[ars]

> 7% extra efficiency per year means electric cars will have twice the capacity in 10 years...and then double up again in 10 more years...and so on.

An example of the dangers of extrapolating.

CPU speeds kept going up - until they didn't.

Battery tech is not going to keep going up 7% each year. Although perhaps his 60% improvement will show exactly at the right time to effectively be 7% better than the previous year.

[semi-extrinsic]

We are actually not so far away from hitting the ceiling with batteries. There are fundamental physical reasons for why you can't put much more than 1eV per atom in a battery. That equates roughly to 850 Wh/kg, if memory serves, and we are currently pushing Li-ion towards 300 Wh/kg. So we can only keep growing at 5% for another couple of decades or so. Regardless of battery chemistry.

What people often do to hide this fact is talk about energy density, Wh/L, where growth can continue for longer, at the expense of making batteries heavier.

[Filligree]

Gasoline manages 12kWh/kg, and I'm sure I've heard people talking about batteries at least approaching that density before, even if only on a theoretical basis.

Leaving aside how safe I'd feel with a supercap of that density under the hood, are you sure about the 800Wh/kg limit? I'd be very interested in seeing a reference.

[semi-extrinsic]

The 1eV per atom is really a rule of thumb, but easy to understand: At that value you are getting dangerously close to the ionization energy 13.6 eV for hydrogen, 5.4 eV for lithium), so you can no longer have a chemical battery. Note also that this is average eV for all atoms in the battery, so those actually providing electricity will be carrying atound 1.5 - 2 eV.

See e.g. here: http://www.ohio.edu/people/piccard/radnotes/radioactive.html

[jontro]

Found this article discussing some limits http://chargedevs.com/features/three-of-a-kind-polyplus-reac...

[mikeyouse]

CPU speeds stopped increasing because the market stopped focusing on CPU speed (coincidentally to this discussion, the change was largely to reduce power demands). Transistor count has kept plugging right along;

http://i.imgur.com/FvVrnn4.png

[yongjik]

I think it's the other way? That is, market stopped focusing on CPU speed because they hit the wall and clock speed could no longer be used as a differentiating factor.

[ajuc]

Yes, market would be more than happy to have 100GHz CPUs by now, it would make scaling everything much easier.

[semi-extrinsic]

I think it is misleading to say it was due to reduce power demands. Rather it was due to needing to cool the damned things. If you were to have a 500W TDP chip, you would either need a crazy huge cooling system or a much bigger surface area. Consumers want light and thin, so Bob's your uncle.

[lkbm]

Even if consumers were fine with big heavy things (and except for maybe phones, I don't think the CPU weight/size are an issue for consumers), light travels roughly 1ft/ns. You want fast, you have to be small.

[mikeyouse]

Surprised by the negative reaction.. More technically, processor speeds required far more electricity to increase clockspeed which required much more energy to disappate in the form of fans and heatsinks.