Author: flipaoXIX

DeepSeek, AWS HPC SDR vs Multipath-TCP, OCSP death, AlphaChip, Visual AI agents, Ollama, Local AI, Bob Bowman

Nice analysis about DeepSeek without hype.

AWS HPC: Didn’t know AWS offered HPC services (articule from 2021). I liked to find more details about SDR: Multipath LB, Out of Order delivery, congestion control similar to BBR. I wonder, this is not the same as UltraEthernet consortium is trying to achieve?

Multipath-tcp: The above probably works in “close” networks (managed by one entity) but maybe it is not going to work in the Wild internet. Still this looks still quite far from production. I believe this like QUID. Somebody like google deploys it and the rest jump in the wagon (more or less)

OCSP death: “OCSP is not making anyone more secure. Browsers are either not checking it or are implementing it in a way that provides no security benefits. “

AlphaChip: As far as I have read, designing chip is one of the most complex things and getting help from AI can even increase the advances in chip design. I read that NVIDIA had something similar. And this should be applied to ASICs too so networking is benefited

Vision Agent AskUI: need to try

ByteDance UI Agent – UI-TARS: as above

Crawl4AI: Interesting for digestion your local knowledge base sites and using with your local LLM….

Run your locally AI: I tried this in my work MacBook and it worked! I want to create an AI agent for a work project (actually i am dreaming to be able to achieve it….)

Open Web UI + Ollama: I tested this too in my MacBook and works like magic! You can even use DeepSeek 🙂

Bolt.diy + DeepSeek: I didnt manage to install bolt.diy ….

Training your AI: My idea is to get an open-source LLM trained with my data so I can use it to do my “job” But in the video there was too much publicity and I dont have access to a GPU… but I dont much data neither (or that’s what I think)

Bob Bowman (Michael Phelps coach): Show up, do the job.

Alchemy

Finished this book, and it is a gem! It is better than expected and very funny. The author is an advertising executive and the book goal is to teach you to think in a non-logical way. As an engineer, I feel that everything has to be logical and reasoned but Rory gives plenty of examples all over the book about logic not being always the only answer. I have highlighted so many lines that I think I can copy the book. It refers to behavioural economics (state agent showing 3 places, so you pick the ones they want you to pick, how to make a train join faster without faster trains (sol: make it more enjoyable)) and politics (Trump 1.0 victory, Brexit victory, etc)

So the two lessons from the book is to understand how the marketing works (as a consumer so you can understand it and it doesnt have to be logical…) and how you should be less logical so you can find different and still valid solutions.

Even this is mentioned to relationships, something that actually hit me…

Some videos from Rory (that I would like to watch)

The lost genius of irrationality

Behavioural Economics, Humas and Advertising.

Chocolate Salami

This is a typical sweet from Portugal too and I wanted to make it for some time, and this was triggered by chance via this video. And then I checked this one for getting an idea about the quantities

Ingredients

  • 250g maria biscuits (dont try to hard biscuits…)
  • 150g dark chocolate chopped
  • 50g cacao powder
  • 50g pistachio chopped
  • 70g butter

Process

  • Smash the biscuits, but dont make it powder!
  • Add the pistachios
  • Melt the chocolate and butter
  • Add to the biscuits
  • Mix well, like a dough
  • Pour the mix over clean film, try to roll like a chorizo
  • Let it cool down in the fridge.

Then you can decorate it like a salami if you wish

Obviously, it doesn’t look like in the videos (although it was tasty). My main mistake was using oat biscuits that are quite hard.

2nd Attempt. A bit better but still not as I would like it:

Chickpeas Chocolate Cake

I was curious about this cake and tried it.

Ingredients
800g Cooked chickpeas (2x cans)
200g peanut butter
120ml maple syrup
2 teaspoons of baking powder
2g salt
100g chocolate 90% chopped roughly

Decoration
150g melted chocolate 90%
150g peanuts crushed

Process:

  • Pre-heat oven at 190C
  • Put all ingredients together but the chopped chocolate in a food processor.
  • Once it is all mixed (it is like a very dense peanut butter), add the chocolate chips and mix with a spoon.
  • Pour the mix into a mold of 25x10cm with parchment paper.
  • Bake at 190C for 40-45m
  • While it cools down, prepare the decoration. Melt chocolate using “mary bath” (you can add a bit of butter to make more glossy).
  • Add the crashed peanuts to the chocolate and pour it over the cold cake. Spread in all sides.
  • Let it cool down again.

Before getting into the oven

After letting it cool down:

To be honest, it is not very sweet. I used 90% chocolate but still is tasty. So you can put a bit more maple syrup and 70-80% chocolate.

It goes great with a glass of milk or even whipped cream!

This cake is a protein bomb!

AusNOG 2024

A bit late to review, but some interesting talks. Agenda.

Arista: Practical AI Networking Innovations:

rail optimized
all reduce

low entropy, 2-3 flows per nic, elephant, bursty

JCT job completion time
TSN time spent networking

overprovisioning 1:1.2

Nokia: NUTS python network testing:

It looks quite nice, it is based on pytest and nornir/napalm. It looks similar to batfish?

Measuring Starlink:

High jitter, each 15sec change satellite -> micro-loss. BBR (non-loss sensitive) is the best flow protocol protocol with Starlink. ECN.

A Dangerous Fortune

I read this book a second time without realising. The first time was several years ago, I think it was one of first English books bough in paper. The second time was a cheap deal at 99p for the ebook.

And, I enjoyed it. The thrill, the twists, really engaging. It reminds to “The Pillars of the Earth” It is funny how this reminds me to all the bank crashes we had in 2008/9 and even last year in Silicon Valley. We don’t learn.

Taxtopia

This book showed me how screw up we are with the tax system, because it doesn’t apply to all. It is overcomplicated, and it you make a mistake, you pay dearly.

I think it is quite radical but at the beginning, the book gives a very radical example of a the tax system as a way to slow down inflation…. because governments have the printing money machine.

One important question is how you define a “company”. It has all the features as a person, it can buy/sell things, sue/be sued, but it is controlled but different people that are putting the money.

At the end, the author suggest to only tax people, remove all other taxes, you only pay if you get wealthier… but still dont fully understand how that could stop “rich”and companies not paying taxes. At the end of the day, all main governments are keeping all the tax heaven places so we are not going anywhere. Is blockchain a solution?

Different ways to pay less taxes is to borrow on your wealth and set up a trust… high level looks fine but can this be done by a “normal” person?

Some recommendations: blog, book

Photonics in Computing, Usb cable hack, Stutz, Building AI Networks Arista

Lightmatter: Based on this video, they are using photonics to connect chips, looks interesting, I remember Google has something with optical but for networking. But It is pretty clear this is not photonics computing.

Hacking USB cable: impressive, and expensive 🙂

Phil Stutz: Interesting conversation. But somehow, I am still looking for that thing that unlocks me…. can’t find it for the life of me….

Building AI Networks Arista:

- allreduce: collect elements from all nodes, apply a reduction operator(eg sum) then distribute reduction to all nodes
-allgather: collect elements from all nodes, and distribute the to all other nodes
- gpu: cpu for parallelization
- RDMA: RoCE2 GPU memory to GPU memory - origin in IB
- issues: flow collision, trafic polarization. low entropy!!! > dificult to ecmp => Dynamic LB
incast: many2one -> ECN + buffering (in spine!)
- use chassis!

With an operations hat on, dealing with chassis is expensive and no efficient. It kind of a vendor lock-in. AWS is all in pizza boxes and I remember one presentation in Cisco Live where the Cisco EVPN authority recommended pizza boxes.

Natural Born Heroes

After reading “After Born To Run”, I decided to buy another of the author’s book. Somehow I had low expectations… I read a bit the intro but somehow it has been beyond what I expected. It mixed history from Ancient Greece, World War II, fitness, nutrition, psychology, philosophy etc. It is a weird cocktel but it worked great!

It is quite brilliant how he tries to mix the fiction of classic Homer‘s Iliad with real facts and personal traits. And how Crete is the origin of all Greece history and heroes, and therefore Western Culture. As well, I never imagined Crete was so important during the WWII as it seems it delayed the invasion of Russia and then doomed the Germans. At the end of the day, the book is about the kidnapping of a Nazi general in Crete. And it is actually an Odyssey. I didnt about Churchill Dirty Tricks and all the characters involved. You dont need to born Rambo to be a war hero. Just check Paddy’s and Xan‘s.

Regarding fitness, he mentions the Fascia Lata and Fat-As-Fuel concepts. Mark Allen is mentioned one of the greatest triathlete that I didnt know and as a curiosity, he had a splash when he discovered the concept of Fat-As-Fuel concept (a.k.a Paleo diet). And the training and fat-as-fuel is mentioned from Phill Maffetone (and it seems he has a very interesting life)

He has a big go against Gym culture (Arnold Schwarzenegger being a target) as it destroyed what humans have done for millennia: natural movement. And the closest to that is parkour . It gives a good history lesson about the origin of Parkour and its figures like Georges Hebert. As well, the author has a go to the Gatorade-link industrie and hydration wasn’t really a problem in long distance running.

Regarding Fascia training he mentions Steve Maxwell (link1, video)

In general, very interesting and entertaining book with a wide range of topics!

Potato Pizza, TCP Conversation Completeness, IBM power10, AI developer kit, 2 not 3

This is a pizza that I tried several years ago, and it seems the restaurant is out of business. I have done some pizzas trying to emulate it but never matching that memory. So this is a placeholder to try:

Original Ingredients: Bechamel, Smoked Mozzarella, red onions, pancetta, sliced potatoes, and Buffalo Mozzarella.

Some example1, example2

This a old for today’s news. IBM Power10 Memory network but looks interesting:

...thanks to the coherence IBM already has across NVLink (which is really BlueLink running a slightly different protocol that makes the GPUs think DRAM is really slow but really fat HBM2, in effect, and also makes the CPUs think the HBM2 on the GPUs is really skinny but really fast DRAM).

Checking some wireshark traces last week, I cam across the concept of TCP Conversation Completeness. This was totally new for me. This video gave some idea too. This was useful for me for finding TCP conversation that showed retransmissions when trying to stablish the TCP handshake, and not just showing the retransmission, so I used “tcp.completeness<33” so I should see TCP flows with a sync + RST.

AI developer Kit by NVIDIA: This card looks nice, I would mind to buy it and give a chance to learn, but it is sold out everywhere…. This is a video about it.

2 not 3 (make a choice!):