A Programming Language for the Next 40 Years: (Rust)

The CPU’s that we use for computers used to roughly double in speed (transistor density) every two years according to something called Moore’s law. This stopped happening around 2013/2014. Some innovations are being made, but right now progress is stagnating. There are a few implications:

1: You can probably keep your old laptop around for longer

2: Programmers have to start writing better, faster code to do more interesting things on single CPU’s

3: For really complicated stuff, we’ll just start trying to use as many CPU’s in parallel as possible (Machine Learning with multi hundred core GPU‘s, server farms, multi-threading, etc.)

For point number 2: Programmers are going to have to start coding closer to the metal. This is possible with C and C++, but the complexity of dealing with pointers and memory safety is difficult to do well consistently with complex programs.

Enter Rust: “a language for the next 40 years” that promises readability similar to Java, run-time similar or sometimes better than C code, and memory safety.

Related:

The Worst Piece of Code I’ve ever Written
Krupczak logo

Home

AI and GPU Silicon Supremacy

NVIDIA has held a clear market and technical lead over AMD with video game graphics cards (GPU’s) in the past few years, having managed to get a chip design a few generations comfortably ahead of AMD. This market dominance, combined with increased demand for graphics cards beyond gaming in applications such as AI and Crypto has allowed NVIDIA to raise prices.

Right now, much of the emerging AI research coming out is being done on NVIDIA GPU’s, but trying to get compute time on these cards can seem like trying to get mainframe time back in the early days of computing (expensive and hard to find the hardware)

Intel ‘s prior acquisition of AMD’s top graphics chip talent and imminent plans to release graphics cards may soon usurp NVIDIA’s dominance, making existing and new more complex AI cheaper to implement.

Now’s a great time to be thinking about intellectual property, in the hopes that implementation will get cheaper in the future

https://www.tomshardware.com/news/intel-xe-gpu-specs-features,38246.html

More:

Toe in the Water: Examining Deep Learning and Data Science
Book Club 1.5
Krupczak logo

Home

Building a Fault-Tolerant Drive Pool with Centos, ZFS, and RAIDZ

  • Shucking
  • Installation
  • tape to cover deprecated 3.3v pins so drive doesn’t receive a “shutdown” signal
  • Setup
  • Monitoring temps during burn-in

Summer break has come, and that means it’s time for personal projects! To get ready, I’ve upgraded the storage array on my home server from 4TB (actual) to ~32 TB (raw). I’m using sub-consumer hardware by taking apart four 8TB WD easystore backup drives to cut down on cost, and I’m using zfzonlinux with CentOS to build a fault-tolerant drive pool with raidz so that if one disk fails, my data will remain intact. ZFS also provides many nice features for snapshotting, data integrity, etc. and the raidz tends to perform better than typical raid solutions. Now that it’s completed, I have about 21TB of space to work with.

Krupczak logo

Home

Cooperative Deep Learning in Quake Capture the Flag:

CTF pic png

Cooperative Deep Learning in Quake CTF:

Google’s deepmind team made some big waves recently beating the world champion at GO, a famously hard board game. They’ve since made a major advancement in agents that can play a modified version of Quake III Capture the Flag.

Capture the Flag (CTF) is a popular first-person shooter game mode where players work in teams to steal a flag from the opposing team’s base while trying to avoid being “tagged”. It requires unspoken teamwork, planning, strategy, and real time interaction with the environment, tasks hard for AI.

The agents play against themselves to train, then play against humans for competition. Just like humans, they’re only allowed to look at the screen and the game score. To make things fair, the AI was given a reaction time and aiming accuracy similar to that of humans.

Even with these handicaps, the AI teams outperformed skilled human teams significantly by cooperating more and having a better sense of strategy.

Humans were still better in some areas though, and when AI teams were augmented with human players they cooperated flawlessly and often outperformed AI-only teams.

Human teams may work with AI augments in many areas in the future. This may already be happening in military applications.

More info:

deepmind.com/blog/article/capture-the-flag-science

arxiv.org/abs/1807.01281

More:

Toe in the Water: Examining Deep Learning and Data Science
AI and GPU Silicon Supremacy
Krupczak logo

Home