Re: strlcpy and how CPUs can defy common sense

ram@zedat.fu-berlin.de (Stefan Ram) wrote or quoted:

This video might be a hit with some folks: "The strange details
of std__string at Facebook" - Nicholas Ormrod, CppCon 2016.

  Also, "Efficiency with Algorithms, Performance with Data Structures"
  - Chandler Carruth, CppCon 2014, from which I take:

CPUS HAVE A HIERARCHICAL CACHE SYSTEM

One cycle on a 3 GHz processor            1   ns
L1 cache reference                        0.5 ns
Branch mispredict                         5   ns
L2 cache reference                        7   ns          14x L1 cache
Mutex lock/unlock                        25   ns
Main memory reference                   100   ns          20xL2, 200xL1
Compress 1K bytes with Snappy         3,000   ns
Send 1K bytes over 1 Gbps network    10,000   ns  0.01 ms
Read 4K randomly from SSD           150,000   ns  0.15 ms
Read 1 MB sequentially from memory  250,000   ns  0.25 ms
Round trip within same datacenter   500,000   ns   0.5 ms
Read 1 MB sequentially From SSD   1,000,000   ns   1   ms 4x memory
Disk seek                        10,000,000   ns  10   ms 20xdatacen. RT
Read 1 MB sequentially from disk 20,000,000   ns  20   ms 80xmem.,20xSSD
Send packet CA->Netherlands->CA 150,000,000   ns 150   ms

 .