3 ms·

what workloads would i run on these?

• krastanov 7y ago
  On this particular one, probably only research code that tests various small scale optimization algorithms and tests ways to "calibrate" the hardware. In a couple of years, on the next gen of such hardware, the hope is to run things like optimization algorithms and chemistry simulations that just begin to outperform what is possible with classical hardware of reasonable size and cost.
  • ptah 7y ago
    so it will eventually run stochastic gradient descent faster than GPU?
    • krastanov 7y ago
      The short answer is no. The long answer is much more interesting. The premise of your question is a bit misleading. Quantum computers do not run classical algorithms faster, rather they run algorithms specifically designed to take advantage of the quantum coherence. For many tasks this does not provide an advantage. For some very special (and occasionally useful) tasks, the advantage seems to be exponential. Simulating quantum mechanics (as in chemistry, material design, drug design) is one of these tasks. Some linear algebra tasks are also faster. Lastly, some researchers are hopeful that novel gradient-free optimizations algorithms exist for quantum hardware, but this is a bit more controversial.