Page 1 of 2 12 LastLast
Results 1 to 15 of 19
  1. #1
    Integrity, whats it to u? Sin22's Avatar
    Join Date
    Jun 2002
    Posts
    884

    TIM Shootout: AS5, AS3, CM ShinEtsu, White Goop



    The four thermal pastes arranged as per the title. All fairly unassuming I'd say.

    I've purchased them all at one point or another through the year with the most recent one being AS5.

    Testing was done on the testbed introduced here. The testbed has its limitations, one of which is the fact that the wattage of the AthlonXP 2000+ in it cannot be accurately determined. Nor can the temperatures that are recorded be considered extremely accurate. They are definitely better than what the motherboard temp probes give, but still not accurate enuff to give a definite scientifically acceptable value. This shall be remedied in due time with the creation of this testbed which I am working on and asking the community for any help that can be offered.

    Now on to the testing. Utilised was the Thermalright SLK900A heatsink recently reviewed here. The reason for this was due to the fact that it utilised a mounting method which I felt would allow for repeatable mounts over a period of time. i.e. the force required for each mount would be similar if not exactly the same thus eliminating one variable in this testing. (The MCX462-V was an unknown though it very likely might have offered similar to identical mounts as well).

    Secondly, each thermal paste was applied to the middle of the CPU in the form of a small blob as depicted below.



    Each paste was tested with only a small blob and utilised the mounting pressure of the clip (which ought to be within 6~7lbsf as per AMD tech papers) to squeeze out the excess. This proved successful for all the pastes except for the CoolerMaster ShinEtsu one whereby the paste was too pastey and resulted in an improper imprint. This was overcome by placing three smaller blobs along the die.

    5 mounts were made for each paste, utilising the SLK900A with a Delta 92mm ASB fan. The averaged idle & load temp differentials with respect to ambient temperatures are given below.



    The accuracy of the probes must be mentioned once more. They are only accurate to +/-1degC so once more take with a pinch of salt the 0.xxx results.

    Anyhow, the results show that the AS5 paste definitely holds its own vs. the rest of the thermal pastes. AS3 and the normal white goop are on par with one another and Coolermaster ShinEtsu bringing up the rear by a fair amount.

    Now, many people and Artic Silver themselves claim that their pastes require a "burn-in" period for the paste to undergo the required phase-change and fill the microscopic pores up sufficiently. This may be the case, or it may just be propaganda. I decided at this point not to pursue that objective but instead to give a straight off test of the pastes. As such, with the results garnered, if what is said to be true, the pastes ought to exhibit better performance over the course of a few days "burn-in" thereby AS3 and AS5 taking the lead very nicely.

    As to the merits and demerits of each paste, its really very hard to quantify them.

    What I can say is that the Coolermaster ShinEtsu paste ought to be steered clear of. Its hard to apply, more like a putty than anything else and retails for S$15 last I checked. Furthermore, its performance was very lack lustre.

    White goop is easily found and found cheaply so if in doubt, just use it. No good or bad thing really, but if one degree drop in temps is essential to you, you can always look for AS3/5.

    AS3 should still be easily available in most places and its performance is still impressive, and furthermore, with the release of AS5, most retailers will be clearing stock and as such it ought to be going relatively cheaply.

    AS5 showed itself to have the best thermal conductivity of all the pastes. Its price is still relatively high due to its recent debut, but in time it will drop. However, I am unsure of the results of how it will hold up under extreme cold. A fair amount of the extreme cooling crowd is using AS Ceramique at the moment due to its good thermal properties at that low temperature. Only time will tell if AS5 will break into that group.


    Personal Opinion
    From here onwards is a small personal opinion. I have for a long time argued that the drastic differences that I have heard of on forums whereby a change in thermal paste resulted in 10degC drop in CPU temps to be outrageous and plain impossible. If there wasnt any thermal paste, yes the temps would be high, possibly in the region of 10deg difference, but as this review and previous experience has shown, normal white goop is more than sufficient for most people.

    Changing over to a more expensive paste will in some cases result in lower temps, but not as drastic as many people expect. Temper your expectations with what is known about physics. Read this article on thermal pastes and notice that normal water gave the best result of 0.41C/W. AS3 obtained 0.48C/W.

    C/W is basically the increase in degree Celcius for each watt of power. So for a 100W CPU, you'd expect a 48degC increase in temps over the temperature of the heatsource.

    Recognising this and the fact water cannot be used as a TIM substance over a long period of time, be ready to question the validity of any manufacturer claiming superb TIM performance. There is only a possible gain of 0.07C/W from here onwards, with the likely possibility of only having 0.04~5C/W actually being physically possible to obtain.

    Another good read would be this editorial by Cathar on the reality of temperatures and of TIM joints. More info can be found on the Procooling forums and the articles on Overclockers.com.

    Basically, once more my advice is, stick to what you can afford and what you truly need. Unless every single degree is essential to you, save your money and just buy normal white goop. If not, splurge a little and get that small temp drop. But remember, there are limits as to how much performance you can extract from TIMs.
    Aircooling History
    DragonOrb3 | V7+ | XDream | SLK800 | SLK900A | MCX462-V | SP94 | CJC 661CA | CNPS7000Cu | MCX478-V | XP90 | Hyper6 | Tuniq Tower | Akasa Evo30/Evo120 |

    Watercooling History
    Cyclone 5 | Jaron Esky (x2) | LRWW | LRCascade | Maze 3 & 4 | Turbine | Clover | MCW5002 | G4 Storm | Apogee | AquaXtreme MP05SP | Cooltrek Vostok | Name a performance block & I've had it....

    Phase Change History
    Evap Tower | Vapochill Classic | DIY Phase-Change | Vapochill PE | nVentive Prometia Mach II |

  2. #2
    :P josh's Avatar
    Join Date
    May 2003
    Location
    Terra, Sol III
    Posts
    7,403
    if memory serves, "white goop" costs around $2 per tube and AS3 costs $15....

  3. #3
    Full-Time Member RiStaR's Avatar
    Join Date
    Jan 2002
    Location
    Lion City
    Posts
    1,196

    Re: TIM Shootout: AS5, AS3, CM ShinEtsu, White Goop

    Originally posted by Sin22


    The four thermal pastes arranged as per the title. All fairly unassuming I'd say.

    I've purchased them all at one point or another through the year with the most recent one being AS5.

    Testing was done on the testbed introduced here. The testbed has its limitations, one of which is the fact that the wattage of the AthlonXP 2000+ in it cannot be accurately determined. Nor can the temperatures that are recorded be considered extremely accurate. They are definitely better than what the motherboard temp probes give, but still not accurate enuff to give a definite scientifically acceptable value. This shall be remedied in due time with the creation of this testbed which I am working on and asking the community for any help that can be offered.

    Now on to the testing. Utilised was the Thermalright SLK900A heatsink recently reviewed here. The reason for this was due to the fact that it utilised a mounting method which I felt would allow for repeatable mounts over a period of time. i.e. the force required for each mount would be similar if not exactly the same thus eliminating one variable in this testing. (The MCX462-V was an unknown though it very likely might have offered similar to identical mounts as well).

    Secondly, each thermal paste was applied to the middle of the CPU in the form of a small blob as depicted below.



    Each paste was tested with only a small blob and utilised the mounting pressure of the clip (which ought to be within 6~7lbsf as per AMD tech papers) to squeeze out the excess. This proved successful for all the pastes except for the CoolerMaster ShinEtsu one whereby the paste was too pastey and resulted in an improper imprint. This was overcome by placing three smaller blobs along the die.

    5 mounts were made for each paste, utilising the SLK900A with a Delta 92mm ASB fan. The averaged idle & load temp differentials with respect to ambient temperatures are given below.



    The accuracy of the probes must be mentioned once more. They are only accurate to +/-1degC so once more take with a pinch of salt the 0.xxx results.

    Anyhow, the results show that the AS5 paste definitely holds its own vs. the rest of the thermal pastes. AS3 and the normal white goop are on par with one another and Coolermaster ShinEtsu bringing up the rear by a fair amount.

    Now, many people and Artic Silver themselves claim that their pastes require a "burn-in" period for the paste to undergo the required phase-change and fill the microscopic pores up sufficiently. This may be the case, or it may just be propaganda. I decided at this point not to pursue that objective but instead to give a straight off test of the pastes. As such, with the results garnered, if what is said to be true, the pastes ought to exhibit better performance over the course of a few days "burn-in" thereby AS3 and AS5 taking the lead very nicely.

    As to the merits and demerits of each paste, its really very hard to quantify them.

    What I can say is that the Coolermaster ShinEtsu paste ought to be steered clear of. Its hard to apply, more like a putty than anything else and retails for S$15 last I checked. Furthermore, its performance was very lack lustre.

    White goop is easily found and found cheaply so if in doubt, just use it. No good or bad thing really, but if one degree drop in temps is essential to you, you can always look for AS3/5.

    AS3 should still be easily available in most places and its performance is still impressive, and furthermore, with the release of AS5, most retailers will be clearing stock and as such it ought to be going relatively cheaply.

    AS5 showed itself to have the best thermal conductivity of all the pastes. Its price is still relatively high due to its recent debut, but in time it will drop. However, I am unsure of the results of how it will hold up under extreme cold. A fair amount of the extreme cooling crowd is using AS Ceramique at the moment due to its good thermal properties at that low temperature. Only time will tell if AS5 will break into that group.


    Personal Opinion
    From here onwards is a small personal opinion. I have for a long time argued that the drastic differences that I have heard of on forums whereby a change in thermal paste resulted in 10degC drop in CPU temps to be outrageous and plain impossible. If there wasnt any thermal paste, yes the temps would be high, possibly in the region of 10deg difference, but as this review and previous experience has shown, normal white goop is more than sufficient for most people.

    Changing over to a more expensive paste will in some cases result in lower temps, but not as drastic as many people expect. Temper your expectations with what is known about physics. Read this article on thermal pastes and notice that normal water gave the best result of 0.41C/W. AS3 obtained 0.48C/W.

    C/W is basically the increase in degree Celcius for each watt of power. So for a 100W CPU, you'd expect a 48degC increase in temps over the temperature of the heatsource.

    Recognising this and the fact water cannot be used as a TIM substance over a long period of time, be ready to question the validity of any manufacturer claiming superb TIM performance. There is only a possible gain of 0.07C/W from here onwards, with the likely possibility of only having 0.04~5C/W actually being physically possible to obtain.

    Another good read would be this editorial by Cathar on the reality of temperatures and of TIM joints. More info can be found on the Procooling forums and the articles on Overclockers.com.

    Basically, once more my advice is, stick to what you can afford and what you truly need. Unless every single degree is essential to you, save your money and just buy normal white goop. If not, splurge a little and get that small temp drop. But remember, there are limits as to how much performance you can extract from TIMs.
    Did you test them below freezing point? Or rather what's their freezing point?
    :p
    P4 2.4C @ 3.?
    P4 2.6C @ 3.64 SL6WS Philli
    Vapochill PE Ti
    Asus P4P800 Deluxe (Vdimm modded)
    Twinmos CH-5 512MB x 2
    Corsair XMS3500C2512 v1.1 x 2
    Hercules 9800 Pro 128MB
    Waterchill(ed) NB and GPU
    VR-Zone Extreme-cooled Member
    Join the Fight with your CPU cycles
    http://d2ol.sengent.com/memberServic...le.jsp?id=1646
    Photography

  4. #4
    Integrity, whats it to u? Sin22's Avatar
    Join Date
    Jun 2002
    Posts
    884
    Pass me your Vapo and I'll do the test for you

    Next thing I can think of, chucking them in my freezer and see which still has its consistency after a few hours in there. That sound acceptable ?
    Aircooling History
    DragonOrb3 | V7+ | XDream | SLK800 | SLK900A | MCX462-V | SP94 | CJC 661CA | CNPS7000Cu | MCX478-V | XP90 | Hyper6 | Tuniq Tower | Akasa Evo30/Evo120 |

    Watercooling History
    Cyclone 5 | Jaron Esky (x2) | LRWW | LRCascade | Maze 3 & 4 | Turbine | Clover | MCW5002 | G4 Storm | Apogee | AquaXtreme MP05SP | Cooltrek Vostok | Name a performance block & I've had it....

    Phase Change History
    Evap Tower | Vapochill Classic | DIY Phase-Change | Vapochill PE | nVentive Prometia Mach II |

  5. #5
    Drug Dealer cccp's Avatar
    Join Date
    Jun 2003
    Location
    Singapore
    Posts
    9,541
    as usual, this thread will go sticky for a while. Nice work Sin22 :thumb1:
    One of the most addictive web-based game

    www.goldentowns.com

    Playing this web game which claims that the in-game gold is backed by real gold. Quite a vibrant community out there. Seems like the whole bitcoin thingie is spreading to the online gaming world. They are coming up with their own crypto-currency as well. Look for me, i am cccp1979lcw inside!

  6. #6
    Integrity, whats it to u? Sin22's Avatar
    Join Date
    Jun 2002
    Posts
    884
    This is for you RiStar

    I chucked all four pastes onto a piece of scotch taped cardboard into my freezer and left it over night.



    A before pic.

    Clarification:
    Top left is white goop.

    Top right is AS3.

    Bottom left is CM ShinEtsu.

    Bottom right is AS5.

    Just then i took it out to have a look.



    An after pic.

    And I touched each one.

    Basically, the white goop still had its basic consistency though it felt slightly thicker.

    AS3 exhibited definitely a more putty like sensation in comparison to when its first applied.

    AS5 also had a slightly more putty like feel to it, but to not the same degree as AS3.

    CM ShinEtsu well. I really dont know what happened. I think someone went into my freezer last night and knocked the cardboard piece off and the ShinEtsu got smeared off. The small problem is, that when I looked around the freezer, I found the ShinEtsu and it appeared to have dribbled onto a plastic bag beneath it. Well basically it looked as though it had become a complete fluid and just flowed off the cardboard piece. I really dont know what to make of it, so, I'll chuck another cardboard piece with just ShinEtsu on it into the freezer later.

    Hope that helps a little, though I dont know how. At least it was sorta fun.
    Last edited by Sin22; Nov 22nd, 03 at 01:52 PM.
    Aircooling History
    DragonOrb3 | V7+ | XDream | SLK800 | SLK900A | MCX462-V | SP94 | CJC 661CA | CNPS7000Cu | MCX478-V | XP90 | Hyper6 | Tuniq Tower | Akasa Evo30/Evo120 |

    Watercooling History
    Cyclone 5 | Jaron Esky (x2) | LRWW | LRCascade | Maze 3 & 4 | Turbine | Clover | MCW5002 | G4 Storm | Apogee | AquaXtreme MP05SP | Cooltrek Vostok | Name a performance block & I've had it....

    Phase Change History
    Evap Tower | Vapochill Classic | DIY Phase-Change | Vapochill PE | nVentive Prometia Mach II |

  7. #7
    RadeonSE
    Guest
    well , seems the white goop is the best for below ZERO . my friend used it and it was excellent in below-ZERO tests.

    seems 2$ is better for below-ZERO users .

  8. #8
    Full-Time Member RiStaR's Avatar
    Join Date
    Jan 2002
    Location
    Lion City
    Posts
    1,196
    Hey Sin,

    I recently ripped apart the vapo... the shinetsu stuff i had on thickened and solidified to about 1mm think with the middle bit basically liquified and boiled--i'm serious there were burnt marks on the heatspreader lol... took alot of scratching to get the solidified gunk off... turpentine couldn't get rid of it...

    Back to using plain old "Unick" stuff till i find something better... as3 doesn't work well below 0 as well :
    P4 2.4C @ 3.?
    P4 2.6C @ 3.64 SL6WS Philli
    Vapochill PE Ti
    Asus P4P800 Deluxe (Vdimm modded)
    Twinmos CH-5 512MB x 2
    Corsair XMS3500C2512 v1.1 x 2
    Hercules 9800 Pro 128MB
    Waterchill(ed) NB and GPU
    VR-Zone Extreme-cooled Member
    Join the Fight with your CPU cycles
    http://d2ol.sengent.com/memberServic...le.jsp?id=1646
    Photography

  9. #9
    Buckminsterfullerene tenankers's Avatar
    Join Date
    Jun 2003
    Location
    Mars
    Posts
    3,744
    geez, i have not finished my artic silver 1 since i bought it years ago.
    it has gone onto at least 50 cpu core or more ....
    still have enuff for 10 -20 cpu.
    think when i finshed using my tube, i can try artic silver 8 liao
    Desktop CPU
    Intel 486DX 33mhz -> AMD 5x86 133mhz -> AMD K6 233mhz -> AMD K6 III 3D now 450mhz -> AMD Duron 700mhz -> AMD Athlon XP 1800+
    Modded CPU case
    IKEA breadbox I , IKEA breadbox II , Guitar PC, Apple G4 PC
    Laptop
    IBM T21 PIII 800mhz->ECS Desknote VIA 1ghz ->FOSA 1Ghz -> Compaq Tablet TC1000 -> Compaq Presario 1532AP 2.4Ghz -> FOSA Centrino -> Dell Inspiron 9300

  10. #10
    chad_sexington
    Guest
    Im using as5 and it dropped my temps 4 or 5 degrees over the pad that came with the hsf. Also, the arctic silver site says that as5 can go to -50 C

  11. #11
    Drug Dealer cccp's Avatar
    Join Date
    Jun 2003
    Location
    Singapore
    Posts
    9,541
    unstuck
    One of the most addictive web-based game

    www.goldentowns.com

    Playing this web game which claims that the in-game gold is backed by real gold. Quite a vibrant community out there. Seems like the whole bitcoin thingie is spreading to the online gaming world. They are coming up with their own crypto-currency as well. Look for me, i am cccp1979lcw inside!

  12. #12
    :P josh's Avatar
    Join Date
    May 2003
    Location
    Terra, Sol III
    Posts
    7,403
    well, belive it or not, theres really is a "burn" in period..


    the day i applied AS3


    today.

    btw can anyone explain the increase in the Sisoftware Sandra score?

  13. #13
    Integrity, whats it to u? Sin22's Avatar
    Join Date
    Jun 2002
    Posts
    884
    I'd rather choose to not believe it considering how notoriously unreliable & inaccurate the Abit temp sensors are added to the fact that they calcuate the final temp values using a mathematical formula based on voltage supplied instead of any form of accurate temperature measurement.

    From the Horses' Mouth

    As to the variance in scores, it could well be that in the previous screenie, there were some background activities occuring that degraded the performance somewhat. Its always a possibility with XP.
    Aircooling History
    DragonOrb3 | V7+ | XDream | SLK800 | SLK900A | MCX462-V | SP94 | CJC 661CA | CNPS7000Cu | MCX478-V | XP90 | Hyper6 | Tuniq Tower | Akasa Evo30/Evo120 |

    Watercooling History
    Cyclone 5 | Jaron Esky (x2) | LRWW | LRCascade | Maze 3 & 4 | Turbine | Clover | MCW5002 | G4 Storm | Apogee | AquaXtreme MP05SP | Cooltrek Vostok | Name a performance block & I've had it....

    Phase Change History
    Evap Tower | Vapochill Classic | DIY Phase-Change | Vapochill PE | nVentive Prometia Mach II |

  14. #14
    :P josh's Avatar
    Join Date
    May 2003
    Location
    Terra, Sol III
    Posts
    7,403
    oops, forgot the date... the day i applied as3, was sat, after CNY...

  15. #15
    :P josh's Avatar
    Join Date
    May 2003
    Location
    Terra, Sol III
    Posts
    7,403
    Originally posted by Sin22
    I'd rather choose to not believe it considering how notoriously unreliable & inaccurate the Abit temp sensors are added to the fact that they calcuate the final temp values using a mathematical formula based on voltage supplied instead of any form of accurate temperature measurement.

    From the Horses' Mouth

    As to the variance in scores, it could well be that in the previous screenie, there were some background activities occuring that degraded the performance somewhat. Its always a possibility with XP.
    interesting..

    hmm any idea where to get a thermal probe, and how much will it cost?

Page 1 of 2 12 LastLast