Why One Simulated Second Took 40 Minutes
A brain simulation does not simply play back a second of activity. It repeatedly calculates how modeled nerve cells change and how signals move between them. Those calculations represent tiny slices of simulated time; the computer must finish one slice before it can reliably determine what happens in the next.
That makes the workload demanding even when it runs across 82,000 processors. A processor can handle part of the model, but the parts are connected: when one group of simulated cells affects another, the computer has to exchange information and keep the calculation coordinated. More processors provide more computing power, but they do not remove that communication or the need to work through the sequence of events.
Simulated Time Is Not Wall-Clock Time
The 40 minutes measure the computer’s time, not the duration represented inside the model. In this case, the system spent 40 minutes calculating one second of modeled brain activity. That gap is a measure of computational cost—not evidence that the simulated brain was moving slowly.
The distinction matters because a model can be useful even when it cannot run in real time. Researchers can examine the modeled activity, test how changing parts of the system affects others, or study processes that would be difficult to observe directly. But the delay also shows the tradeoff: a detailed, interconnected model can demand enormous computing resources just to advance a small stretch of simulated time.
The striking implication is that a vast machine may still need far longer than a living brain to calculate what happens in it.
