Editorial: Gaming / Facebook / Twitter / Coverage / Instagram / Discord
Three years ago, Google unleashed the power of its quantum computer called Sycamore, which carried out a fairly complex calculation in just 200 seconds. Processing that according to the company itself, would take something like 10,000 years to the Summit supercomputer, which was developed by IBM for use at the OAK Ridge National Laboratory in the United States.
Google’s Sycamore Quantum Computer
Quantum computer that reached a performance of 53 qubits, which they claimed in 2019, was no match for any computer that will use conventional hardware. Something that some Chinese scientists have thrown overboard.
You can read: Quantum computers will reach universities, schools and educational training centers in Chile
This is Pan Zhang, a physicist at the Chinese Academy of Sciences, who, together with his team, demonstrated that they can beat Google’s quantum computer, using 512 GPUs. To do this, as explained on the El Chapuzas Informático site, the scientists created a 3D mathematical matrix called a tensor network, which has 20 layers and each of them has 53 points, one for each qubit. Thanks to this simulation, they were able to multiply all tensors, using the aforementioned 512 graphics cards.
In the end, they were able to calculate a million of the 9 quadrillion possible strings of numbers in 15 hours using all 512 GPUs, blowing Google’s argument to the ground and showing that they can do what their quantum computer could do, on a conventional one.
#build #GPU #Graphics #cards #manages #overcome #power #Googles #quantum #computer #Tarreo