The numbers are so vast they need analogies to land.
1,000×
RAM vs. cache
A main memory access (100ns) vs. L1 cache hit (0.5ns) is like waiting 3 minutes for coffee vs. 8 hours at the DMV.
100,000×
SSD vs. cache
An NVMe SSD read (~10μs) vs. an L1 cache hit (0.5ns). If the cache hit is one second, the SSD read is a full day.
10,000,000×
HDD seek vs. cache
A spinning disk seek (~5ms) vs. an L1 cache hit. If the cache is one second, the HDD seek is four months.
~1,000×
Cross-country vs. same-rack
A cross-country round trip (~40ms) vs. same-rack network hop (~40μs). Local is a thousand times faster.
31.5M
seconds in a year
π × 10⁷ is the classic engineering approximation. Close enough for napkin math, elegant enough to remember.
10¹⁷
ns in 10 years
The full span of this chart. A single nanosecond to a decade is 100 quadrillion to one. Your CPU does billions of operations before you finish reading this sentence.