Five credible estimates of brain computing capacity in terms of FLOPS by others that we are aware of are spread across the range from 3 x 1013 to 1025 FLOPS. The median estimate is 1018 FLOPS.
Our own estimate of brain communication capacity in terms of TEPS suggests a figure of roughly 1-30 * 1016 FLOPS, with high uncertainty.
We have not investigated the brain’s performance in FLOPS in detail. This page summarizes others’ estimates that we are aware of, as well as the implications of our investigation into brain performance in TEPS. Text on this page was heavily borrowed from a blog post, Preliminary prices for human-level hardware.
Sandberg and Bostrom 2008
Sandberg and Bostrom project the processing required to emulate a human brain at different levels of detail.1 For the three levels that their workshop participants considered most plausible, their estimates are 1018, 1022, and 1025 FLOPS. These would cost around $100K/hour, $1bn/hour and $1T/hour in 2015.
Moravec (2009) estimates that the brain performs around 100 million MIPS.2 MIPS are not directly comparable to MFLOPS (millions of FLOPS), and have deficiencies as a measure, but the empirical relationship in computers is something like MFLOPS = 2.3 x MIPS0.89, according to Sandberg and Bostrom.3 This suggests Moravec’s estimate coincides with around 3.0 x 1013 FLOPS. Given that an order of magnitude increase in computing power per dollar corresponds to about four years, knowing that MFLOPS and MIPS are roughly comparable is plenty of precision.
Conversion from brain performance in TEPS
Among a small number of computers we compared5, FLOPS and TEPS seem to vary proportionally, at a rate of around 1.7 GTEPS/TFLOP. We also estimate that the human brain performs around 0.18 – 6.4 * 1014 TEPS. So this gives us 0.9 – 33.7 * 1016 FLOPS.6
- From Sandberg and Bostrom, table 9: Processing demands (emulation only, human brain)(p80):
- spiking neural network: 1018 FLOPS (Earliest year, $1 million: commodity computer estimate: 2042, supercomputer estimate: 2019)
- electrophysiology: 1022 FLOPS (Earliest year, $1 million: commodity computer estimate: 2068, supercomputer estimate: 2033)
- metabolome: 1025 FLOPS (Earliest year, $1 million: commodity computer estimate: 2087, supercomputer estimate: 2044)
- ” it would take, in round numbers, 100 million MIPS (100 trillion instructions per second) to emulate the 1,500-gram human brain.” – Moravec, 2009
- See p89. It actually says FLOPS not MFLOPS, but this appears to be an error, given the graph.
- ‘If we use the figure of 1016 cps that I believe will be sufficient for functional emulation of human intelligence…’ – Kurzweil, The Singularity is Near, p121
- “The [eight] supercomputers measured here consistently achieve around 1-2 GTEPS per scaled TFLOPS (see Figure 3). The median ratio is 1.9 GTEPS/TFLOPS, the mean is 1.7 GTEPS/TFLOP, and the variance 0.14 GTEPS/TFLOP. ” See Relationship between FLOPS and TEPS here for more details
- 0.18 – 6.4 * 1014 TEPS =0.18 – 6.4 * 105 GTEPS =0.18 – 6.4 * 105 GTEPS * 1TFLOPS/1.9GTEPS = 9,000-337,000 TFLOPS = 0.9 – 33.7 * 1016FLOPS