dcl's comments

On: Understanding LucasArts' iMUSE System

[dcl]

You have awakened some incredible memories. I know exactly what you are talking about.

On: New material gives copper superalloy-like strength

[dcl]

Rearden metal...?

[sanex → dcl]

Not as strong as steel, real or imaginary.

[nine_k → sanex]

Milder steels have yield strength in the 200-300 MPa range, while this alloy reaches nearly 1000 MPa.

[AngryData → sanex]

Well not as strong as the best steels, but still stronger than many common steels. Even some less special bronze alloys can beat common steels in strength.

[mitthrowaway2 → dcl]

That's a copper-iron alloy.

On: 4chan Sharty Hack And Janitor Email Leak

[dcl]

thank you for this laugh

On: DolphinGemma: How Google AI is helping decode dolphin communication

[dcl]

Wouldn't have even been a believable April fools joke even 5 years ago.

On: Ask HN: Do you still use search engines?

[dcl]

Search engines are still required for me. LLM's still get lots of very important things wrong.

Last night, I asked Claude 3.7 Sonnet to obtain historical gold prices in AUD and the ASX200 TR index values and plot the ratio of them, it got all of the tickers wrong - I had to google (it then got a bunch of other stuff wrong in the code).

Also yesterday, I was preparing a brief summary of forecasting metrics/measures for a stakeholder and it incorrectly described the properties of SMAPE (easily validated by checking Wikipedia).

I constantly have issues with my direct reports writing code using LLM's. They constantly hallucinate things for some of the SDK's we use.

[throwaway422432 → dcl]

Asking for a list of companies in a specific sector also gives you made up tickers, or at best a list it found on some blog.

Was a bit more useful at questions like "Rank these stocks by exposure to the Chinese market", as you can prioritise your own research but in the end you just have to go through the individual company filings yourself.

On: The Llama 4 herd

[dcl]

But how good is it at Pokemon?

On: Recent AI model progress feels mostly like bullshit

[dcl]

I like this bit:

> Personally, when I want to get a sense of capability improvements in the future, I'm going to be looking almost exclusively at benchmarks like Claude Plays Pokemon.

Definitely interested to see how the best models from Anthropics competitors do at this.,

On: The race is on to build the most complex machine

[dcl]

6 months?

On: Fastplotlib: GPU-accelerated, fast, and interactive plotting library

[dcl]

I always thought it was interesting that my modern CPU takes ages to plot 100,000 or so points in R or Python (ggplot2, seaborn, plotnine, etc) and yet somehow my 486DX 50Mhz could pump out all those pixels to play Doom interactively and smoothly.

[sieste → dcl]

This SO thread [1] analyses how much time ggplot spends on various tasks. Not sure if a better GPU integration to produce the visual output would help speed it up significantly.

[1] https://stackoverflow.com/questions/73470828/ggplot2-is-slow...

[kkoncevicius → dcl]

From R side i think this is mainly because ggplot2 is really really slow.

Base R graphics would plot 100,000 points in about 100 milliseconds.

    x <- rnorm(100000)
    plot(x)

A quick benchmark with writing to a file:

    x <- rnorm(100000)
    system.time({
      png("file.png")
      plot(x)
      dev.off()
    })

     user  system elapsed
    0.179   0.002   0.180

[stackedinserter → dcl]

Nobody cares about optimization for relatively big datasets like million points, maybe it's not a very popular use case. Even libraries that do able to render these datasets, do that incorrectly e.g. skip peaks, show black rectangles instead of showing internal distribution of noisy data, etc.

I ended up with writing my own tool that's able to show millions of points and never looked back.

On: Microsoft Windows 1.0 with Steve Ballmer (1986)

[dcl]

He kept more of his MSFT stock.