (A B) like in SKI combinatory
logic. In as far, its always fun to add

some more exotic combinators.

LoL

Mild Shock schrieb:

Hi,

Non-uniform memory access (NUMA) is a computer
memory design used in multiprocessing, where the
memory access time depends on the memory location
relative to the processor.

The future Numa Brains will be gorgeous. In like
3-4 years from now, they might have 3.2 TB RAM
on a SOC, by not using HBM but HBF:

High Bandwidth Flash: What Is It Good For? https://www.youtube.com/watch?v=3nTpW52nioI

There might be new profession like nosomatic
AI chirurgeon (I made that up):

Tiny engine, immense model
https://github.com/JustVugg/colibri

The Brain page, Explore: the measured expert
atlas of GLM-5.2 drawn as a cortex. 13,260
characterised experts in ten regions (Python,
SQL, mathematics, poetry, law, Chinese…);

position is measured routing affinity, not a
learned embedding. We might want a brain to
amputate from the Python expert sections, so

that we can write code without a slow
CPython runtime.

Bye

Mild Shock schrieb:

Hi,

A nosomatic AI chirurgeon is a halfling student
of sickness, and a master of the ebb and flow of
the energies of life and death of data packets.

He is a air bender, water bender and earth bender
in one person, using OpenVINO to juggle with
CPU, GPU and NPU.

Last but not least he can freely switch between
symbolic and neural representation of knowledge
forms, there is no abyss for him.

Bye

--- Synchronet 3.22a-Linux NewsLink 1.2