Tempest
100M Challenge
Parse a dataset of 100 million page visits from CSV into a sorted JSON file.
Fork the repo, implement Parser::parse(), and submit a PR.
How fast can you go?
Sponsored by JetBrains PhpStorm & Tideways
Top 3 multithread performers based on the latest benchmark run. Positions may change until the challenge ends. Single-thread has a separate prize.
Performance improvements over time for the top contenders.
All submissions ranked by execution time.
| Rank ▲ | Participant ▲ | Time ▲ | Submitted ▲ |
|---|
Rules, prizes, and server specifications.
Parse a dataset of 100 million page visits from CSV into a sorted JSON file using pure PHP. Each entry maps a URL path to daily visit counts, sorted by date ascending.
Fork the repo, implement Parser::parse(), and submit a PR. Results are benchmarked on a dedicated Mac Mini M1 with 12GB RAM.
Solutions must be written in PHP only. No FFI, no external tools, no internet access. Your script may only work within the project directory.
The solution must validate against the real dataset. Improvements of 0.01s or more are counted. Ties are broken by submission time.
Mac Mini M1 with 12GB available RAM. JIT is disabled. OPcache is disabled.
Each submission is run individually on the dedicated server to prevent bias. Results are manually verified before benchmarking.