Vegan Chocolate Tart

Last weekend made some vegan desserts for a friend. This recipe was really good. Much better than I expected.

Ingredients:

Base:

  • 350g vegan biscuits (ie Lotus Biscoff)
  • 2 tbsp cocoa powder
  • 1 tbsp plain flour
  • 1 tbsp maple syrup
  • 170g margarine + extra for greasing

Filling:

  • 2x400g can coconut milk, CHILLED!!!
  • 200g dark chocolate
  • 1 tbsp maple syrup

Process:

  • Pre-heat oven at 200C.
  • Crush biscuits to crumbs (I used my food processor).
  • Add the cocoa powder, flour, syrup and margarine to the biscuits. Mix and create a dough.
  • Butter with margarine the bottom of a 23cm aprox tart tin.
  • Spread the biscuit dough into the the tin. Press it carefully so it is thick and consistent.
  • Bake in the oven for 10-12 minutes. Then let it cool, VERY IMPORTANT!!!
  • Make the filling. Open the coconut cans (I put them in the freezer for a bit). They should have separated the coconut water from the cococut cream. Make a hole with a knife and save the water. You can use it later for cereals or drink it directly.
  • Put the solid coconut cream in a bowl and move to the next step
  • Melt the chocolate using bain-marie or microwave.
  • Add the melted chocolate and syrup to the coconut cream. And whisk until fully combined.
  • Pour the filling over the base ONLY if the base is properly cool down. Spread it evenly and put it in the fridge for at least 3-4 hours to set.
  • Optional you can decorate the top with fresh red berries.

This is the result. Very tasty and not very sugary.

Sardinia – Dolomites

Visiting Sardinia I found a type of bread that I didnt expect: pane carasau, that is a flat, crunchy bread. You could find it in restaurants and supermarkets! As well, the typical sweets were very good. I was lucky to try some of them from this local pasticcerie, and reminded me to the “old” classics where people used only the ingredients they had around. I tasted papassini, gueffus (like marzipan), pardulas (small ricotta based pies), etc. Very good all of them

And I was very lucky to trek with a friend through the Dolomites. Really, really nice. It was a bit harder than I expected, but totally worth it! We did the Alta Via del Granito, the “giro verde”. A bit more info about that route.

Stolen Focus

This is a book I wanted to read because it is very important subject for me. I want to be more efficient and better at whatever I want to do and to be honest, sometimes I find it difficult to sit down and try to learn something new with my computer. As well, I dont like unnecessary distractions, mainly at work…. As usual, I expected to find the magic pill to get focus. But the book was better than that, it was the author’s journey from not having focus to understand why we are in this situation and the options we have. Spoiler alert: there is no magic pill, and it is not easy (neither impossible). At some points looks quite dramatic, but I see the point.

  • Multitasking – This is a lie we tell ourselves. We “can” do several “easy” things at the same time, only. That reminds me a sentence from a person I respect highly: ATP “Make one change at a time”.
  • Flow: This chapter was mainly based on the book “Flow” from Mihaly Csikszentmihalyi. Summary to get into “flow” mode you need: Clear goals, meaningful goals and at edge of own ability.
  • Sleep: Very important for hour healt and sanity. As well, I liked the concept of less sleep => more consume. And that reminded to something I heard once that the Netflix CEO said the only competitor of Netflix was sleep….
  • Wandering mind: Sometimes we think that is bad, but depends on the circunstancies. Meditation is a good things it calms the mind. But the point here, using Poincare example resolving an important problem, for having focus, we need to give space to our mind too.
  • Aza Raskin (infinite scroll) and Tristan Harris (Google). They are two important figures from the tech work showing the devils of social media / big tech and how works inside. These companies they want you as much as possible in their platform. And that is not always the best for you! It was interesting to read about “positive reinforcement” that is the main practice to get you hooked to social media. This was developed by BF Skinner. As well, there is a reference that one of the founder of Instagram was in the class of Skinner. As well, I liked the reference to Magic (powa.academy) as it works tricking your focus!
  • These companies harvest so much data… that is easy to polarize people. Examples in Brazil (Bolsonaro supporters shouting “Facebook!!” and USA (Trump) elections.
  • Nir Eval: I read his book “Indistractable” and I follow several of his pieces of advice. He sais we have to adapt to the new circunstances. We need to find the distraction triggers. And all is about avoidance. I was surprised with the author showed the conflicts between Nir (tech is ok, it is our fault we got distructed) vs Tristan (tech is bad). As well, it was interesting to read Nir’s background in the tech industry. I dont think all tech is evil but I am clear about something. If I dont pay for a service (gmail, youtube, etc) then I am the currency (aka: If you are not paying for the product, then you are the product). So you have to be conscious about what you are using it for.
  • How to change things: change business model, instead of getting money from advertisers, you pay for the service (although dont think that would world outside the rich countries), so they will server you! not the advertisers. Government regulation. I liked the concept of “Surveillance Capitalism” that is based on “human (psychology) hacking”
  • You get focus when in a safe environment. If not, you are always in a stress mode. I liked the story of Nadine Burke Harris and her work to get kids on track in school from difficult neighbourhoods. There is a point to the “Universal Basic Income” concept that I think it is great.
  • Work hours: longer is not better. More focus, more efficiency. Win/win for employer and employees. Again this is something that is not going to work for everybody, just to office workers. Although I fully agree with it. There is too much stupidity in the office culture….
  • Food: This is our source of energy, and with the wrong diet, we can get focus. The evil of ultra processed food. There is a mention to Michael Pollan for cook books.
  • Pollution: This affects too our capacity of attention.
  • ADHD: attention illness in kids. There is a huge diagnosis of ADHD and that means a lof of prescription pills. The book mention that is not always and illness that is most of the time the environment, mainly for kids.
  • Play! It is critical for kinds to spend time outside and without supervision to be able to develop personal skill and thought process. The education system and society is getting worse at this. And I agree, I always remember when I was a kid that my best time was when going to my hometown where I could be outside all day. In the big city, this was very difficult, although I was lucky to have a football group and a small park where I spent a lot of time too!
  • Intrinsic Values: Social media is mainly based on showing-off. And we lost track of what is really important. Having the “correct values”for their own sake get us off that wheel that doesnt give us anything and always keep us unfulfilled.
  • James William (Google) mentions that there are three types of attention: spotlight, startlight and daylight.
  • Constant economy growth: we need to consume more because there are no more markets! You need to do more in the same amount of time. Professor Thomas Hylland Eriksen. Steady-state economy: different goals. Pushing our mind and nature to the limit. W.H. Auden “We must love one another, or die” WWII.

In general, it is a good book that gives you a lot of information about this problem and show you examples of people of fighting it from all points: from the 4 days work week, freedom to play for kids, etc.

The author says he hasnt solved it but he has learned a lot in this journey.

The Genetic Lottery

I finished this book today. And to be honest I have struggled a bit with it from the scientific aspect to the social one.

DNA is our instruction set but we depend on the our environment too for developing those instruction. But that DNA is a random mix from our parents. So we could look like them but we are not them, we are unique. Each person is unique.

It is difficult to accept that whatever I have achieved is based on genes and environment, and luck. Many times I tell myself when looking at the mirror that I am the luckiest person in the world, so I came into terms with that point. As well, that doesn’t mean that we are pre-programmed and there is nothing to do, that there is no “free” will. This is a tough philosophical topic and again, I got a hard time reading about it in the book but the author says there is elbow room between our genes/environment and what we become or do with it. So it was a bit of a relief as your ego is not totally destroyed.

People vary in ability, energy, heath, character, and other socially important traits, and there is good, although not absolutely conclusive, evidence that the variance of all these traits is in part genetically conditioned. Conditioned, mind you, not fixed or predestined”

Theodosius Dobzhansky

So in the book, there is a strong emphases to the fact that the genetic lottery defines much of the inequality in the society we live. We live in the system where educational success, work success, etc follows one standard. The book wants to change the idea that one system fits all is not possible, as we are all different so we need/have different ways to evolve, learn, etc. So for achieving a more fair society we need to provide a different education method to children that can’t learn/develop in the “standard” way. For doing that, we need to have a better understanding of our DNA. And obviously that is a bit scary because it can be misused by companies, governments, etc. That means a change in mentality in social politics. It is kind of being more “socialist” instead of a more cut-through capitalist society. And I think that makes sense, the social improvements we have from the “socialist” politics like free education, free health system, holidays, worker rights, etc has improved our societies compared with the earliest ones from the Industrial Revolution. But at some point, the inequality gap is increasing again. So accepting that genetics is a lottery and we need a new approach to close that gap, is the first step.

These are two pictures that helped me to understand what they book was trying to achieve. Equality is giving to all, the same. Equity is given each one what it needs, and that is win-win situation for all.

The author refers to three types of positions when dealing with social policies:

  • eugenic: we are defined by our genetics and we should do nothing to change it.
  • genome-blind: ignore genetics difference, waste time/money without really improving the inequality gap (or making it worse)
  • anti-eugenic: use genetic data to search for effective processes that improve people’s live and reduce inequality in society.

Something that surprised me is the mention that there are deaf couples that wanted their children deaf as they dont see deafness as defect, and they used genetic help for that. I struggle to accept that is correct.

Still got the feeling that I am not explaining all properly or if I have understood everything properly. I need to take notes, highlight things, etc. How difficult is to have a pencil around when most of the time I am reading at home 🙂

Infiniband Essentials

NVIDIA provides this course for free. Although I surprised that there is no much “free” documentation about this technology. I wish they follow the same path as most networking vendors where they want you to learn their technology without much barriers. And it is quite pathetic that you can’t really find books about it…

The course is very very high level and very very short. So I didnt become an Infiniband CCIE…

  • Intro to IB

— Elements of IB: IB switch, Subnet Manager (it is like a SDN controller), hosts (clients), adaptors (NICs), gateways (convert IB <> Ethernet) and IB routers.

  • Key features

— Simplify mgmt: because of the Subnet Manager

— High bw: up to 400G)

— Cpu offload: RDMA, bypass OS.

— Ultra low latency: 1us host to host.

— Network scale-out: 48k nodes in a single subnet. You can connect subnets using IB router.

— QoS: achieve loss-less flows.

— Fabric resilience: Fast-ReRouting at switch level takes 1ms compared with 5s using Traffic Manager => Self-Healing

— Optimal load-balancing: using AR (adaptive routing). Rebalance packets and flows.

–MPI super performance (SHARP – scalable hierarchical aggregation and reduction protocol): off-load operations from cpu/gpu to switches -> decrease the retransmissions from end hosts -> less data sent. Dont really understand this.

— Variety of supported topologies: fat-tree, dragonfly+, torus, hypercurve and hyperx.

  • Architecture:

— Similar layers as OSI model: application, transport, network, link and physical.

— In IB, applications connect to NIC, bypass OS.

— Upper layer protocol:

— MPI: Message Passing Interface

— NCCL: NVIDIA Collective Communication Library

— iSEB: RDMA storage protocols.

— IPoIB: IP over IB

— Transport Layer: diff from tcp/ip, it creates an end-to-end virtual channel between applications (source and destination), bypassing OS in both ends.

— Network Layer: This is mainly at IB routers to connect IB subnets. Routers use GID as identifier for source and destinations.

— Link Layer: each node is identified by a LID (local ID), managed by the Subnet Manager. Switch has a forwarding table with “port vs LID” <- generated by Subnet Manager. You have flow-control for providing loss-less connections.

— Physical Layer: Support for copper (DAC) and optical (AOC) connectors.

AI Supercomputer – NVLink

So NVIDIA has an AI supercomputer via this. Meta, Google and MS making comments about it. And based on this, it is a 24 racks setup using 900GBps NVLink-C2C interface, so no ethernet and no infiniband. Here, there is a bit more info about NVLink:

NVLink Switch System forms a two-level, non-blocking, fat-tree NVLink fabric to fully connect 256 Grace Hopper Superchips in a DGX GH200 system. Every GPU in DGX GH200 can access the memory of other GPUs and extended GPU memory of all NVIDIA Grace CPUs at 900 GBps. 

This is the official page for NVlink but only with the above I understood this is like a “new” switching infrastructure.

But looks like if you want to connect up those supercomputers, you need to use infiniband. And again power/cooling is a important subject.

Jamaican Rum Cake

I have been lucky to try some Jamaican Rum Cake brought from Jamaica so I decided if I could make it myself. I found some recipes online like this (my main source) and this.

Ingredients:

  • 200g butter at room temperature + a bit for greasing
  • 1 cup of brown sugar
  • 4 eggs
  • 1 tbsp lime juice
  • 1 tbsp lime zest
  • 1 cup of blended fruits: raisins, cherries, mixed fruit, etc. Pre-soak the fruits earlier with water and a bit of white rum.
  • 1 tsp vanilla paste
  • 1 tbsp almond liquor
  • 1/4 cup of white rum + a bit for brushing
  • 1/2 cup of Port wine (I dont have red label / sweet red wine)
  • 1 cup plain flour
  • 1 tbsp cinnamon
  • 1 tbsp mixes spice
  • 1 tbsp gratted nutmeg
  • 1.5 tbsp baking powder
  • 1/2 cup bread crumbs
  • 3 tbsp black treacle (I dont have “browning liquid”)

Process:

  • Pre-heat oven at 180C. Grease a cake tin.
  • Cream butter and brown sugar in a bowl. Use a wooden spoon initially and then you can use this whisk. The video use an electric whisker but I think I managed a decent mixture. You want something creamy and fluffy.
  • In another bowl, mix the eggs, lime juice and lime zest.
  • Add the egg mix to the butter mix bit by bit, whisking constantly.
  • In another bowl, mix the blended fruit, vanilla, almond liquor, rum and Port.
  • And the fruit mix to the butter/egg mix bit by bit, whisking constantly.
  • Clean one of the bowl. Add the flour, cinnamon, mixes spice, nutmeg, baking powder and bread crumbs. Mix well.
  • Add the flour to the wet mix, bit by bit and mixing constantly.
  • Finally, add the black treacle that should bit the dark color to the cake. Mix well.
  • Pour the cake mix into the tin. Shake until level.
  • Put a small bowl with water in the oven or spray with water the oven to create extra moisture.
  • Bake for 1h 15m aprox. Remove from oven only when a skewer comes out clean from the center of the cake.
  • Once you take the cake from the oven, brush it with white rum while hot.
  • Leave it cool down for 1h.

The real thing:

My thing:

To be honest, although my version doesnt look like the original one, it was tasty. I think these were my errors:

  • I didnt soak the dry fruits so they didnt blend properly. I need to find more info about how to prepare this part properly. I think this is the reason the cake is not as “dense” as the original.
  • I think I over baked it. I lost track of time and it was 1h 30m I think.
  • The black treacle doesnt give the same black color as in the video. Or I need to put more?
  • Use more water in the oven. The first video didnt use any but the second did it so I though the second version was more moist and I wanted that.
  • Although I didnt use Jamaican rum neither Jamaican red wine, the taste was good.

Jericho3-vs-Infiniband

Jericho3 is the new chip from Broadcom to take into NVIDIA infiniband. From that article, I dont really understand the “Ramon3” fabric. It seems it can support 18 ports at 800G (based on 144 serdes at 100G). It has 160 SerDes (16Tbs) for uplink to Ramon3. The goal is to reduce the time the nodes wait on the network so it is not just (port to port) latency. Based on Broadcom testing swapping a 200Gb Infiniband switch with a Jericho3 is 10% better. As well, dont understand what they mean by “perfect load balancing” because the flow size matters (from my point of view) and “congestion free”. Having this working at scale… looks interesting…

But then we have the answer from NVIDIA: spectrum-X. So it is Spectrum-4 switches, with Bluefield3 DPU and software optimization. This is an Ethernet platform. Spectrum-4 looks very impressive definitely. But this sentence, puzzles me “The world’s top hyperscalers are adopting NVIDIA Spectrum-X, including industry-leading cloud innovators.” But most of links I have been reading lately are saying that Azure, Meta, Google are using Infiniband. Now NVIDIA says top hyperscales are adopting Spectrum-X, when Spectrum-4 started shipping this quarter?

And finally, why NVIDIA is pushing for Ethernet and Infiniband? I think this is a good link for that. Based on NVIDIA CEO, Infiniband is great and nearly “free” if you build for very specific application (supercomputers, etc). But for multi-tenant, you want Ethernet. So that kind of explains why hyperscalers likeAWS, GCP, Azure want at the end of the day Ethernet, at least for customers access. At the end of the day, if you have just one (commodity) network, it is cheaper and easier to run/maintain. You dont have a vendor lock like IB.

Will see what happens with all these crazy AI/LLM/ML etc.

AMD MI300 + Meta DC

Reading different articles: 1, 2, 3 I was made aware of this new architecture of CPU-GPU-HMB3 from AMD.

As well, Meta has a new DC design for ML/AI using Nvidia and Infiniband.

Now, Meta – working with Nvidia, Penguin Computing and Pure Storage – has completed the second phase of the RSC. The full system includes 2,000 DGX A100 systems, totaling a staggering 16,000 A100 GPUs. Each node has dual AMD Epyc “Rome” CPUs and 2TB of memory. The RSC has up to half an exabyte of storage and, according to Meta, one of the largest known flat InfiniBand fabrics in the world, with 48,000 links and 2,000 switches. (“AI training at scale is nothing if we cannot supply the data fast enough to the GPUs, right?” said Kalyan Saladi – a software engineer at Meta – in a presentation at the event.)

An again, cooling is critical.

Fat Tree – Drangonfly – OpenAI infra

I haven’t played much with ChatGPT but my first question was “how is the network infrastructure for building something like ChatGPT?” or similar. Obviously I didnt have the answer I was looking for and I think i think ask properly neither.

Today, I came to this video and at 3:30 starts something very interesting as this is an official video as says the OpenAI cluster built in 2020 for ChatGTP was actullay based on 285k AMD CPU “infinibad” plus 10k V100 GPU “infiniband” connected. They dont mention more lower level details but looks like two separated networks? And I have seen in several other pages/videos, M$ is hardcode in infiniband.

Then regarding the infiniband architectures, it seems the most common are “fat-tree” and “dragon-fly”. This video is quite good although I have to watch it again (or more) to fully understand.

These blog, pdf and wikipedia (high level) are good for learning about “Fat-Tree”.

Although most info I found is “old”, these technologies are not old. Frontier and looks like most of supercomputers use it.