I fail to understand some of the advice given here on the subject of TIM and replacement intervals. Heat...and especially prolonged intervals of continous heat like say a cpu.....drys everything including the thermal paste...especially the thermal paste@! Just because one hasn't changed the paste in 6 years..or more...doesn't mean that it was a good thing for their processors or the operating efficiency of said silicon and all the surrounding components that suffered efficiency loss due to the thermal envelope inside their machines ever increasing over time. The hotter silicon gets the less efficiently it can conduct electrons thus reducing overall efficiency and sometimes a loss of efficiency that is heavily detrimental to the performance of part or all of the components in ones machine, never mind the reduction in mtbf...mean time before failure...which decreases as operating temperatures in the operating environment reach higher and higher levels and the components that function in those environments suffer from the increase in heat load.
As a guy with a degree in technology and a total of two decades of working on electronics and pc's...read..electronics and networking...I fully understand that "HEAT" is the destroyer of all electronics and the transferrence and dissipation of heat...as much as you can...is a very desirable trait one wants a computer system to be extremely capable and efficient at performing.
Since ALL processing performed in a computer is performed by one type of processing unit or another..ie cpu or gpu and other controllers that exist in the Northbridge or Southbridge or by any type of chipset all produce large quantities of heat and one should always endeavor to remove as much of that heat as is efficiently possible! GOOD ENOUGH... ISN'T... REALLY!
With the CPU being the "brains" of the pc platform...using a quality thermal paste, in the correct amounts, per spec design...and REGULARLY changing the paste...does nothing but ENHANCE performance over the long haul. The colder you can keep your cpu, gpu, N or S Bridges... or chipsets that generate large amounts of heat..the absolute better the entire system performs...the longer the usuable life the components have...and the longer it takes to reach unacceptable performance and degradation of the components leading to failure, bsods, unexplained freezes and drop outs and so on.
At a minimum...I would (and do) change the paste on my machines at least once a year...my laptop...every 6-8 months...and yes...I do see a difference in performance and idle temps as well as "under load condition" temps. The most gain realized is always under load conditions...I usually drop 2-4 C with new paste...in and of itself that doesn't seem like much of a drop...but when you add that difference back into the equation when considering longetivity of your components....that can lead to a considerable lengthening of the silicons life expectancy. In this economy...I'm all for enhancing my chances of a better ROI on my choice of computers and the ensuing maintenance intervals that are responsible for that enhancing!!!
The cost of materials is insignificant vs the cost of a lost cpu...chipset...gpu...and associated hardware like the socket for the cpu...the cpu cooler fan or equipment and especially the bus traces that interconnect the cpu to the peripheals...when a cpu or gpu fails...and with added heat loads...they do regularly fail...these failures almost always contain enhanced risk to the components directly connected via those bus paths...ie the ram slots and ram...the pcie slots the expansion cards are in...the controllers that handle jobs for usb...raid...the onboard nics...all are easy to blow with failures in the circuits due to heat build up that could be disastrous and expensive.
These are "Worst case scenarios" agreed...but they are also common as failure points for the pc industry and the designs being utilized today...high frequency processors...with very little headroom for heat buildup outside the thermal profiles of these high speed components per manufacturer specs.
As a once semi-profecient backyard auto mechanic...grease is what helps manage heat in the rearend that is always turning if moving...the wheel bearings as they are always turning if moving...the drive shaft u-joints as they are always turning if moving...and as a mechanic...if people would repack the bearings and u-joints once a year and replace the rearend grease at intervals of 3 years or so...then the risk of failure goes heavily DOWN...just like the thermal paste in your machines...be they laptops or desktops or htpcs...heat management is critical for longevity and efficiently guarding against performance degradation.
[FONT=arial black] I would recommend waiting on gpu's through the warranty periods as removal of outer covers and heatsinks often times voids the warranty...it's getting better as time passes and manufacturers are recognizing that overclocking and mods are being utilized more often and are providing for simple maintenance such as tim replacement and are beginning to allow the end-user to perform some basic preventive maintenance on these types of components without voiding the warranty...read carefully all warranty provisions and beware...if in doubt email or contact the manufacturer of your component and inquire as to if what you are attempting will void your warranty PRIOR to any work being performed.[/FONT]
Why oh Why would you neglect to perform a very simple "in most cases" maintenance operation that is both cheap and extremely effective insurance against catastrophic loss or failure?
For me...it's a no-brainer...and a simple task...an individual with basic hand tools and a small modicum of ability...can watch a video on any type of cpu socket or gpu heatsink removal and tim replacement as well as recommendations for application of the tim and perform what is at the most a fifteen minute operation in a vast majority of pc layouts.
You should change (or have changed) your paste as often as whatever makes you comfortable...but when dealing with high speed electronic components, I never believe in pushing my luck or taking a unnecessary risk due to heat build up that can be easily neutralized via a simple 15 min task in most cases.
As I said in the beginning of this post...I do not understand why one would risk thousands of dollars and all their personal data on their computers by failing to perform simple "preventive" maintenance. That's a risk that is easily avoided and should be part of any pc owners efforts to maintain their computers. Educate oneself a bit in the art of electronics and do a bit of reading on the subject of heat vs electroncis and how engineers design the TP...the thermal envelope and operating designs of a circuit or component or group of components...and the value of keeping all that silicone as cool as possible and operating at peak efficiency for the money spent on your electric bills.. thereby improving efficiency, enhancing roi along with increasing the longevity of the components and investment in their computers.
Need is a subjective descriptor...I "need" to adhere to a maintenance schedule for my machines that includes a healthy dose of common sense, an educated interest in the components thermal envelope that resides in my computers and since I don't have the funds to burn...protecting my hard earned dollar invested in said machines.
A year is a good time frame if heavy computing...a bit longer if surfing and email is your average online experience...less if heavily into gaming, encoding, or cad...or any other processor intensive activities are the norm for you.
Risk is all around....why invite it in?
I know others don't feel this way...but from one who has seen a lot of machines in educational environs...governmental environs.. and industrial environs... and many a home....I feel comfortable with my time frame as a least risk option for me...Not everyone will see it the same way but I suspect there are a lot of techies out there that practice much the same time tables as I do...JIC...just in case. It also has the caveat of being one more risk reduced as a point of failure further reducing expenses to ones pocket book.
Respectfully,
tekman42