5ersi

Long time casual wine drinker (and occasional beer), I find myself drinking less and less. Two reasons: first drinking a glass of average wine for the sake of having a drink makes no sense anymore. I prefer to have a drink more seldom but pick a better wine, that I actually enjoy. Second, as I get older I see that even a modest amount of alcohol has negative effect on me: during the day I get groggy, during the night my sleep is worse (I know, counterfactual, but so it is).

For a true high performance you should use thread busy-spinning, CPU pinning and SPSC/MPSC ring buffers.

MCPs are diminishing in value a bit, because AI Agents are getting smarter about using API/CLIs. For example, I use gh cli via Claude instead of their MCP, because I already had cli setup, so no need to use MCP.

MCPs can potentially have great value if they cross multiple sources and combine results. For example at work we use an in-house MCP for log/metrics search across five different (legacy) systems. It finds correlation across events in different system within minutes.

Just bought AMD AI 395+ 64GB RAM for 2000eur. Mac mini would be 2x for about similar performance. Pass.

UltimaDB, a high performance full-featured MVCC transactional embedded database, with special focus for use on SMR clusters.

UDB has full transactions, selectable persistence (in-memory, eventual, strict) and consistency (SI, SSI). A lot of time was spent on performance.

Written in Rust, with 8k line of code and 15k lines of tests, 96% coverage. Formal verification with Lean 4 is in progress (see /formal), also consistency checking with elle/jepsen is in progress.

It uses phase change (solid to liquid) to store heat at about 200 kJ/kg. Compare this to heating water in a boiler from 10c to 60c - stores 209 kJ/kg.

So we already have an effective way to store heat which can work for decades without servicing and is also cheap to produce (in terms of money and energy consumption).

Launching alone consumes about 75-150kWh per tonne of energy for fuels only (as per ChatGPT).

Planned lifespan of Starlink satellites is 5years.

This AI 8GB HAT is $150, while better performing Pi5 8GB is $100. Makes no sense to buy the AI HAT.

Java?

Is supported on more platforms, has more developers, more jobs, more OSS projects, is more widely used (Tiobe 2024). Performance was historically better, but c# caught up.

LLMs are learning in a very similar way humans are learning. So if humans can read a text (or view a video), learn from it and then use the knowledge to produce something, so can LLMs?

Copyright laws have quite strict rules on what constitutes a copy, and this was tested in courts many times. This rules also apply to works produced by LLMs.