[Beowulf] [EXTERNAL] Re: Is Crowd Computing the Next Big Thing?

[Lux, Jim (US 337K)]

On 11/27/19, 12:43 PM, "Beowulf on behalf of David Mathog" <beowulf-bounces at beowulf.org on behalf of mathog at caltech.edu> wrote:

    On 2019-11-27 11:23, beowulf-request at beowulf.org wrote:
    > Send Beowulf mailing list submissions to
    > 	beowulf at beowulf.org
    > 
    > To subscribe or unsubscribe via the World Wide Web, visit
    > 	https://beowulf.org/cgi-bin/mailman/listinfo/beowulf
    > or, via email, send a message with subject or body 'help' to
    > 	beowulf-request at beowulf.org
    > 
    > You can reach the person managing the list at
    > 	beowulf-owner at beowulf.org
    > 
    > When replying, please edit your Subject line so it is more specific
    > than "Re: Contents of Beowulf digest..."
    > 
    > 
    > Today's Topics:
    > 
    >    1. Re: Is Crowd Computing the Next Big Thing? (Chuck Petras)
    >    2. Re: Is Crowd Computing the Next Big Thing? (Alexander Antoniades)
    > 
    > 
    > ----------------------------------------------------------------------
    > 
    > Message: 1
    On Wed, 27 Nov 2019 18:19:20, Chuck Petras wrote:
    > My question re financial viability was prompted by this statement in
    > the Neocortix article:
    > 
    > “And phone owners could be paid for the service to rent out their
    > phones’ computing capacity. Neocortix claims on their website that top
    > users can earn up to $80 a year for a phone that’s engaged in
    > computing for 8 hours a day; if available for 24 hours, it can earn up
    > to $240 a year.”
    > 
    > So that works out to around US$0.023/hour.
    
    Whatever payment they make must also be weighed against the phone 
    battery running down much faster than it otherwise would.  Seems like 
    the sort of background application which, if one forgot to turn it off, 
    could easily result in a dead phone just when it was needed at the end 
    of the work day.
    
    The idea that unused cycles are somehow "free" I think dates way back to 
    the time when computers had fixed clock speeds and the amount of power 
    the CPU used was nearly independent of what they were doing.  These days 
    unless set otherwise ("max performance" or the like) most machines turn 
    their clocks way down when they are not busy.  So burning all of those 
    "free" cycles will result in substantially higher power consumption.  
    Phones do that even more than other computers.  It seems likely that if 
    the application was only running when the phone was plugged into its 
    charger that level of payment could cover those extra electricity costs.
    

--
And, batteries are "cycle life" limited, so running the battery up and down will cost you in earlier battery/phone replacement.  Actually the life is more like "joules passed through the battery" (i.e. 100 cycles from 80% to 20% is similar to 200 cycles from 80 to 50 or 50 to 20)

With respect to "free cycles" in desktop computers - back in the day, 10-15 years ago, a bunch of folks made measurements on cluster nodes of one sort or another.  As I recall, there *is* a power consumption change between full load and not, but there's a significant "background load" that is more than 50% of the total power consumption.

If you start looking at things like large FPGAs, the old Virtex II family had a very strong dependence between number of gates clocking and power consumption. The Virtex 6, though, has almost no dependence on clock rate or circuit complexity. It's dominated by quiescent current (the leakage through 600 million devices on the die), and that is very strongly affected by die temperature.  It rises exponentially - so the increase from 0 to 20C is a tiny fraction of 40C to 60C or more.