Episode 111 – 18th November 2011

We speak to TOM MORELLO from Rage Against the Machine about Occupy, Mark Brandon the academic advisor on Frozen Planet, to Lewis Dartnell about Phobos-Grunt and much, much more!

[Direct MP3 Link] [Podcast Feed] [Add to iTunes]

Like the show? Support us by subscribing for £5/month here (or £10/month here).

Tom Morello on Occupy (1:50) by James O’Malley & Liz Lutgendorff (ft Tom Morello!)
Phobos-Grunt (13:52) by Kash Farooq (ft Lewis Dartnell)
Frozen Planet by Liz Lutgendorff (ft Mark Brandon)
Microprocessors at 40 by Sean Ellis (ft Greg Yeric)
Political Lies (43:01) by Steven Sumpter
The Christ Myth Theory (50:51) by Salim Fadhley (ft Robert M Price)
Engaging with others (59:32) by Tannice Pendegrass
The sketch at the end is by David Lovesy & Brian Two

Follow-Up Links:

8 thoughts on “Episode 111 – 18th November 2011

  1. Greg Yeric also pointed me at a couple of websites that may be useful to anyone interested in how chips are made: Chip Making’s Virtual Museum and also 40 Years and Holding – A History of Optical Lithography.

    And just for clarification, when making the connections on the chip, it’s the aluminium layer that goes on first, then the photoresist, then it’s exposed, then it’s etched. Apologies – the order wasn’t as clear in the edit as in the original inverview.

  2. Liked the bit on chip making. As awe-inspring as they are though, the technological innovations that have brought us this far and are helping us into the near future don’t alter the underlying physical limits on Moore’s law. There isn’t that much room at the bottom.

  3. Could you include direct MP3 links to the longer segments? I only really listen to podcasts in the car, so the embedded player isn’t a solution for me.

    For things in the extra feed, I can get them with my regular podcatching software, but I haven’t figured out how to do it for things included as subsidiary links.

  4. Pete,

    While I agree that there must, eventually, be fundamental limits to Moore’s law, I disagree that we are near them. There won’t be a hard stop – things will get harder for a while, then there will be breakthroughs that give us more compute power, and the process will repeat.

    The thing is, it’s not just feature size scaling that gives us more compute power. Other possible technologies that we didn’t go into (because I wanted to focus on the manufacturing process) include 3D, parallelism, and optical interconnect.

    As Greg said, it’s an interesting time to be in the industry – there are big changes coming and it’s up to us to shape how they are used.

    Finally, just to clarify my position, I work for ARM but I’m primarily a software person. I’m not associated with the actual design of fabrication of logic gates – I’m about three levels of abstraction away from that world. In any case, ARM does not actually manufacture chips, instead we design them, then these designs are incorporated into system designs by our design partners, and then they are manufactured by our manufacturing partners. Greg, as he says in the longer interview, works with those manufacturing partners to ensure that our chip designs will work efficiently on their processes as they evolve.

  5. Sean,

    Moore’s law is inexorably linked to features size – because its continuation is dependent not just on transistor count growing exponentially, but on side effects of transistor count not doing so.

    So long as you can keep shrinking feature size (we have maybe a decade or two of that before the Heisenberg Uncertainty Principle starts being a real bitch, as Michio Kaku pointed out) you can get more computing power without having your electrical power requirements, your heat production, or the mass or volume of chip increasing exponentially as well. Once feature size stops shrinking, you have maybe another doubling or two from clever component design, but then you are faced with future doublings leading you inexorably to a unwieldy solid brick of silicon that puts out more heat than a car engine, or home computers the size of filing cabinets with thousands of cores.

    I have no doubt that very smart people at ARM, and other chip designers, will think of innovative ways to squeeze the last drops of exponential growth out of semiconductor technology – but the fact that people are having to be so clever with component design instead of just saying “more of them, and smaller” indicates an approaching limit.

    This is a problem with going for exponential growth of any kind; geometric advancements only buy you linear amounts of time to continue growing. Its not even a matter of science or technology, just one of maths.

  6. Nothing to do with this podcast, but remebering the BC/BCE argument from a few weeks back. Here’s a chance to laugh at the Daily Mail

    Noticed this when looking at the Daily Mail website http://www.dailymail.co.uk/sciencetech/article-2065254/Coins-Jerusalems-Western-Wall-hints-sacred-site-older-Herod.html

    “Coins dated 17/18 CE were discovered beneath the Western Wall of Temple Mount, providing scientific confirmation that the Western Wall and Robinsons Arch construction were not completed in King Herods lifetime”

Leave a Reply

Your email address will not be published. Required fields are marked *