I knew the initial claims IRAN made were absurd and baseless. HOWEVER, deep down, somehow, there
might be some truth to it, "Because since quantum computers are inherently parallel, FPGAs can make a good fit for
small-scale quantum simulations".
So that means emulating qubits in a FPGA fabric is feasible ? YES. Turns out, this CAN be true to some extent, at least technically. But was Iran showcasing this ? Hello NO.
But anyway, for this we need to go back in history. So since, it is a known that
classical computers are slow at running quantum computing algorithms, FPGAs on the other hand, could have a chance at
qubit simulation, for improving performance.
This was already demonstrated in a paper, dated 2004, by A. U. Khalid. Here's the link:
Quantum computing offers immense speedup in performing tasks such as data encryption and searching. The quantum algorithms can be modeled using classical computing devices, however classical computer simulations cannot deal efficiently with the parallelism present in quantum algorithms. The...
ieeexplore.ieee.org
The PDF version of the Master Thesis can be downloaded here. Luckily I also had this on my backup drive, lol:
🙄
And the next submission if I can recall correctly, was an FPGA-based real quantum computer emulator paper/thesis, by Jakub Pilch.
While we cannot efficiently emulate quantum algorithms on classical architectures, we can move the weight of complexity from time to hardware resources. This paper describes a proposition of a universal and scalable quantum computer emulator, in which the FPGA hardware emulates the behavior of a...
link.springer.com
In the above paper by Pilch, emphasis was given on
shifting the processing from time to space, by using slow sequential CPU processing to hardware complexity, and FPGA's fabric parallelism which could have been configured. But at what cost or compromise ?
Turns out, the tradeoff between time-space was limited to two qubits, since FPGAs have a limited number of cells. So that would be four states. So by filling up the entire FPGA, they were essentially able to implement a
two-bit quantum XOR. So much ado for real-world military field applications ? Basically, not.
In the previous paper by Khalid, they took a more parallel approach, and the concept was to allow for nine simulated qubits on the same FPGA, but this was far slower and impractical. Conclusion based on both these papers, showed that the
tradeoff between speed and FPGA fabric space was quite large or aggressive, so any FPGA simulation has to be either "very small" or "very slow".
Take your pick by choosing just one between these two !
That's why Google recommends renting a bunch of GPUs instead, for practical simulations.
quantumai.google
So in the end, classical simulations will always come up short, because we are looking for parallelism and speed in a real quantum computer. So the previous Iranian claims were completely off track. Glad it halted !
I still can't and won't overlook "FPGA-based qubit simulation" though.
Facebook
Twitter
Instagram