ClarkThan

(This is a translation of the original article in Japanese by moratorium08.)

(UPDATE (22/3/2019): Added some corrections provided by the original author.)

Writing your own OS to run on a handmade CPU is a pretty ambitious project, but I've managed to get it working pretty well so I'm going to write some notes about how I did it.

ClarkThan

type term =
  | Lam of (term -> term)
  | Pi of term * (term -> term)
  | Appl of term * term
  | Ann of term * term
  | FreeVar of int
  | Star
  | Box

let unfurl lvl f = f (FreeVar lvl)

ClarkThan

#include <iostream>

#ifdef __AVX__
  #include <immintrin.h>
#else
  #warning No AVX support - will not compile
#endif

int main(int argc, char **argv)
{

ClarkThan

const std = @import("std");
const builtin = @import("builtin");

pub const Config = struct {
    /// Whether to synchronize usage of this allocator.
    /// For actual thread safety, the backing allocator must also be thread safe.
    thread_safe: bool = !builtin.single_threaded,

    /// Whether to warn about leaked memory on deinit.
    /// This reporting is extremely limited; for proper leak checking use GeneralPurposeAllocator.

ClarkThan

type binder = Lam | Pi

(* The syntax of our calculus. Notice that types are represented in the same way
   as terms, which is the essence of CoC. *)
type term =
  | Var of string
  | Appl of term * term
  | Binder of binder * string * term * term
  | Star
  | Box

ClarkThan

Memory Optimization (Christer Ericson, GDC 2003)
http://realtimecollisiondetection.net/pubs/GDC03_Ericson_Memory_Optimization.ppt

Cache coherency primer (Fabian Giesen)
https://fgiesen.wordpress.com/2014/07/07/cache-coherency/

Code Clinic 2015: How to Write Code the Compiler Can Actually Optimize (Mike Acton)
http://gdcvault.com/play/1021866/Code-Clinic-2015-How-to

ClarkThan

Achieving warp speed with Rust

Contents:

• Number one optimization tip: don't
• Never optimize blindly
• Don't bother optimizing one-time costs
• Improve your algorithms
• CPU architecture primer
• Keep as much as possible in cache
• Keep as much as possible in registers

ClarkThan

#![feature(test)]
extern crate test;

#[cfg(test)]
mod tests {
    use super::*;
    use std::fs::{File};
    use std::io::{BufWriter, Write, BufRead, BufReader};

    #[bench]

ClarkThan

Tucan - Bibliography

Majority of the resources I used to build Tucan, my toy optimizing compiler in Rust. This list is not complete but most of the things listed here are things I really read through and used.

Books

• Engineering a compiler (I use this a lot! For SSA, dominance and optimizations)
• [Static Single Assignment Book][ssabook] (I use this a lot!)
• Types And Programming Languages

ClarkThan

Latency Comparison Numbers (~2012)
----------------------------------
L1 cache reference                           0.5 ns
Branch mispredict                            5   ns
L2 cache reference                           7   ns                      14x L1 cache
Mutex lock/unlock                           25   ns
Main memory reference                      100   ns                      20x L2 cache, 200x L1 cache
Compress 1K bytes with Zippy             3,000   ns        3 us
Send 1K bytes over 1 Gbps network       10,000   ns       10 us
Read 4K randomly from SSD*             150,000   ns      150 us          ~1GB/sec SSD