Huge Breakthrough in Quantum Computing

Показать описание

Google just announced a crazy new quantum computing chip. It can solve a problem in 5 minutes that would take one of the fastest supercomputers 10 septillion years! BUT that problem was a “benchmark” called “RCS”, chosen specifically because it’s extremely hard for a supercomputer - which are still better at basically everything practical. The real reason you should care about quantum computers and this new chip is less about beating supercomputers, and more about doing new things - like really accurate simulations of nature at a quantum level that could be used for drug discovery or battery development or more.

And this new chip matters for how we might get there because of how it deals with mistakes. One of the biggest challenges with quantum computers is errors. The information inside of the “qubits” can get messed up by their environment. Typically the more “qubits” of information, the more errors.

Google’s chip has 105 qubits, much smaller than the largest quantum processors, but it’s the first to show the number of errors reduce as more qubits are added - which could be a big deal for the future of quantum computers.

If you want to know more about the real future of quantum computing, Marques and I made a longer video! For optimistic science and tech stories, subscribe for more of our show Huge If True.

#quantum #willow #scitech #science #quantumcomputing

Cleo Abram

Рекомендации по теме

Комментарии

Finally I can run Crysis on medium graphics

arbitrario

Finally someone actually addressing the fact that it's 10 septillion times faster at solving a VERY SPECIFIC problem. Most media is saying it like we now have a general purpose computer that's somehow 10 septillion times faster.

wlockuz

If you don't know it actually requires -273°C to operate

FACTOPTENZA

Can't wait to simulate me shitting my pants in 56 degree weather with 2h34m remaining in my walk home from Taco Bell to know if it's worth it or not

Marusii

I read the original arXiv paper. I think that Google did a deservice to itself by focusing on simulating the dynamics of the physical device instead of publicizing their HUGE improvement in logical error rate.

tahahammadia

The real question is-
can it run doom?

ItayBroitman

I dropped my phone behind my bed and had to hear this for like 7 minutes 😭😭😭

anjuscuccos

I read an article about this, but in the first 10 seconds of this video, you explained this better than that entire article did. Awesome content!

cgifford

The way I shouted “Roblox?!” When I saw that chip. I’m addicted.

Tree

Bro tilt that 15% and you got a lawsuit

Reallillycurxe

Remember: quantum computing is for graphics cards and password breaking, not general computation.

crimsondragon

For those wondering what the next actual commercial computing breakthrough will happen before we crack quantum, check out photonic CPU blocks.

FrotLopOfficial

Two things: 1, the biggest thing is cracking RSA, making any "secure" message from the last 30ish years readable. People have saved messages for exactly this reason, particularly around places where government documents might be shared. 2) Qubits can be used for error correction, so this only matters if it's overall more efficient. (An oversimplified example: if you have 15% of error, 85% correct per qubit, 10 qubits has a 0.8^10 or only about a 19.7% chance of no error. If that goes down to 10% error, 90% correct at 100 qubits, that's still 1/37648 chance to succeed. If those 100 qubits cost as much as 1000 normal qubits, you can just assign multiple qubits to a task. Assigning 10 qubits and taking the consensus (most common answer), you get better than 1/2 success (57%), instead of 1/37648 (0.000027%)

collinguo

Starting in 1943, the Allies used a dozen SIGSALY stations to allow encrypted voice transmission, using a digital systm to essentially frequency-hop the voice audio rather than the RF. Rather than a paper tape it used a large record with the encrypting/decrypting key as an audio recording. It required a very large support facility (30 full-height vacuum tube racks and 30 kW of electrical power), and a given record could only contain 12 minutes worth of data.
And the issue of RF reception underwater was less important than the need for the operator to see both the torpedo AND the target. As you might expect, a torpedo running underwater is not easy to see.

wirebrushproductions