4 Replies Latest reply on Dec 30, 2009 6:26 AM by aliensporebomb

    Will 100% CPU Utilization for 12 straight hours per 24 decrease lifetime?


      I am trying to do estiamate the lifetime of a CPU that will be used for 12 hours continually at 98% - 100% (all cores) per 24 hours, 7 days a week.  It will be cooled to within standard specifications, but I expect that such a long max. utilization will result in lower lifetime due to interspatial heating (between cores) and a general degredation due to constant maximum useage.  Is there any data on this, or does anyone have any practical experience?  I know that the standard RoT is 7 years, but my 'gut' is saying about 3 - 4 for such a setup, if that.



        • 1. Re: Will 100% CPU Utilization for 12 straight hours per 24 decrease lifetime?

          Just a thought:  I've had numerous computers, both Intel driven and their competitors running 100% cpu, 24 hours a day for about the

          last eleven years as I've been using the Seti at Home project on my machines. 


          In terms of Intel, I've got a 2.2 ghz Core2duo T7500 that's been running 24/7 in a laptop for over two years now with the only item needing

          repair is cooling fans inside the machine (cheap Chinese muffin fans sourced as cheaply as possible I suspect).


          I've never once ever had a CPU failure due to this type of usage.  It's the peripheral components that may start wearing out such as

          bearings in CPU fans and the like.  Just make sure you know that running this way will cause the CPUs to generate a bit of heat and

          as such you'll get a lot of dust in the innards so make sure you clean the machine out with compressed air if you notice a lot of dust



          But 12 hours a day on and 12 hours off?  I don't see a problem there.  Now if you're overclocking all bets are off.

          • 2. Re: Will 100% CPU Utilization for 12 straight hours per 24 decrease lifetime?

            SETI uses 100% of both cores?  I did not think it was that intensive, though I have not seen it running on a system for a few years now.


            That is encouraging, but hoperfully someone will have a desktop example.  Laptop CPUs are much more efficient than desktop CPUs for energy use and therefore run at lower temps.  My largest concern is that the chips in some of these systems will not have the budget to be replaced if they do start to fail, and while it will be within the warranty of the processor, I would not be surprised to see a chip inspection void the warranty.  I may be paranoid, but I remember the good old days when processors came with 7 year warrantie, and I have heard rumors that current three year may be replaced with only a two pretty soon.  That indicates a much higher failure rate than single cores had, and I feel fairly secure in assuming is it the increased heat that they produce.  I was surprised to find out how hot newer quad cores run.


            Thanks for your reply!

            • 3. Re: Will 100% CPU Utilization for 12 straight hours per 24 decrease lifetime?



              The Intel cooling solutions and ratings are based on typical usage which isn't having all cores ramped up to 100% usage for extended periods of time.  To cope for this modern CPUs will become hot and will then start to slow themselves down (thermal throttling).


              This will mean you may want to use a better cooling solution than the standard Intel heatsinks provided even then the chip may overheight causing thermal throttling as the heat often can't transfer quick enough away from the chip.


              The power regulation circuit on the motherboard will also be under strain as well as other components as already explained.  For the sort of usage of you require a computer designed for the job such as a server, these cost a lot more simply because everything is over engineered to allow them to used and abused.


              You get what you pay for.





              • 4. Re: Will 100% CPU Utilization for 12 straight hours per 24 decrease lifetime?

                One thought I would mention is for my systems that are designed for maximum uptime and heavy CPU utilization all have more fans than

                necessary (one system actually has 9 fans so each part of the machine is cooled so the CPU is running at comfortable temperatures when

                under load.  The other machine, the overbuilt one has 4 powerful fans internally but I actually have an ATX size case butted up against the

                side of the open case with a modified air purifier with hepa filtering that pushes large amounts of air into the machine and you can visibly

                see the temperature graph drop abruptly when that is in use.    Phil is right, you get what you pay for.  Good luck!