3 ms·
It's a real shame they don't talk more specifically about Luhn's role in the 'record addressing' problem. Back in 1953, IBM was investigating disk storage for
by mschaef 9y ago
It's a real shame they don't talk more specifically about Luhn's role in the 'record addressing' problem.
Back in 1953, IBM was investigating disk storage for record retrieval in large volumes of data. Seek/read time on these machines was on the order of a second, so a 10 level binary search could take around 10 seconds of total time. Unfortunately for IBM, this wasn't an improvement over the best human-driven manual lookup processes... which is an issue if you're trying to sell/lease expensive computer hardware.
With this as motivation, Luhn then essentially invented what we now would know as a hash table, which was key to the 1956 RAMAC release.
- bogomipz 9y ago>"Seek/read time on these machines was on the order of a second, so a 10 level binary search could take around 10 seconds of total time. Unfortunately for IBM, this wasn't an improvement over the best human-driven manual lookup processes" I am curious what human-driven manual lookup processes existed that could be in a single second?
- brian_cloutier 9y agoI have no information as to what the parent is referring to but 2^10 is 1024. That seems roughly in line with what a double-page spread of a phone book could show you; I don't think it's unreasonable that a trained human could search through one of those in under 10 seconds.
- mschaef 9y agoA tub file: https://en.wikipedia.org/wiki/Tub_file https://en.wikipedia.org/wiki/Tub_file (Note that the overall search takes 10 seconds, not the one second of an individual seek... so the human had a bit more time to be competitive with the disk than a single second. :-) )