Leaving Mac On -- Advantages / Disadvantages
in [Mac Systems]
|
From: Nick Naym on 17 Feb 2010 03:30 In article 2010021622124216807-not(a)dotcom, thepixelfreak at not(a)dot.com wrote on 2/17/10 1:12 AM: > On 2010-02-16 19:03:15 -0800, Davoud <star(a)sky.net> said: > >> Nick Naym: >>> There seems to be two schools of thought: One, which takes the position that >>> leaving the machine on 24X7 shortens its life due to the wear & tear of >>> constant operation; the other, which believes that the wear & tear of >>> constant operation is minimal compared to the cumulative stress caused by >>> repeated, daily powering up and shutting down. >> >> Bzzzzzt! The "stress" caused to a piece of electronic gear by turning >> it on and off is pure myth. Study after study, some published and some >> not, and tons of empirical evidence support that. > > Bzzzzzzzzt! I've worked for a Supercomputer manufacturer for the past > 15 years and a total of 20 in the computer industry. I _routinely_ see > plenty of real world evidence to the contrary. The machines we make > have many thousands (8192 cores currently) and many many Terabytes of > memory and Petabytes to Exabytes of attached rotational disk media. I > am also a professional member of the ACM (Association of Computing > Machinery) and have read numerous articles suggesting what you do. > These articles DO NOT agree with my experience. > > Many of the studies (not all) you posit deal with statistically > insignificant numbers of components. The very large component count > machines we make (whether they be large NUMA SMP machines or large > cluster supercomputers) exhibit statistically significant component > failure after power has been removed for significant periods of time. > These effect tend to manifest themselves more in rotational media than > integrated circuits but nonetheless I see it in all components. > > That said, untoward effects of the power cycle on desktop computers > with relatively very few components these days is likely to be rather > minimal. That would beg the question as to how the effect of cycling the power on/off -- however minimal that effect may be -- compares to the effect of leaving the power on 24X7. Your _actual_ experience suggests that the comparison still would favor letting the machines -- especially the HDs -- run full time. -- iMac (24", 2.8 GHz Intel Core 2 Duo, 2GB RAM, 320 GB HDD) � OS X (10.5.8)
From: Christoph Gartmann on 17 Feb 2010 04:44 In article <C7A0BD2F.53478%nicknaym@_remove_this_gmail.com.invalid>, Nick Naym <nicknaym@_remove_this_gmail.com.invalid> writes: >There seems to be two schools of thought: One, which takes the position that >leaving the machine on 24X7 shortens its life due to the wear & tear of >constant operation; the other, which believes that the wear & tear of >constant operation is minimal compared to the cumulative stress caused by >repeated, daily powering up and shutting down. I've always tended to agree >with the latter (ever notice that light bulbs tend to fail more often during >the electrical surge that occurs when you flip the switch, rather that while >they're running constantly in steady state?). > >The optimum likely is somewhere in between, depending on how often you turn >the machine on & off. Everyone has their own prescription, and mine is to >leave the machine running (with scheduled sleep periods every night) unless >I expect not to be using it for an entire day or more. Here we run a few hundred Macs and our experience so far is the following: - they last longer if they run continuously but - you have to put the drive to sleep when possible (with SCSI drives this wasn't an issue). To save energy you may put the monitor to sleep but again, only after a reasonable amount of time (eg. 15 minutes). Otherwise it will wear out earlier. Regards, Christoph Gartmann -- Max-Planck-Institut fuer Phone : +49-761-5108-464 Fax: -80464 Immunbiologie Postfach 1169 Internet: gartmann(a)immunbio dot mpg dot de D-79011 Freiburg, Germany http://www.immunbio.mpg.de/home/menue.html
From: thepixelfreak on 17 Feb 2010 10:59 On 2010-02-17 00:30:52 -0800, Nick Naym <nicknaym@_remove_this_gmail.com.invalid> said: > In article 2010021622124216807-not(a)dotcom, thepixelfreak at not(a)dot.com > wrote on 2/17/10 1:12 AM: > >> On 2010-02-16 19:03:15 -0800, Davoud <star(a)sky.net> said: >> >>> Nick Naym: >>>> There seems to be two schools of thought: One, which takes the position that >>>> leaving the machine on 24X7 shortens its life due to the wear & tear of >>>> constant operation; the other, which believes that the wear & tear of >>>> constant operation is minimal compared to the cumulative stress caused by >>>> repeated, daily powering up and shutting down. >>> >>> Bzzzzzt! The "stress" caused to a piece of electronic gear by turning >>> it on and off is pure myth. Study after study, some published and some >>> not, and tons of empirical evidence support that. >> >> Bzzzzzzzzt! I've worked for a Supercomputer manufacturer for the past >> 15 years and a total of 20 in the computer industry. I _routinely_ see >> plenty of real world evidence to the contrary. The machines we make >> have many thousands (8192 cores currently) and many many Terabytes of >> memory and Petabytes to Exabytes of attached rotational disk media. I >> am also a professional member of the ACM (Association of Computing >> Machinery) and have read numerous articles suggesting what you do. >> These articles DO NOT agree with my experience. >> >> Many of the studies (not all) you posit deal with statistically >> insignificant numbers of components. The very large component count >> machines we make (whether they be large NUMA SMP machines or large >> cluster supercomputers) exhibit statistically significant component >> failure after power has been removed for significant periods of time. >> These effect tend to manifest themselves more in rotational media than >> integrated circuits but nonetheless I see it in all components. >> >> That said, untoward effects of the power cycle on desktop computers >> with relatively very few components these days is likely to be rather >> minimal. > > That would beg the question as to how the effect of cycling the power on/off > -- however minimal that effect may be -- compares to the effect of leaving > the power on 24X7. Your _actual_ experience suggests that the comparison > still would favor letting the machines -- especially the HDs -- run full > time. My experience suggest that it can be a very bad thing to let a large system cool off for protracted periods of time after long periods of steady state runtime. My position is that small desktop computers don't have a large enough component part count (memory gates, processor transistors etc.) to make much of a difference whether or not they are left running. I have 5 computers here in my home office. 3 are running all the time and are never turned off (granted the iMac does go to sleep on the weekends). Two others are always off until some project requires their service and are promptly turned off when no longer needed. There are also two laptops elsewhere in the house that are in various states of power so they feel the brunt of the power cycle more than any others. All are fine. -- thepixelfreak
From: Davoud on 17 Feb 2010 11:21 thepixelfreak: > Bzzzzzzzzt! I've worked for a Supercomputer manufacturer for the past > 15 years and a total of 20 in the computer industry. I _routinely_ see > plenty of real world evidence to the contrary. Maybe you do, but your esoteric world has no application to the appliances -- and a home computer is just another appliance -- in the every-day world. Davoud -- I agree with almost everything that you have said and almost everything that you will say in your entire life. usenet *at* davidillig dawt cawm
From: Nick Naym on 17 Feb 2010 11:37
In article 2010021707590316807-not(a)dotcom, thepixelfreak at not(a)dot.com wrote on 2/17/10 10:59 AM: > On 2010-02-17 00:30:52 -0800, Nick Naym > <nicknaym@_remove_this_gmail.com.invalid> said: > >> In article 2010021622124216807-not(a)dotcom, thepixelfreak at not(a)dot.com >> wrote on 2/17/10 1:12 AM: >> >>> On 2010-02-16 19:03:15 -0800, Davoud <star(a)sky.net> said: >>> >>>> Nick Naym: >>>>> There seems to be two schools of thought: One, which takes the position >>>>> that >>>>> leaving the machine on 24X7 shortens its life due to the wear & tear of >>>>> constant operation; the other, which believes that the wear & tear of >>>>> constant operation is minimal compared to the cumulative stress caused by >>>>> repeated, daily powering up and shutting down. >>>> >>>> Bzzzzzt! The "stress" caused to a piece of electronic gear by turning >>>> it on and off is pure myth. Study after study, some published and some >>>> not, and tons of empirical evidence support that. >>> >>> Bzzzzzzzzt! I've worked for a Supercomputer manufacturer for the past >>> 15 years and a total of 20 in the computer industry. I _routinely_ see >>> plenty of real world evidence to the contrary. The machines we make >>> have many thousands (8192 cores currently) and many many Terabytes of >>> memory and Petabytes to Exabytes of attached rotational disk media. I >>> am also a professional member of the ACM (Association of Computing >>> Machinery) and have read numerous articles suggesting what you do. >>> These articles DO NOT agree with my experience. >>> >>> Many of the studies (not all) you posit deal with statistically >>> insignificant numbers of components. The very large component count >>> machines we make (whether they be large NUMA SMP machines or large >>> cluster supercomputers) exhibit statistically significant component >>> failure after power has been removed for significant periods of time. >>> These effect tend to manifest themselves more in rotational media than >>> integrated circuits but nonetheless I see it in all components. >>> >>> That said, untoward effects of the power cycle on desktop computers >>> with relatively very few components these days is likely to be rather >>> minimal. >> >> That would beg the question as to how the effect of cycling the power on/off >> -- however minimal that effect may be -- compares to the effect of leaving >> the power on 24X7. Your _actual_ experience suggests that the comparison >> still would favor letting the machines -- especially the HDs -- run full >> time. > > My experience suggest that it can be a very bad thing to let a large > system cool off for protracted periods of time after long periods of > steady state runtime. My position is that small desktop computers don't > have a large enough component part count (memory gates, processor > transistors etc.) to make much of a difference whether or not they are > left running. > > I have 5 computers here in my home office. 3 are running all the time > and are never turned off (granted the iMac does go to sleep on the > weekends). Two others are always off until some project requires their > service and are promptly turned off when no longer needed. There are > also two laptops elsewhere in the house that are in various states of > power so they feel the brunt of the power cycle more than any others. > All are fine. > I assume that the machines are of various ages, have been running the way you describe for different periods of time (long enough to reach a meaningful conclusion), and that neither their ages nor operating times appear to have any relevance to their "health." Correct? -- iMac (24", 2.8 GHz Intel Core 2 Duo, 2GB RAM, 320 GB HDD) � OS X (10.5.8) |