mpickering / languagecomparison

Understand the performance differences between several languages, and illustrate semantic differences between the languages.

Language Performance Comparison

Purpose

Understand the performance differences between several languages, and illustrate symantic differences between the languages.

Languages being compared:

1. C#
2. Rust
3. PureScript
4. Haskell
5. Elixir

Performance Comparison

• 1.2M Accounts
• 10M Transactions

Machine	C#	Rust	PureScript	Haskell	Elixir
Alfredo	42s	18s	378s	156s	185s
Dave	49s	20s	334s	184s	322s
Relative (Apprx.)	1	½	8	3.5	5
(Memory)	4 GB	1.6 GB	x	12 GB	3 GB

Potential Next Steps

To run a dockerized version of these tests

./docker-build-test-images.sh
./docker-test-runner.sh

On linux it looks like atop might be a good tool

• https://haydenjames.io/use-atop-linux-server-performance-analysis/
• https://www.atoptool.nl/

C#

• Pros
  • Performance without optimization
• Cons
  • Large holes in logic - potential failure points
• Code
• Running: dotnet build; time dotnet run

Rust

• Pros
  • Correctness
  • Tool ecosystem
• Cons
  • Borrow checker / need to understand memory layout
  • Lack of libraries
• Code
• Running: cargo build --release; time cargo run --release

PureScript

• Pros
  • Correctness
• Cons
  • Performance concerns on large datasets (not because of Node)
• Code
• Running: `pulp build; time node -e "require('output/Main').main()"

Haskell

• Do Account\Transaction problem
• Look at record syntax
• What kind of speed to we get?
• Code
• Running: stack build --fast; time stack run

Elixir

• Pros
  • Concurrent
  • Fault-tolerant
  • Clusteable
• Cons
  • Dynamic
  • Static types (Dyalizer) are an optional add-on
• Code
• Running: mix deps.get; mix clean; MIX_ENV=prod mix release;mv _build/prod/rel/test_elixir . ; ./test_elixir/bin/test_elixir foreground