Skip to content

benchmarks

Directory actions

More options

Directory actions

More options

Latest commit

 

History

History

benchmarks

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
README.md
README.md
 
 
ULTRA-FAST-MODE.md
ULTRA-FAST-MODE.md
 
 
memory.ts
memory.ts
 
 
redis.ts
redis.ts
 
 

README.md

Benchmarks

Performance benchmarks for ts-cache comparing with popular alternatives.

Platform: Apple M3 Pro @ 3.53 GHz Runtime: Bun 1.2.24 (arm64-darwin) Tool: mitata

Quick Results

Memory Cache Performance

Library GET SET Notes
ts-cache (ultra-fast) 3.90 ns 🥇 30 ns 🥇 No stats, no events, no TTL, Map storage
lru-cache 12 ns 40 ns Industry standard LRU cache
ts-cache (no-clone) 38 ns 218 ns With TTL, stats, events enabled
ts-cache (default) 630 ns 532 ns Full features + cloning for safety
node-cache 549 ns 692 ns Popular sync cache library

Redis Performance

Client GET SET Batch GET (100)
ts-cache 19.7 µs 19.7 µs 61 µs 🥇
Bun native 18.4 µs 19.0 µs 80 µs
ioredis 21.3 µs 19.7 µs 91 µs
node-redis 21.7 µs 22.0 µs 118 µs

ts-cache Redis driver wins on batch operations!

Ultra-Fast Mode

Enable maximum performance by disabling optional features:

import { Cache } from 'ts-cache'

const cache = new Cache({
  useClones: false, // No cloning (store references)
  enableStats: false, // No statistics
  enableEvents: false, // No event emission
  stdTTL: 0, // No TTL checking
  checkPeriod: 0, // No expiration checks
  maxPerformance: true, // Use Map storage
})

// Now 3.2x faster than lru-cache!
cache.set('key', value) // 30ns vs lru-cache 40ns
cache.get('key') // 4ns vs lru-cache 12ns

Performance Modes

Choose the right mode for your needs:

Mode GET SET Features Use Case
Ultra-Fast 4 ns 30 ns None Hot paths, maximum speed
No-Clone 38 ns 218 ns TTL, stats, events Controlled data, good speed
Default 630 ns 532 ns All + cloning Full features, data safety

Running Benchmarks

# Memory cache benchmarks
bun benchmarks/memory.ts

# Redis benchmarks (requires Redis running)
docker run -d -p 6379:6379 redis:alpine
bun benchmarks/redis.ts

What Makes Ultra-Fast Mode So Fast?

Optimizations:

  1. No EventEmitter (~25ns saved) - Biggest win
  2. No stats tracking (~15ns saved)
  3. No TTL checking (~20ns saved)
  4. Map storage (~10ns saved) - Guaranteed O(1)
  5. Cached flags (~5ns saved)

Total: ~75ns savings per operation

When to Use Each Mode

✅ Ultra-Fast Mode

  • Hot paths (called millions of times/sec)
  • No TTL needed (data doesn't expire)
  • Don't need statistics or monitoring
  • Control the data (safe to store references)

✅ No-Clone Mode

  • Good performance + features needed
  • TTL expiration required
  • Need statistics/monitoring
  • Control the data (won't mutate)

✅ Default Mode

  • Data safety is critical
  • Working with untrusted data
  • Multiple parts of code access cache
  • Need full feature set

Key Findings

  1. ts-cache ultra-fast mode beats lru-cache by 3.2x on GET operations
  2. Cloning is expensive: 179x slower for large objects
  3. EventEmitter overhead is significant: ~25ns per operation even with zero listeners
  4. ts-cache Redis driver wins on batch operations: 24% faster than Bun native
  5. Features are now optional: Progressive performance based on needs

Real-World Impact

For 1 million operations per second:

  • lru-cache: 12.37ms total
  • ts-cache (ultra-fast): 3.90ms total
  • Savings: 8.47ms per million ops

For a service handling 10K requests/sec with 10 cache ops each:

  • 100K cache ops/second
  • ~850µs saved per second
  • Equivalent to 8.5 more requests/sec with same CPU

Detailed Results

See ULTRA-FAST-MODE.md for complete analysis and configuration guide.