Why no benchmarks against pg_vector, DuckDB with extension?

For benchmarks you may just prepare Phoronix Test Suite module (https://www.phoronix-test-suite.com/) to facilitate replication on a variety of machines.

Also compare with LKH3 which seems much faster and closer to optimal.

Unfortunately the default configuration poses significant security problems, allowing for sidestepping 2-factor authentication schemes, and sideloading of applications without user permission (!)

Can you elaborate?

Do they change IP numbers so often?

Oh, lord yes! Frequently they're scraping us from multiple distinct CIDR blocks simultaneously. But we can tell it's the same organization doing it not just because the requests look similar, but it's even possible occasionally to see a request for a search from one CIDR that's followed up immediately by requests for details for the products that had been returned by the search.

While at the same time, because our site is B2B ecommerce, where our typical customer is a decent-sized corporation, it's not uncommon for a single legit user to have consecutive requests originate from different IPs, as their internal proxies use different egress points.

In bussiness it is not called "losing", but "investment".

Because 8GiB is insufficient for most work in 2025, so cheap Macbook is simply better Chromebook at this point: https://videocardz.com/newz/pcgh-demonstrates-why-8gb-gpus-a...

And 16GiB VRAM insufficient for games:

https://videocardz.com/newz/pcgh-demonstrates-why-8gb-gpus-a...

Also, 3nm M4 is going head-to-head with older Ryzen AI 365 in everything except for power efficiency: https://nanoreview.net/en/cpu-compare/apple-m4-vs-amd-ryzen-...

When compared in multicore against Ryzen AI Pro laptops (high end), even Apple M5 are behind in the dust...

https://www.cpu-monkey.com/en/compare_cpu-amd_ryzen_ai_max_p...

Despite awesome progress with its latest ARM processors, Apple was caught behind Ryzen, and is threatened by next generation of Zen processors.

except real life performance is not about nm process (this is almost completely irrelevant) or benchmarks.

I load my DuckDB into RAM and I'm mindblown at the speed that I can run data analysis on it due to RAM bandwidth and the fact that it's SoC, so almost no bottleneck between RAM and the CPU. Dozens of seconds on a $5k PC compared to milliseconds on a $2k Mac.

Not to mention that Windows 11 is a very weird OS.

Linux is great for some usecases, but doesn't utilize the hardware in the same way as other OSs.

I have all OSs at home and the difference is stupendous.

If the task takes "dozens of seconds" on the PC, there is no way it is shortened to millisecond on the Mac. Unless you purposefully make it a bad comparison to favor the Mac.

I'm going to call it and say that you are full of shit. The improvements seem to come from the software: https://duckdb.org/2024/06/26/benchmarks-over-time

If you were to benchmark 2 current comparable machines, the difference would be negligible.

How much faster or slower than 400 lines compiler? https://news.ycombinator.com/item?id=26191243

Fails to recover after the first parse error.

While I admire social-network-friendly websites, I am afraid that performance is too bad to allow its use instead of a social network.

Indeed it is so performance sensitive, that it has blocked my region. Would it not be better to get a static site generator from a standard Markdown posts, and thus assure it is both performant and accessible?

Inventions of SpaceX made space trips much cheaper: https://spaceinsider.tech/2023/08/16/how-much-does-it-cost-t...

Now it reaches just 2k$ per kilogram.

Hotel I’m staying in Fiji is using starlink for about 30 rooms. It works incredibly well.

There’s a fiber there too, but I assume someone is so incompetent to make it reliable and affordable that a $500 dish is able to serve entire resort.