Quanticles

My biggest concern about F-35's is how quickly they can be made in wartime situations. They have so many sophisticated instruments... can they really be mass produced? In WW2, Germany had the best tanks, but the Soviets and the Americans overwhelmed them with numbers. You can say that WW2 is an old war, but if you're wrong do we really have a reliable plane that we can mass produce anymore?

As noted in the FAQ, it was actually against the rules for an already existing company to show up.

That's part of the reason why contracts are not allowed - everyone owns the IP then so there cannot be an owner-employee relationship in any sense.

Analog weights can save a lot of delay/power/cost if you can implement them right, easier said than done

Sorry, I'm going to keep those details secret for now :)

These are all good questions/points

History is something we need to contend with, not just for neural networks, but also for analog computing which has a similarly troubled past.

For NN history, there has not actually been a market for NN accelerators until recently. You can see this because:

1. No NN algorithm was worth accelerating until AlexNet came along in 2012

2. What commercial products even use NN now? Currently it is mostly just voice recognition which is processed server-side.

Right now we are not attempting to go after any markets that a GPU would be sufficient for the reasons you mention; we're sticking to products that can only work with our technology. By the time we went after an overlapping market our credibility would be established and that wouldn't be an issue.

Actually, I can give you a better answer...

An ASIC is always going to be at least 10x better than a CPU/GPU for performing the same algorithm. The question isn't whether or not an ASIC can beat NVIDIA, the question is whether the target market is large enough to support an ASIC company.

At Isocline we assume that this market IS big enough to support an ASIC. Our competition is not NVIDIA, it's the future all-digital ASIC company that can do the same thing, but without all of the whiz-bang technology. If we have to, we could probably fall-back to be that all-digital company, but I'd prefer to maintain our technology advantage.

NVIDIA's advantage is flexibility, there's always going to be a lot of demand for that.

We make use of non-volatile memories throughout the chip, which stores the configuration and weights

I guess you'll have to wait and see

We're in the process of fabricating a prototype and are not publicly releasing detailed estimates at this time.

I can say that the cost depends on what you want to do - systems can range from less than 1mm^2 to the entire reticle depending how much performance you want.

We do the learning on GPU and efficiency is not a concern because the learning result goes out as a firmware update. Let's say it takes 2 weeks to train a neural network - the customer never sees that, they just get the firmware update. Similarly, we can use a lot of GPUs to train one network because that training result goes out to many chips.

The products that we are creating are reprogrammable and reconfigurable, just like a GPU or FPGA. Updates are like a firmware update. Our hardware would be no more obsolete over time than a GPU or CPU running in its place, and given the huge improvements over CPU/GPU, it would be many years before CPU/GPU would catch up to any particular product anyway.

They are not able learn on chip - that is a non-starter and not particularly useful anyway. Customers dont want self-driving cars that need to learn how to drive, they want self-driving cars that already know how to drive.

Target applications like self-driving cars would require deep convolutional neural networks like NVIDIA's Drive PX

There is a lot of interesting research out there on analog computing, analog neural networks, and far-out stuff like transistor-free computing. Going from research project to product on Digikey is a really huge leap for most research though. Designing a chip is very expensive, so the product better be a slam dunk. Most of these analog neural network projects can do some sort of learning with small black and white patterns, which does not approach the accuracy or scale of software neural networks.

What we're working on is an accelerator for the convolutional neural networks that are winning competitions like ILSVRC. Even that by itself is insufficient for a business case, though. You also have to have end application in mind too, and that end application better be power intensive or performance constrained enough that software cannot accomplish what you need it to do. Because, if software is good enough, then why take a risk on a fancy new hardware component?

Our company works on this kind of stuff for those who are interested (http://isosemi.com)

We're seeing more like 10-100x improvements in energy efficiency and performance, not 10000x, unless the comparison point is a full blown CPU/GPU.

Agreed. You could say that public libraries have a natural limit to their degree of sharing, whereas online filesharing does not.

For instance, when a popular book comes out the library might have a couple copies. If hundreds want to read those copies then they'll have to wait. Many of those people will be unwilling to wait and buy the book instead, which supports the author.

1. If you're going to tax universities, then you need to tax churches too. Good luck.

2. If you're going to tax non-profits, then why have non-profits?

3. Do we really want to penalize fiscally responsible organizations when even governments are going bankrupt? What values are we trying to support here?

Yeah, you see a few people with wealthy beginning say that.

You also see plenty of success stories of people with much humbler beginnings say that.

Which you choose to focus on is either going to motivate you to work hard or be an excuse to not.

These lines of reasoning (it's talent/money, not work/grit/focus) appeal to those who don't want to take responsibility for their own success or failure.

Even if you had the same body as when you were 25?

Corporations are supposed to operate on the behalf of their shareholders - insider trading is in direct conflict with that

Makes sense, thanks

> "Oh, so your company doesent produce anything real?"

The euros are real

If you're in deep space then you're not near any particular star. That's going to reduce the available starlight by several orders of magnitude, right?

This article relies on carefully constructed images that maximize one particular outcome by summing up lots of small errors into it.

For it to work, the pixels have to be very accurately tweaked. If the tweaks were off by one pixel, the whole thing would fall apart.

The assumption is that this cannot be done to a person. But there is no way to put in a pixel-level "exploit of sorts" into a person to test that theory.

The real answer is probably that a little bit of noise on the input probably disrupts the exploit. It could never happen to a person because eyes have noise. At the same time, it could never happen to a robot either because cameras have noise.

Economic inequality is a hot topic right now. Employers knowing everyone's salaries, but not the employees, puts the employees at a disadvantage when negotiating salary.

One option would be to require all companies to publish employees salaries. Would this requirement be more or less onerous than other options?

One argument is that it hurts businesses - but if every business is required to do it, then it is an even playing field. Obviously that cannot happen in an international business setting - so would it put the countries that implement it at a disadvantage, or would the economic gains outweigh the decrease in competitiveness?

At least the drama is well understood

/s

Finding and hiring the right person is very time consuming, training up that person is very timing consuming, and when someone leaves, training someone to take over their responsibilities is very time consuming. Combining all of those these together, each time a person leaves a company it incurs a cost of about 6 months of their salary. I'd rather pay more for someone that's going to stick around then hire people who are likely going to get bored and leave. Many companies try to figure out way to reduce turnover - it's very expensive.

This sounds like a blunt combination of popular terms

Austin is a great city with great food, music, and brews. And you can even afford a nice house too

3 days for 12-14 hours a day would be nice - does he think those should be M/W/F or T/W/Th ?