Skip to main content

Databend Vectorized Engine Performance

tip
  • Memory SIMD-Vector processing performance only
  • Dataset: 100,000,000,000 (100 Billion)
  • Hardware: AMD Ryzen 9 5950X 16-Core Processor, 32 CPUs
  • Rust: rustc 1.61.0-nightly (8769f4ef2 2022-03-02)
QueryDatabendQuery (v0.6.87-nightly)
SELECT avg(number) FROM numbers_mt(100000000000)1.682 s.
(59.47 billion rows/s., 475.76 GB/s.)
SELECT sum(number) FROM numbers_mt(100000000000)1.621 s.
(61.67 billion rows/s., 493.37 GB/s.)
SELECT min(number) FROM numbers_mt(100000000000)3.962 s.
(25.24 billion rows/s., 201.93 GB/s.)
SELECT max(number) FROM numbers_mt(100000000000)2.792 s.
(35.82 billion rows/s., 286.54 GB/s.)
SELECT count(number) FROM numbers_mt(100000000000)1.172 s.
(85.31 billion rows/s., 682.46 GB/s.)
SELECT sum(number+number+number) FROM numbers_mt(100000000000)6.032 s.
(16.58 billion rows/s., 132.63 GB/s.)
SELECT sum(number) / count(number) FROM numbers_mt(100000000000)1.652 s.
(60.52 billion rows/s., 484.16 GB/s.)
SELECT sum(number) / count(number), max(number), min(number) FROM numbers_mt(100000000000)6.212 s.
(16.10 billion rows/s., 128.78 GB/s.)
SELECT number FROM numbers_mt(10000000000) ORDER BY number DESC LIMIT 101.414 s.
(8.76 billion rows/s., 70.09 GB/s.)
SELECT max(number), sum(number) FROM numbers_mt(10000000000) GROUP BY number % 3, number % 4, number % 5 LIMIT 105.791 s.
(1.73 billion rows/s., 13.81 GB/s.)

Experience 100 billion performance on your laptop, talk is cheap just bench it