First Browser-Based Quantum Computer Simulator Released

← Back to Stories (view on slashdot.org)

First Browser-Based Quantum Computer Simulator Released

Posted by samzenpus on Wednesday May 21, 2014 @02:11PM from the give-it-a-try dept.

greg65535 (1209048) writes "Following the trend of on-line coding playgrounds like JSFiddle or CodePen, Google researchers unveiled the first browser-based, GPU-powered Quantum Computing Playground. With a typical GPU card you can simulate up to 22 qubits, write, debug, and share your programs, visualize the quantum state in 2D and 3D, see quantum factorization and quantum search in action, and even... execute your code backwards."

61 comments

Min score:

Reason:

Sort:

Well, not exactly. by Anonymous Coward · 2014-05-21 14:16 · Score: 5, Funny

They both released it and didn't release it simultaneously.
1. Re:Well, not exactly. by Travis+Mansbridge · 2014-05-21 14:21 · Score: 5, Funny
  
  Until we observed this article.
2. Re:Well, not exactly. by canadiannomad · 2014-05-21 14:30 · Score: 2, Funny
  
  At which point it collapsed into a state where it has been released, but fails to be able to do anything useful.
  
  --
  Hmm, the humour and sarcasm seem to have been be lost on you.
3. Re:Well, not exactly. by Anonymous Coward · 2014-05-21 14:58 · Score: 1
  
  So, like most Google Betas then.
4. Re:Well, not exactly. by dudpixel · 2014-05-21 15:21 · Score: 1
  
  It fails to do anything useful backwards as well :)
  
  --
  This seemed like a reasonable sig at the time.
5. Re:Well, not exactly. by wonkey_monkey · 2014-05-21 19:23 · Score: 1
  
  This joke is both funny and not funny now.
  Oh, wait... its waveform just collapsed. Guess what to?
  
  --
  systemd is Roko's Basilisk.
6. Re:Well, not exactly. by Anonymous Coward · 2014-05-21 21:53 · Score: 0
  
  Until we.. observed... this article?
  Come on now, you know that will never happen, this is Slashdot.
  Well this experiment sure failed. It will never come to an end.
7. Re:Well, not exactly. by bluefoxlucid · 2014-05-22 02:36 · Score: 1
  
  Actually, quantum theory says nearly that.
  In Chaos theory, things appear random because they are deterministic but you don't have perfect information to calculate the result. Your lack of information introduces randomness. Dice, for example, fall based on their mass, their momentum, air density, the shape and material properties of the surface, etc. These things are themselves imparted by how they're thrown, by the temperature and humidity and make-up of the air, and so on. If you could know all of these things, you could predict exactly how the dice will land: the physical interaction between the dice, its thrower, the air, gravity, and the surface determine how the dice tumble and where they stop.
  In quantum theory, things are actually random. A particle may or may not decay; an electron may or may not drop to a lower energy state and emit a photon. These quantum events represent the infinite variation in precise timing of the firing of neurons, impacting human decision and propagating to other decisions or physical outcomes (i.e. the split-second reaction to having a firearm pointed at you: one tenth of a second can decide whether you get shot or tackle your assailant unharmed).
  At the edge of your field, you reach events which can occur from any number of inputs. Too many states could cause what you're seeing, and the likely events become unrestricted. Thus, to an extent, parts of the world you have such insignificant knowledge of are unknown, and do not exist. Whether a Chinese factory worker knocks a fountain pen nib off the line or not makes approximately zero difference to anything; it has and hasn't happened until the news is brought to you that it has. Quantum theory states that all such outcomes occur simultaneously, but that the observer is restricted to one set of outcomes: if the outcome is unobserved, it hasn't physically happened yet.
  That means your world is unfixed until you observe it. Your friends who know of these things but have said nothing know of both states, and the state that has happened doesn't occur until they communicate one or the other to you--the only state that person has ever known. It's a tough concept, but think of it this way: there isn't just more than one world, but rather more than one outcome. The world is cobbled together from pieces.
  
  --
  Support my political activism on Patreon.
When you execute it backwards by NoNonAlphaCharsHere · 2014-05-21 14:52 · Score: 1

It prints "I buried Paul".
Maybe it killed your cat by Anonymous Coward · 2014-05-21 15:05 · Score: 0

Better go check...
If each of those is 22 qubits... by thieh · 2014-05-21 15:22 · Score: 0

Would a simple botnet be able to easily crack all encryption crackable by quantum computing, or are there better ways to go at it given a botnet?
1. Re:If each of those is 22 qubits... by Anonymous Coward · 2014-05-21 15:28 · Score: 0
  
  Simulates is the operative word you missed.
2. Re:If each of those is 22 qubits... by Anonymous Coward · 2014-05-21 15:53 · Score: 1
  
  They must mean simulates as in sim-city, not simulates as in actual simulation.
3. Re:If each of those is 22 qubits... by stoploss · 2014-05-21 16:22 · Score: 4, Informative
  
  Would a simple botnet be able to easily crack all encryption crackable by quantum computing, or are there better ways to go at it given a botnet?
  Yes it is crackable using a bother simulating a quantum computer, in the same sense that you would be able to simulate a quantum computer solving the traveling salesman problem by using a botnet. Or by using a massively parallel supercomputer.
  That is to say, the quantum computer simulation is Turing computable. This really doesn't help for anything more than trivial problems, much like pointing out the Halting Problem is decidable if you "simply" observe the Turing machine for the appropriate Busy Beaver function's number of execution steps.
  More succinctly, the simulation would gain you nothing over a direct parallel processing attack on the key space, and in fact the quantum computer simulation would add execution overhead that would reduce efficiency compared to straightforward brute force attacks.
4. Re:If each of those is 22 qubits... by retchdog · 2014-05-21 17:41 · Score: 0
  
  simulation does not mean it is in real-time, you ignorant slut.
  
  --
  "They were pure niggers." – Noam Chomsky
5. Re:If each of those is 22 qubits... by Anonymous Coward · 2014-05-21 17:54 · Score: 0
  
  Yes it is crackable using a bother simulating a quantum computer...
  I'm not sure whether to be bothered by brilliance, or bothered by bullshit.
  In either case, nice work!
6. Re:If each of those is 22 qubits... by mwvdlee · 2014-05-21 22:02 · Score: 1
  
  Simulation as in "simulation", not simulation as in "emulation".
  
  --
  Slashdot social media options: AIM, ICQ, Yahoo, Jabber and Mobile Text. Why no MySpace?
7. Re:If each of those is 22 qubits... by ld+a,b · 2014-05-22 04:58 · Score: 1
  
  You mean a simulation like this?
  # cat /dev/random | grep "the answer to life, the universe and everything" | sed -e 's/the answer to life, the universe and everything/42/'
  
  --
  10 little-endian boys went out to dine, a big-endian carp ate one, and then there were -246.
8. Re:If each of those is 22 qubits... by n-russo · 2014-05-22 06:39 · Score: 1
  
  It didn't output 42 for me:
  # cat /dev/random | grep "the answer to life, the universe and everything" | sed -e 's/the answer to life, the universe and everything/42/'
  Binary file (standard input) matches
9. Re:If each of those is 22 qubits... by stoploss · 2014-05-22 06:59 · Score: 1
  
  Yes, I saw that right after I posted. Autocorrect stymied me as I spelled "botnet". Fixed it once, but then it "helped me out" again when I edited the sentence later.
even... execute your code backwards. by quax · 2014-05-21 15:56 · Score: 4, Insightful

This is actually a requirement for such a simulator as all unitary QM transformations are reversible.
It's kind of ironic that Google released this project given that they are at the same time heavily betting on D-Wave with a radically different approach to QM than the Gate based model.
The D-Wave founder Geordie Rose is know for disparaging the Quantum Gate based model as completely impractical, and in turn other QC researchers have been very critical of his approach to the matter. Spawning a contentious controversy almost as old as the Canadian start-up itself.
1. Re:even... execute your code backwards. by by+(1706743) · 2014-05-21 16:34 · Score: 2
  
  It might be interesting if they introduced some user-selectable amounts of simulated decoherance, though -- perhaps to allow for simulation of quantum error correction, etc. Looking at this locally, it could be non-unitary (though I'm not sure the extent of the environment that one would model for such a computer simulator). Fun stuff, in any event.
2. Re:even... execute your code backwards. by quax · 2014-05-21 16:58 · Score: 1
  
  A simple way to simulate this to some extend would be to just add some random noise in form of qubit flips.
  But with just 20 qubits you unfortunately can't push this very far.
3. Re:even... execute your code backwards. by retchdog · 2014-05-21 17:47 · Score: 1
  
  no, it's not ironic. the simulator is just a fucking project they're hosting because it's 'cool'. they are not investing anything in it beyond a smidgen of bandwidth and disk space, and they are not endorsing it.
  they're also not "betting heavily" on D-wave. it was a stab-in-the-dark just-in-case thing which they could afford with the coins under Sergei's couch cushions, and despite that i wouldn't be surprised if they're still regretting how hopeless their investment turned out to be. D-wave is bullshit.
  *: unless D-wave secretly paid google a huge pile of cash for this implied endorsement. it's the only thing that would really make sense, but still seems really unlikely.
  
  --
  "They were pure niggers." – Noam Chomsky
4. Re:even... execute your code backwards. by Beck_Neard · 2014-05-21 17:53 · Score: 3, Interesting
  
  Well let's compare. Geordie Rose spent years and millions of dollars trying (and succeeding) in building a computational device that works on radically different principles than existing computer tech, is actually useful for a lot of real-world tasks, and consumes virtually zero power - a huge feat in itself, even if it's not really a "quantum computer" in the traditional sense of the word. Whereas those people disagreeing with him are all ivory tower academics who have not built and do not plan to build any hardware. The most egregious of which is Scott Aaronson who is known for his delusional rants on everything from neuroscience to fundamental physics. I wonder which one has their head grounded more firmly in reality.
  But seriously though, the fundamental principles of gate-based and adiabatic quantum computing aren't that different; it's more a continuum where on one end you have highly decoherent classical behavior, on the other you have pure quantum behavior, and in the middle you have quantum+noise behavior where tiny entanglements are being generated and decohered on a rapid scale that is too short to do quantum computing but long enough to do adiabatic quantum computing. It's possible that by investing in AQC technology, as the technology matures it will give better and better entanglement and eventually approach a pure quantum computer in capability.
  
  --
  A fool and his hard drive are soon parted.
5. Re:even... execute your code backwards. by Anonymous Coward · 2014-05-21 19:37 · Score: 0, Insightful
  
  How much does DWave pay you?
6. Re:even... execute your code backwards. by Beck_Neard · 2014-05-21 19:59 · Score: 1
  
  I'm sorry if my post sounded like a commercial, it's just that I've done a lot of research on D-wave's hardware and it's really impressive what such a small team managed to pull off. At least they're doing something.
  
  --
  A fool and his hard drive are soon parted.
7. Re:even... execute your code backwards. by StripedCow · 2014-05-21 20:44 · Score: 2
  
  All of physics is reversible (except perhaps black holes and such), so your regular computer should be reversible too.
  
  --
  If Pandora's box is destined to be opened, *I* want to be the one to open it.
8. Re:even... execute your code backwards. by Anonymous Coward · 2014-05-21 20:51 · Score: 0
  
  Controversy? More that it kind-of smells like snake oil, but everyone who has a financial stake in it really wants to persuade the world that it isn't :-)
9. Re:even... execute your code backwards. by Anonymous Coward · 2014-05-21 22:46 · Score: 0
  
  Sure, let's replace the scientific method with sniffing.
10. Re:even... execute your code backwards. by Anonymous Coward · 2014-05-22 00:34 · Score: 2, Informative
  
  ...so your regular computer should be reversible too.
  For a regular computer to be reversible it needs reversible logic gates. For example, a standard XOR gate loses one bit of information, so given the output you cannot construct the input perfectly (as there are two possible inputs for each output).
11. Re:even... execute your code backwards. by VortexCortex · 2014-05-22 00:54 · Score: 1
  
  When you reverse a black hole it's called a white hole, AKA, big bang.
12. Re:even... execute your code backwards. by Anonymous Coward · 2014-05-22 01:46 · Score: 0
  
  Time, thermodynamics, and enthropy seem exceptions to this everything is reversible rule.
13. Re:even... execute your code backwards. by bluefoxlucid · 2014-05-22 02:38 · Score: 1
  
  The big bang was a hell of a lot more complex than that. If you ignited a miniature big bang in our universe, it would likely destroy the universe. Yes, even a really tiny one.
  
  --
  Support my political activism on Patreon.
14. Re:even... execute your code backwards. by VortexCortex · 2014-05-22 02:55 · Score: 0
  
  ...so your regular computer should be reversible too.
  For a regular computer to be reversible it needs reversible logic gates. For example, a standard XOR gate loses one bit of information, so given the output you cannot construct the input perfectly (as there are two possible inputs for each output).
  But the output from the opcode isn't stored back to both input memory locations at once ergo, XOR itself is reversible at the chip level, even if it writes back to one of the inputs just XOR the output with the other input. You're conflating the theory of computation with the actual computation. In THEORY you can delete bits, but in practice you actually can't -- Well, using the arrow of time created by sub-atomic entropy (quantum foam) you might be able to... but that will remain beyond your grasp for some time yet. When you write zeros over the data the exact opposite process would restore the data because its remnants are still there encoded into everything from slight resistance in potential of the RAM or repulsion of the writehead, etc. you leave behind sub-bit signatures. Let's not even get into in-memory attempts to erase memory that can fail due to caching, paging, another thread with a copy, etc. and just talk about on-disk data.
  Let's say I have these bits: 1 0 1 0 and I write over them with 0 0 1 1. For the sake of argument let's say that each write is affected by one tenth of the origin data's signal. Our existing initial state may actually not be so clean, and our write signal may not be so perfect, but let's assume they are just for example. Here's the overwrite:1.00 <- 0 = 0.100 0.10 <- 0 = 0.010 1.01 <- 1 = 1.101 0.00 <- 1 = 1.000
  We're allowing bits to go above 1 because in reality there's a threshold for the bit value one, and you can exceed it (obviously). Really, the zeros should be negative ones, but this is just an oversimplified example. Let's say we wanted to reverse the process. We read back what is apparently stored there which is rounded to the whole bits (0 0 1 1) and subtract that out of the analog signal (decimals). That zeros the whole number threshold place, but it would reveal the tenths place I've emboldened above. You amplify that signal beyond the threshold and you've got our origin signal: 1 0 1 0. See, the theory of the computer would have said those bits are lost forever, but even without resorting to full reversal of everything at the quantum scale I can get your overwritten bits back in practice. With an even more sensitive system you could get what was written in a prior pass than this, revealing what's in the hundredths and thousandths place, etc., though each layer down is more entropic.
  This is just one reason why writing zeros all over the disk doesn't really erase your data, that's actually the worst thing to write. Your neighbor likely wouldn't be able to get the data back, but the drive itself may have just marked that sector entry in its look up table as full of zeros without actually changing the data on the disk -- read it back and the table could tell the controller to fill the buffer with zeros without touching the actual disk data (sort of how POSIX file systems are allowed to do with files full of zeros, and may stop your zero write at the FS level, thus we needed to go deeper).
  To erase data so that it's unreachable by police or thieves you'll have to write random noise all over the disk to erase it. However, state-level & enemy governments could remove the drive platters from their enclosures and place them in highly sensitive drive reading tech with heads that could pick up the analog signal and perform the top-layer subtraction method I mentioned above. So, to really erase the bits you want to write over the surface with multiple passes of random bits.
  Ah, but SSDs employ ware leveling and even magnetic spinning disks frequently swap out a sector from use. The logi
15. Re:even... execute your code backwards. by Anonymous Coward · 2014-05-22 03:49 · Score: 0
  
  Wrong. Second Law of Thermodynamics.
16. Re:even... execute your code backwards. by Anonymous Coward · 2014-05-22 05:13 · Score: 0
  
  Whereas those people disagreeing with him are all ivory tower academics who have not built and do not plan to build any hardware.
  Some of us academic actually do build hardware and use them for various tests and experiments, including coworkers I've had that spun off companies form the hardware they've made, while still disagreeing there is any value to D-wave's products or even orders of magnitude improvement upon it...
  So what hardware have you made, or are you trying to live up to the stereotype your username is based on?
17. Re:even... execute your code backwards. by flargleblarg · 2014-05-22 06:46 · Score: 1
  
  Seriously, this was one of the most interesting comments I've ever read here on Slashdot. Thanks for taking the time to write it.
18. Re:even... execute your code backwards. by quax · 2014-05-22 10:25 · Score: 1
  
  "D-wave is bullshit."
  Tell us how you really feel.
19. Re:even... execute your code backwards. by quax · 2014-05-22 10:39 · Score: 1
  
  Mighty big roar for an AC.
20. Re:even... execute your code backwards. by quax · 2014-05-22 10:47 · Score: 1
  
  While he makes some good points he is unfortunately completely missing the point with this statement:
  " In THEORY you can delete bits, but in practice you actually can't ."
21. Re:even... execute your code backwards. by quax · 2014-05-22 10:59 · Score: 2
  
  " In THEORY you can delete bits, but in practice you actually can't ."
  If I give you a bunch of RAM SIMs there's no way you can tell me what was written on them.
  At any rate, fully reversible computing means the ability to completely reverse arbitrarily complex algos, being able to reconstruct a couple of previous bit states isn't cutting it.
  And yes, you actually can delete bits, the entropy heat signature this produces is theoretically well understood, and Landauer's principle has recently been experimentally confirmed.
22. Re:even... execute your code backwards. by quax · 2014-05-22 11:10 · Score: 1
  
  Have been blogging about them for a while and visited them on site.
  Full Disclosure: One of their board members paid a beer for me.
  It's because of dudes like you that I am cross with Scott A. He has every right to be critical but his rhetoric is so over the top that he created a kind of parallel universe, that doesn't even allow for this kind of adiabatic quantum computation to be tried and tested.
23. Re:even... execute your code backwards. by Anonymous Coward · 2014-05-22 15:09 · Score: 0
  
  That is an even more empty statement than saying academics don't actually build anything or work toward accomplishing anything. Even if you can find an asshole riding his tenure and otherwise being an ivory tower idiot, there are a dozen more with lots of actual hands on experience and many of which that reach out to do or run industry work. Might as well argue anyone who posts on Slashdot has never done anything because they never left their basement.
24. Re:even... execute your code backwards. by quax · 2014-05-22 18:57 · Score: 1
  
  What's empty is your straw-man argument. Of course most academics do excellent work.
  What the original poster claimed was academics in the QC hardware business dismissing D-Wave. The most outspoken critic is a theorists. Is it too much to ask to get a link to a more hardware oriented academic going on the record with regards to D-Wave?
  MIT included them in the list of the fifty smartest companies, so we know there are plenty of academics who think highly of D-Wave.
25. Re:even... execute your code backwards. by Anonymous Coward · 2014-05-24 03:52 · Score: 0
  
  "And such"?
  You make a sweeping generalization with "all" and then a sweeping qualification, with no explanation as to why black holes "and such" are different.
  That's not to say that anyone I know of has a correct or complete answer about _whether_ black holes (or anything else) deny _local_ microscopic reversibility (in the general sense of CPT symmetry or even more generally Poincaré symmetry violation) or whether _local_ non-reversibility (versus any normal sense such as T, TCP, or Lorentz symmetry) is real or only emergent or apparent behaviour (e.g., where non-reversibility is not strictly occurring _locally_, i.e. in the limit where spacetime intervals go to zero or equivalently in a local section of the fibre bundle), although your statement suggests that you do, so I'm all eyeballs.
  That is to say that you do not seem to know that there are implications to what you said, which suggests that you do not know what you are talking about.
Would have come in handy ... by quax · 2014-05-21 16:00 · Score: 0

... when I took the EdX's CS191x Quantum Mechanics and Quantum Computation course.
1. Re:Would have come in handy ... by Anonymous Coward · 2014-05-21 16:14 · Score: 0
  
  He godwined his blog right off.
2. Re:Would have come in handy ... by quax · 2014-05-21 16:15 · Score: 0
  
  Show me yours, I showed you mine ...
3. Re:Would have come in handy ... by quax · 2014-05-21 16:23 · Score: 1
  
  Should have included this in the previous comment, but couldn't find the link at first.
  What I did use occasionally when taking the course was this little browser based gem. While certainly not nearly as powerful as this Google simulator it was still quite useful.
Neat! by Nemyst · 2014-05-21 17:36 · Score: 1

Ignoring the typical Slashdot cynicism (and often lack of understanding disguised as such), this is actually pretty damn neat! Quantum mechanics and quantum computing using the gates model aren't intuitive, especially not for people without a physics background, so this could really help learning the fundamentals of quantum computing. Being able to visualize the state of the qubits at each step of the process as something other than a big formula is a pretty big deal.

As it is right now, QC is pretty much just taught using pen and paper, but I think this deserves some attention too. I don't think many people in the classroom understood what the hell Shor's algorithm was doing when the prof presented it (I know I didn't), but perhaps with a more interactive demo it'd be a bit easier to grasp. Grover's algorithm would also be extremely cool to watch unfold, I think.

Isn't it ironic that a consumer graphics card can simulate more qubits than most actual quantum computers have right now?
1. Re:Neat! by hweimer · 2014-05-21 20:04 · Score: 1
  
  Isn't it ironic that a consumer graphics card can simulate more qubits than most actual quantum computers have right now?
  No. If it were the other way around then quantum computing wouldn't be an open research problem but a multi-billion dollar industry.
  
  --
  OS Reviews: Free and Open Source Software
Wait...WTF, crypto dead? P=NP? by Anonymous Coward · 2014-05-21 18:54 · Score: 0, Interesting

Hold on a minute. If it's possible to simulate qubits using, at the bottom, bits, and, if qubits and quantum computing allow for performing NP calcs in parametric time (and hence breaking crypto), then haven't we already been able to do all of these things for decades?
1. Re:Wait...WTF, crypto dead? P=NP? by Anonymous Coward · 2014-05-21 19:20 · Score: 0
  
  That assumes that the simulator can run all operations in parametric time.
2. Re:Wait...WTF, crypto dead? P=NP? by wonkey_monkey · 2014-05-21 19:28 · Score: 1
  
  If it's possible to simulate qubits using, at the bottom, bits, and, if qubits and quantum computing allow for performing NP calcs in parametric time
  Being able to simulate qubits doesn't mean you can do so in parametric time.
  One can simulate a few molecules chemically reacting, but you can't reasonably do so at a molecular level for a macroscopic sample - yet in reality both would take a similar amount of time.
  * the above is an uninformed guess
  
  --
  systemd is Roko's Basilisk.
3. Re:Wait...WTF, crypto dead? P=NP? by Anonymous Coward · 2014-05-22 00:35 · Score: 0
  
  This is such a common point of confusion that the tagline from Scott Aaronson's blog reads: "Quantum computers are not known to be able to solve NP-complete problems in polynomial time, and can be simulated classically with exponential slowdown."
  http://www.scottaaronson.com/blog/
4. Re:Wait...WTF, crypto dead? P=NP? by neverutterwhen · 2014-05-22 01:13 · Score: 1
  
  Hold on a minute. If it's possible to simulate qubits using, at the bottom, bits, and, if qubits and quantum computing allow for performing NP calcs in parametric time (and hence breaking crypto), then haven't we already been able to do all of these things for decades?
  Oblig xkcd - http://www.xkcd.com/505/
  
  --
  My appreciation of Douglas Adams is far deeper than yours.
parent is full of disinformation by Anonymous Coward · 2014-05-22 00:04 · Score: 1

Parent post is so full of (intentional?) disinformation that it hurts.
Why haven't we been doing this for decades? We have. The only novel part here is "in a web browser." Simulation is not a new concept. Any nondeterministic computing problem can be simulated by a deterministic machine, and vice versa.
Second, instruction runtime on the simulated machine does not correlate with the runtime on the physical machine -- at all. A deterministic machine can simulate a nondeterministic one in O(2^n) by trying every possible combination. More on this later.
Third, integer factorization and graph isomorphism are two algorithms known to be be in NP ^ coNP. If P != NP, that means these algorithms are not NP hard, so the fact that a QC can do integer factorization faster than a general purpose machine says nothing about how fast a QC can solve NP complete problems.
Finally, a QC with only 22 qbits is hilariously pointless, because 22 bits can be brute forced in O(1) with 512kiB state table. I won't even be impressed when someone makes a 44 qbit machine, because that state table will fit on a 2TiB hard drive. I will be impressed if anyone ever builds 128 qbit quantum computer in my lifetime that (1) can solve 128qbit problems in less than a second, (2) runs faster than the fastest known conventional supercomputer at the time, (3) uses less power than a desktop PC, and (3) costs less than a desktop PC. Until then, meh.
1. Re:parent is full of disinformation by flargleblarg · 2014-05-22 06:59 · Score: 1
  
  I agree with you, but I have a nit to pick:
  You wrote "512kiB". This is incorrect. It should be "512KiB". Although "k" is the prefix for "kilo-", there is nothing such as an "iB", so the use of "k" is inappropriate here. Note that the prefix "Ki" is for "kibi-" and it applies here to "B" for "bytes."