Performance

PerformanceΒΆ

TODO: performance model

memory, time, allocations

FLOPS, computing FLOPS, how many FLOPS can gou get?

system scheduling, measuring over multiple runs

cache, blocking

parallel model