Topics
Performance optimization techniques, by topic
Every technique in adtomic was mined from a real commit or paper and comes with its measured speedup. Browse the main families; ask Claude for the answers.
Better algorithms
6,505 techniques
Avoiding allocations
4,644 techniques
Precomputation and memoization
4,523 techniques
SIMD vectorization
4,448 techniques
Early exit and pruning
4,376 techniques
Bit tricks and strength reduction
1,973 techniques
Batching
1,931 techniques
Specialization, templates and JIT
1,716 techniques
Reducing synchronization
1,462 techniques
Pipelining and async overlap
1,374 techniques
Lazy evaluation
1,354 techniques
Bit packing and smaller types
1,306 techniques
Zero-copy
1,179 techniques
Inlining and devirtualization
951 techniques
Lookup tables
909 techniques
Cache locality and data reordering
897 techniques
Branchless programming
885 techniques
Multithreading and work stealing
795 techniques
GPU kernel fusion
766 techniques
Arena and pool allocation
715 techniques
Loop unrolling and instruction-level parallelism
705 techniques
I/O batching
604 techniques
Fast math and approximation
511 techniques
Ask Claude for the full answer
Each technique comes with its bottleneck, measured speedup, trade-offs, conditions and the commit or paper behind it, served to Claude through MCP.
claude mcp add --transport http adtomic https://www.adtomic.co/mcp