I know I’m going to get Crazy nerd points for this, but hey, that’s a good thing.
I have calculated Data’s Read Speed.
According to the Wiki, which I helped update on precisely the quoted material, “800 quadrillion bits” or 100 petabytes” is Data’s storage capacity.
What’s interesting about this figure is that it is well beyond current super computer norms, now that’s not shocking given that this is a show about hundreds of years in the future. However, Data’s processing speed is actually well below modern super computers. “60 trillion operations per second” or 60 teraflops.” Data’s speed compared to 1456.7 teraflops for the Roadrunner at Los Alamos.
I wondered about how Data could host such a profoundly advanced AI on such limited processor speed.
Now every nerd knows that a computer’s output is a function of many things not just clock speed, but also architecture and software.
But given how little we are told about this in the show these factors can be regarded as magic, what we can know is as before his capacity, and his clock speed, which are directly stated but his read speed is not.
Something else we know is that Data almost certainly has a single type of solid state memory, as demonstrated by his ability to be instantly shut down and reactivated with no memory loss.
Read time is how long it takes to read a volume of data.
This was hard to nail down because typically Data knows something about which he is asked, so it could be constructed when a result is located he stops searching, like your keys are always in the last place you look because once you find them you stop looking.
But in episode 28 “Where Silence Has Lease” we have the following useful clue.
In studying a stellar phenomenon the following conversation takes place.
“Data is there any record anywhere of any occurrence even vaguely similar to this.” – Riker
“Accessing…” – Data
Elapsed time 3 seconds.
“Negative sir.” – Data
I think that search input is qualitatively close enough to “search your entire memory” for an estimate.
And we know that no result was found so he did search it from beginning to end.
We also know he did not search the main computer, it’s seek time is much longer. Probably due to database size. But that’s another nerd rant π
Thus…
800 quadrillion bits searched in 3 seconds in megabytes
WolframAlpha returns Data’s final read speed on or about 3.333×10^10 MB or 33.33 PB, per second.
Data’s processor speed may be a little slow, but his capacity and this monstrous seek time launch him squarely into future land for the time being.
I predict his seek time will be the last thing we actually match about his hardware performance statistics.
I suspect we’ll achieve AI first.
Links.
http://www04.wolframalpha.com/input/?i=800+quadrillion+bits+divided+by+3+in+megabytes
http://en.wikipedia.org/wiki/Data_star_trek
http://en.wikipedia.org/wiki/TOP500#List_as_of_November_2008
Probably due to all the computational elements being run in parallel. Billions of them.
I did some calculations and an array of 128,000 CPUs not much faster than a Pi Zero (600MHz) drawing less than 30W would be able to search a data volume this large in around 3 seconds but only if the information was distributed evenly among the CPUs.
I wonder if holographic storage would accomplish that.
https://en.wikipedia.org/wiki/Holographic_data_storage
Thanks for the comment, I really appreciate it π
Sidenote: Vote in the primary! #BernieOrBust
As well they are not telling us anything about how complex datas “instructions” are…
a single operation on his positronic matrix could be the equivelent of hundreds or even thousands of binary operations.
That’s true. Apparently instructions in computer science have no minimum simplicity. I did not know this until researching for this reply hehe.
https://simple.wikipedia.org/wiki/Instruction_(computer_science)
I guess that’s one way they could retcon the gap closed.
Thanks for the comment π