Skip to content

Instantly share code, notes, and snippets.

Embed
What would you like to do?
Latency Numbers Every Programmer Should Know - MarkDown Fork

Latency Comparison Numbers

Note: "Forked" from Latency Numbers Every Programmer Should Know

Event Nanoseconds Microseconds Milliseconds Comparison
L1 cache reference 0.5 - - -
Branch mispredict 5.0 - - -
L2 cache reference 7.0 - - 14x L1 cache
Mutex lock/unlock 25.0 - - -
Main memory reference 100.0 - - 20x L2 cache, 200x L1 cache
Compress 1K bytes with Zippy 3,000.0 3 - -
Send 1K bytes over 1 Gbps network 10,000.0 10 - -
Read 4K randomly from SSD 150,000.0 150 - ~1GB/sec SSD
Read 1 MB sequentially from memory 250,000.0 250 - -
Round trip within same datacenter 500,000.0 500 - -
Read 1 MB sequentially from SSD 1,000,000.0 1,000 1 ~1GB/sec SSD, 4X memory
Disk seek 10,000,000.0 10,000 10 20x datacenter roundtrip
Read 1 MB sequentially from disk 20,000,000.0 20,000 20 80x memory, 20X SSD
Send packet CA → Netherlands → CA 150,000,000.0 150,000 150 -

Unit Reference

  • 1 ns = 10^-9 seconds
  • 1 µs = 10^-6 seconds = 1,000 ns
  • 1 ms = 10^-3 seconds = 1,000 µs = 1,000,000 ns

Credit/Notes

Visual Comparison Charts

Visual Comparison Chart

Scale of Computing Latencies

@jgoldschrafe

This comment has been minimized.

Copy link

@jgoldschrafe jgoldschrafe commented Mar 13, 2017

The numbers for main memory reference don't add up. They are 100x L1, but only 14.28x the L2 numbers given, not the 20x stated.

@AnatoliiStepaniuk

This comment has been minimized.

Copy link

@AnatoliiStepaniuk AnatoliiStepaniuk commented Aug 6, 2017

How long will it take to read 1kb from SSD? (Definitely not 1/1024 of time you need to read 1Mb, right?)

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment