{"id":50624,"date":"2026-01-05T17:18:07","date_gmt":"2026-01-05T22:18:07","guid":{"rendered":"https:\/\/mjtsai.com\/blog\/?p=50624"},"modified":"2026-01-07T15:10:57","modified_gmt":"2026-01-07T20:10:57","slug":"latency-numbers-every-programmer-should-know","status":"publish","type":"post","link":"https:\/\/mjtsai.com\/blog\/2026\/01\/05\/latency-numbers-every-programmer-should-know\/","title":{"rendered":"Latency Numbers Every Programmer Should Know"},"content":{"rendered":"

Jonas Bonér<\/a> (based on work by Peter Norvig and Jeff Dean from 2012):<\/p>\n

\n
\nL1 cache reference                           0.5 ns\nBranch mispredict                            5   ns\nL2 cache reference                           7   ns                      14x L1 cache\nMutex lock\/unlock                           25   ns\nMain memory reference                      100   ns                      20x L2 cache, 200x L1 cache\nCompress 1K bytes with Zippy             3,000   ns        3 us\nSend 1K bytes over 1 Gbps network       10,000   ns       10 us\nRead 4K randomly from SSD*             150,000   ns      150 us          ~1GB\/sec SSD\nRead 1 MB sequentially from memory     250,000   ns      250 us\nRound trip within same datacenter      500,000   ns      500 us\nRead 1 MB sequentially from SSD*     1,000,000   ns    1,000 us    1 ms  ~1GB\/sec SSD, 4X memory\nDisk seek                           10,000,000   ns   10,000 us   10 ms  20x datacenter roundtrip\nRead 1 MB sequentially from disk    20,000,000   ns   20,000 us   20 ms  80x memory, 20X SSD\nSend packet CA->Netherlands->CA    150,000,000   ns  150,000 us  150 ms<\/pre>\n<\/blockquote>\n\n
Colin Scott<\/a> has a page that helps visualize how these types of numbers have changed over time (Hacker News<\/a>).<\/p>\n\n
Jon Snader<\/a>:<\/p>\n
\n
Mohammad Zeya Ahmad has an informative post<\/a> [archive<\/a>] that answers that question. He has a list of how much time various common operations take. That’s interesting but what make his list stand out is that he draws conclusions from his results.\n<\/p>\n
\nFor example, SSDs are about 30 times faster than HDDs so if you have a high performance disk-based task, it makes sense to use SSDs. Of course, there are reasons to prefer HDDs but if performance is your controlling metric, SSDs are probably your best choice.\n<\/p>\n
\nFor each group of comparable metrics, Ahmad offers an actionable suggestion. Those groups range from CPU versus Cache and Memory speeds to network transfer times.<\/p>\n<\/blockquote>\n\n
Previously:<\/p>\n