That’s such a great way to communicate the current state of the field. Do you have a reference for that? And so you know what the largest number factored for algorithms that don scale into the cryptographic levels is?
"Here we experimentally demonstrate the factorization of two bi-primes, 4088459 and 966887 using IBM's 5- and 16-qubit quantum processors, hence making those the largest numbers that have been factorized on a quantum device."[1]
It looks like some real progress has been made with qbit quality and stability in the last few years! I think that 21 figure held for quite a while. Maybe it still does hold and I'm misunderstanding some of the nuance.
I was in a meeting with other developers and we were discussing pros/cons for a module that needs to be rewritten. The discussion came to the point where the team has different opinions on network overhead. That was the time actually I remember using concrete numbers/values can help you win debates/discussions easily. :)
How come a meaningless question, which is probably just a click-bait for promoting a web site, is the first result on HN?
I'm worried about HN. This is totally subjective, but I feel that technical content related to computers is now sparse in the top pages, while it was preponderant. And the culture of ads and lucrative visits is damaging the site, like everywhere online.
Maybe this is false and the change is within myself, anyway I should keep away from HN.
Wow. So, I was preparing for technical interviews I came across that post from 2009. And the values in the tables seemed relevant but potentially dated. And so the question to HN is directed toward whether those values are still valid, and whether other values are relevant now. The title of the post is a reference to the title of the OP.
Oh, I think I completely missed that it was a question. Having this comment in the original question probably would have helped. So, to answer the question, there are definitely some numbers in that chart that are closely related to the speed of light, so they can’t ever change. Most other numbers are still reasonably close due to the end of Moore’s law. But, the list itself feels a little dated since it references spinning disks and doesn’t mention SSDs. These days, I’d maybe include something about modern GPUs, number of flops & typical memory/bus bandwidth. It’s also fun to extend your posted chart with longer times like around the world, or out to geosynchronous orbit or the moon and back. I do sometimes ask an interview question about what is the fastest way to get a terabyte of data from LA or NY with specific assumptions, or something along those lines.
27 comments
[ 4.5 ms ] story [ 73.5 ms ] threadmaybe "What are the computing performance numbers developers should know in 2019?"
You're going to have to qualify this more, because I can factor 2^googol pretty well…
It looks like some real progress has been made with qbit quality and stability in the last few years! I think that 21 figure held for quite a while. Maybe it still does hold and I'm misunderstanding some of the nuance.
1 - https://arxiv.org/abs/1805.10478
53 - approx number of those known to be associated with advertising and tracking
Sorry to post something negative, I think the idea in the post is cool, but in 2019 everyone should know this...
I'm worried about HN. This is totally subjective, but I feel that technical content related to computers is now sparse in the top pages, while it was preponderant. And the culture of ads and lucrative visits is damaging the site, like everywhere online.
Maybe this is false and the change is within myself, anyway I should keep away from HN.
In case anyone wants to read any of the many previous discussions: https://hn.algolia.com/?query=latency%20numbers%20every%20pr...