Current software is very energy efficient -- and on so many levels. I worked developing code used in the Windows Smartphone and it was during that time that I had to first think hard about the energy efficiency dimension in computing -- as measured by useful work done per unit of energy. The engineering management in that group was constantly harping on the need to produce energy efficient code. Programmers are deeply engrained with a lot of bad habits -- and not only in terms of producing energy efficient software. For example most developers will instinctively grab large chunks of resources -- in order to ensure that their processes are not starved of resources in some kind of peak scenario. While this may be good for the application -- when measured by itself -- it is bad for the overall footprint of the application on the device (bloat) and for the energy requirements that that software will impose on the hardware. Another example of a common bad practice poorly written synchronization code (or synchronized containers). These bad practices (anti-patterns in the jargon) can not only have a huge impact on performance in peak usage scenarios, but also act to increase the energy requirements for that software to run. I think that -- with a lot of programming effort of course (which is why it will never happen) that the current code base, and not only in the mobile small device space, where it is clearly important, but in datacenter scale applications and service (exposed) applications as well -- that the energy efficiency of software has a huge headroom for improvement. But in order for this to happen there has to first be a profound cultural change amongst software developers who are being driven by speed to market, and other draconian economic and marketing imperatives and are producing code under these types od deadlines and constraints. If there is a theoretical minimum that derives from the second law of thermodynamics it must be exceedingly far below what the current practical minimums are for actual real world computing systems. And I do not see how a minimum can be determined without reference to the physical medium in which the computing system being measured is implemented. In fact how could a switch be implemented without it being implemented in some medium that contains the switch? -Chris
________________________________ From: L.W. Sterritt <lannysterr...@comcast.net> To: everything-list@googlegroups.com Cc: L.W. Sterritt <lannysterr...@comcast.net> Sent: Friday, September 20, 2013 3:27 PM Subject: Re: What gives philosophers a bad name? Chris, It's the Landauer argument relating energy to information, as Frank wrote. There is a summary article in the same issue of Nature: Philip Ball, "The unavoidable cost of computation revealed," Nature (March 07, 2012). Ball references the analysis mentioned in my last post; It's the ultimate thermodynamic limit, and how close we approach that limit will depend upon technology as you discuss. Intel, Apple and others are making significant improvements in the efficiency of their chip sets, but there is never enough battery life in a smart phone. LW On Sep 20, 2013, at 2:40 PM, Chris de Morsella <cdemorse...@yahoo.com> wrote:, Interesting. Do you know what assumptions went into their analysis? > >I would think that this is a medium dependent value; i.e. what underlying >medium is the relying on to hold its logical state. Did the researchers >attempt to figure out the minimum scale system (say an electron spin for >example in a spintronics device -- which have not yet been built, but which is >very much one of the future paths to the ever more small scale and which is >sucking down big R&D money to try to find an economic means of building the >single electron type gates that would rely on the electron's spin. > >What if some hypothetical future technology is able to access more fundamental >sub-atomic systems and a logic state is able to be contained in something far >smaller than an electron. I am speculating here of course because such >technology does not exist (as far as I know), but what if a state could >reliably be inferred -- even if not directly measured -- in something that >begins to approach the Planck scale -- say some property of a vibrating >string. Wouldn't the minimum energy to flip a bit, in this new hypothetical >and vastly smaller scale circuit be considerably less (by many orders of >magnitude) than the energy required to flip a macro gate (which by comparison >even the most miniaturized transistor would be)? > >My point in replying is that the medium and the scale in which the logic is >etched (or some equivalent process for non-lithography based production) are >other drivers that need to be considered, and that lower energy does not >necessarily equate to slower performance. The twenty watt human computer can >solve (especially subtle pattern recognition) problems that bring a >super-computer to its knees; admittedly this is changing as computer >hardware/software improves and better pattern recognition algorithms are >developed, but our energy frugal computing machines cannot be said to be slow >-- yes I know nerve impulses travel at a vastly reduced speed as compared to >electrons flipping logic circuits, but the brain is a massively parallel >architecture and is performing thousands maybe millions of tasks each and >every second. > >-Chris > > >________________________________ > From: L.W. Sterritt <lannysterr...@comcast.net> >To: everything-list@googlegroups.com >Cc: L.W. Sterritt <lannysterr...@comcast.net> >Sent: Friday, September 20, 2013 1:50 PM >Subject: Re: What gives philosophers a bad name? > > > > >Chris, >An article in Nature last year presents a calculation of the theoretical >minimum energy required to erase a bit - independent of the computer: > * Antoine Bérut, * Artak Arakelyan, * Artyom Petrosyan, > * Sergio Ciliberto, * Raoul Dillenschneider& * + et al. >>Nature 483, 187-189 doi:10.1038/nature10872 >L.W.Sterritt >On Sep 20, 2013, at 1:22 PM, Chris de Morsella <cdemorse...@yahoo.com> wrote: > >>> A computation always takes a nonzero amount of energy to perform, >>> theoretically you can make the energy used be as close to zero as you like, >>> but the less energy you use the slower the calculation. >> >>How does that square with the increased (well measured) energy efficiency per >>fundamental unit of logic (single machine operation) -- it takes far less >>energy to perform an elementary logic operation on a modern CPU than it did >>on say a CPU from ten years ago (even if the modern CPU may suck down more >>total power -- it is performing far more work) >> >>Modern CPUs clearly are also operating at much higher speeds. I think you are >>not factoring in the dimension of scale or the physical size of the logic >>container/state-machine. As the size of a logic gate is scaled down it takes >>less energy and can operate at a higher clock speed. >> >>http://en.wikipedia.org/wiki/Performance_per_watt >>"For example, the early UNIVAC I computer performed approximately 0.015 >>operations per watt-second (performing 1,905 operations per second (OPS), >>while consuming 125 kW). The Fujitsu FR-V VLIW/vector processor system on a >>chip in the 4 FR550 core variant released 2005 performs 51 Giga-OPS with 3 >>watts of power consumption resulting in 17 billion operations per >>watt-second.[1][2] This is an improvement byover a trillion times in 54 >>years." >> >>Size (or rather the lack of it) matters in this equation. >>-Chris >> >> >>________________________________ >> From: John Clark <johnkcl...@gmail.com> >>To: everything-list@googlegroups.com >>Sent: Friday, September 20, 2013 10:38 AM >>Subject: Re: What gives philosophers a bad name? >> >> >> >>On Thu, Sep 19, 2013 at 6:10 PM, LizR <lizj...@gmail.com> wrote: >> >> >> >>>> As Rolf Landauer said "Computation is physical", all computations must use >>>> energy and generate heat. And what's the difference between a physical >>>> process and a non-physical process anyway? >>>> >>>> >>>> >>>> I thought it was only erasing the results of computations that had to use >>>> energy and increase entropy? - if so - quibbling, I know, but sometimes >>>> quibbles have important consequences. >> >> >>A computation always takes a nonzero amount of energy to perform, >>theoretically you can make the energy used be as close to zero as you like, >>but the less energy you use the slower the calculation. >> >> >> John K Clark >> >> >> >> >> >> >> >> >>> >>>-- >>>You received this message because you are subscribed to the Google Groups >>>"Everything List" group. >>>To unsubscribe from this group and stop receiving emails from it, send an >>>email to mailto:everything-list%2bunsubscr...@googlegroups.com. >>>To post to this group, send email to everything-list@googlegroups.com. >>>Visit this group at http://groups.google.com/group/everything-list. >>>For more options, visit https://groups.google.com/groups/opt_out. >>> >> -- >>You received this message because you are subscribed to the Google Groups >>"Everything List" group. >>To unsubscribe from this group and stop receiving emails from it, send an >>email to everything-list+unsubscr...@googlegroups.com. >>To post to this group, send email to everything-list@googlegroups.com. >>Visit this group at http://groups.google.com/group/everything-list. >>For more options, visit https://groups.google.com/groups/opt_out. >> >> >> >> >> -- >>You received this message because you are subscribed to the Google Groups >>"Everything List" group. >>To unsubscribe from this group and stop receiving emails from it, send an >>email to everything-list+unsubscr...@googlegroups.com. >>To post to this group, send email to everything-list@googlegroups.com. >>Visit this group at http://groups.google.com/group/everything-list. >>For more options, visit https://groups.google.com/groups/opt_out. >> > -- >You received this message because you are subscribed to the Google Groups >"Everything List" group. >To unsubscribe from this group and stop receiving emails from it, send an >email to everything-list+unsubscr...@googlegroups.com. >To post to this group, send email to everything-list@googlegroups.com. >Visit this group at http://groups.google.com/group/everything-list. >For more options, visit https://groups.google.com/groups/opt_out. > > > > > -- >You received this message because you are subscribed to the Google Groups >"Everything List" group. >To unsubscribe from this group and stop receiving emails from it, send an >email to everything-list+unsubscr...@googlegroups.com. >To post to this group, send email to everything-list@googlegroups.com. >Visit this group at http://groups.google.com/group/everything-list. >For more options, visit https://groups.google.com/groups/opt_out. > -- You received this message because you are subscribed to the Google Groups "Everything List" group. To unsubscribe from this group and stop receiving emails from it, send an email to everything-list+unsubscr...@googlegroups.com. To post to this group, send email to everything-list@googlegroups.com. Visit this group at http://groups.google.com/group/everything-list. For more options, visit https://groups.google.com/groups/opt_out. -- You received this message because you are subscribed to the Google Groups "Everything List" group. To unsubscribe from this group and stop receiving emails from it, send an email to everything-list+unsubscr...@googlegroups.com. To post to this group, send email to everything-list@googlegroups.com. Visit this group at http://groups.google.com/group/everything-list. For more options, visit https://groups.google.com/groups/opt_out.
