Exit the dragon: US govt blows $325m on China-beating 300PFLOPS monster computer The US government has chosen IBM and Nvidia chips to build the world's fastest supercomputer – a 300 petaFLOPS beast that would trounce today's most powerful super: China's Tianhe-2. The Department of Energy has commissioned two supercomputers, it was revealed on Friday: one is a system codenamed "Summit", which will be …
Strategic Computing: DARPA and the Quest for Machine Intelligence, 1983-1993
Actually FORTRAN is the very opposite of your proverbial dead parrot :
http://arstechnica.com/science/2014/05/scientific-computings-future-can-any-coding-language-top-a-1950s-behemoth/
Some examples are atmospheric modeling and weather prediction carried out by the National Center for Atmospheric Research; classified nuclear weapons and laser fusion codes at Los Alamos and Lawrence Livermore National Labs; NASA models of global climate change; and an international consortium of Quantum Chromodynamics researchers, calculating the behavior of quarks, the constituents of protons and neutrons. These projects are just a few random examples from a large computational universe, but all use some version of Fortran as the main language.
"The US's fastest publicly known supercomputer is the Cray-built 27 petaFLOPS Titan at Oak Ridge, which is number two in the world rankings. Number three is the 20 teraFLOPS Sequoia at the Lawrence Livermore."
As the link makes clear, Sequoia turns in just over 20 petaFLOPS, not 20 teraFLOPS; measured in teraFLOPS, its score is just over twenty-thousand ("20,132.7 TFlop/s"), not twenty.
The article also employs curious rounding when ascribing only 54 petaFLOPS to China's Tianhe-2, instead of 55:
"It is designed to peak at 150 to 300 petaFLOPS – that's 300 quadrillion calculations per second, or about five times faster than the 54 PFLOPS Tianhe-2."
The Top 500 site quite clearly reports Tianhe-2's theoretical peak performance as just over 54.9 petaFLOPS ("54,902.4 TFlop/s"), which almost any of us would round up to 55.
On the other hand, this is the first instance I can think of in which someone who wrote something like "about five times faster than" may actually have meant it, for 300 is closer to 6 times as fast as 54 than it is to 5 times as fast (= 4 times faster), which is what people who say "five times faster than" usually mean.
Of course, if we use the more accurate 55 instead of 54, then 300 is slightly closer to 5 times as fast, or 4 times faster, than it is to 6 times as fast, or 5 times faster. The problem with the "__ times faster" usage is that the reader or listener can never tell for sure whether the writer or speaker truly meant it or not. More than 9 times out of 10, the writer or speaker actually only means N times as fast, which is actually N-1 times faster.
(One time faster = 100% faster = twice as fast = two times as fast. Etc.)
"for 300 is closer to 6 times as fast as 54 than it is to 5 times as fast (= 4 times faster)"
Especially if one mentally treats 54 as roughly 50, and does the entire mental calculation using rounded numbers, since 300 is exactly 6 times as much as 50, and exactly 5 times more than 50.
"US govt blows $325m on China-beating 300PFLOPS monster computer"
Blows? Why "blows"?
When we say that someone "blew" $X on something, we assert that the expenditure was a complete waste of money, yet there is no discussion in the article at all that suggests that the US government overpaid for the computers, or that expenditures on supercomputers are intrinsically a waste of taxpayer money.
So where is the evidence of wastefulness to back up the claim made in the headline?
Cf. "US blows $174m on new Cray to simulate nukes":
http://www.theregister.co.uk/2014/07/13/cray_super_to_help_us_simulate_nuke_tests/
Evidently El Reg has been proclaiming such expenditures intrinsically wasteful for a while. Has the editorial staff finally seen the light of libertarianism? Or at least of public choice theory (a branch of economics that explains, among other things, why bureaucracies and legislatures always tend to overspend)? Is it ready at last to call the scientists at Oak Ridge and Los Alamos "welfare queens in lab coats"?
Well, either that or else El Reg just mis-uses "blows" as a synonym for "spends" or "pays" (or "lays out"). A quick search of El Reg for articles with "blows" in the title suggests this is actually the case:
.
Restless PC biz Lenovo blows $100m to gobble 2,500 mobe tech patents
http://www.theregister.co.uk/2014/03/21/lenovo_lays_out_100_meeellion_in_patent_buyup/
.
Crown Castle blows $4.85 BEEELLION for rights to AT&T's cell towers
http://www.theregister.co.uk/2013/10/21/att_to_spend_24bn_leasing_its_own_infrastructure/
.
Chinese search giant Baidu blows $1.9 BEELION on app store
http://www.theregister.co.uk/2013/07/17/baidu_acquires_mobile_app_91_wireless/
.
Big Blue blows big green in SoftLayer public cloud gobble
http://www.theregister.co.uk/2013/06/04/ibm_acquires_softlayer_cloud/
.
.
On the other hand, here is an example — the only one I could find* — where it appears that "blows" was actually called for, and the content of the article fully supports its use:
.
Carmack blows 'crazy money' on hibernating Armadillo
http://www.theregister.co.uk/2013/08/02/armadillo_aerospace_hibernation/
.
Carmack himself calls the expenditure he made "crazy" and acknowledges that it was wasted. Which is why he cut it off.
.
.
* Excepting cases in which "blows" was used in some different sense unrelated to the expenditure of money, as where someone blew a deadline, or something was blown across a border.
And my apologies for the ugly, space-wasting kludge I had to use to make this last post readable. An inevitable consequence of El Reg's present policy of inserting an extra line break where none is wanted, yet removing extra lines where the composer has inserted them on purpose.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
