Visuall matches or beats C++ on 6 of 10 macro benchmarks. Compute-bound integer and loop work is within 1.0–1.4× of C++. Recursion-heavy workloads (Ackermann, TreeSum) are 4–9× due to GC stack scanning overhead.

Macro Benchmarks

Test C++ (ms) Visuall (ms) Python (ms) V / C++ Py / C++
Primes (trial div, 100K ×3) 23.2 31.7 415 1.4× 18×
TreeSum (recursive, depth 22) 15.2 63.8 1,298 4.2× 85×
Collatz (1..100K) 30.6 23.1 1,758 0.8× 57×
Strings (200K f-strings) 27.9 47.9 162 1.7×
Pi (Leibniz, 10M terms) 23.1 23.7 2,884 1.0× 125×
Nested loops (2000×2000) 11.0 10.7 705 1.0× 64×
Ackermann (3,11) 178.8 1,589.3 28,098 8.9× 157×
GCD sum (1..2000) 56.2 61.9 724 1.1× 13×
Fibonacci (fib(35) ×500K) 9.2 9.7 684 1.0× 74×
Float distance (1M) 10.4 10.1 403 1.0× 39×

Micro Benchmarks

Test Iterations C++ Visuall vs C++ Python
Match dispatch 1M 11.9 ms 9.2 ms 0.8× 222 ms
Function call 1M 8.5 ms 12.6 ms 1.5× 251 ms
Tight loop sum 100K 7.1 ms 7.7 ms 1.1× 47 ms

All tests run against equivalent C++ compiled with g++ -O2 and Python 3.14. Best of 3 runs reported. Test machine: not documented; these are README benchmarks. Benchmarks updated for Visuall v1.1.0.