- Joined
- Dec 13, 2002
- Location
- Tucson, AZ
Hello everyone,
I'm going to start off this show with a bang - be warned, for this is a very long read:
First, I was recently informed that an article was published on www.overclockers.com regarding the comparison of thermal compounds. In this short, one-page review, several thermal compounds were reportedly tested. However, the results didn't seem to be congruent with any other professional reviewers findings (or even user reports for that matter), anywhere else in the world. I found this to be quite odd, considering that www.overclockers.com was generally considered as a very reputable site amongst the overclocking enthusiast community.
Upon further investigating the review's test results, and the methods used to gain those results, I subsequently found them lacking both consistency and thoroughness. Now normally, I wouldn't be inclined to personally write the author of any review - even a bad review, such as this one obviously was. However, this review was simply too insulting to anyone who knew anything about thermal compounds.
So I emailed the author directly, commenting politely on the obvious mistakes made in the article which really should be corrected (on an immediate basis), and which really wouldn't have been too difficult to correct - with no tangible "loss of face" suffered. Here's the email I sent:
"Hello Joe,
Recently I learned that overclockers.com had made a thermal compound review, posted here:
http://www.overclockers.com/articles660/
Now please compare yours with the review that I conducted for Inside Project:
http://www.insideproject.com/showreview.cfm?reviewid=81
While I've always enjoyed your site, I unfortunately have distinct issues with your review. First, you should already know that you must allow the compounds to settle and cure for at least 3 days while on medium to high heat, and then re-take your temperatures again. The additional fact that there was no control mechanism in place for verifying temperatures also could lend skepticism towards your reported results.
There are also several versions of Shin Etsu, with the G-751 being the best performer. Your review unfortunately didn't clearly specify what version was tested, yet pointed to an older version (G-749) via a hyper-link.
Frankly put, Shin Etsu G-751 blew away everything else in its class in our tests, and in similar tests conducted by www.meetthegeeks.org. Shin Etsu G-751 is available at www.chillblast.com, and a few other places around the world.
As a recommendation, perhaps you should've used the most recent version of Shin Etsu, and conducted further tests after a period of at least 72 hours, and you may have found the same results that we did.
In summary, I found this review pretty short-sighted, and potentially misleading, unfortunately.
I am VERY surprised to have seen this actually posted on overclockers.com... You guys should know better - you have a very good reputation amongst the community.
My friendly .02,
Wizard
[email protected] "
I however, received no response.
A second "follow-up" article appeared shortly thereafter, which apparently (and vainly) attempted to support the first article's findings by using a potentially incomplete and basically non-applicable mathematical formula from over 6 years ago, out of context, with substituted and relative W/mK values for compounds which weren't obtained by any uniform testing method.
Also included in the review was the statement:
"The "ultimate" thermal grease will reduce CPU core temps by about 1 C."
Yet later in the same article, this was self-contradicted by the following statement:
"Don't expect any quality grease to cost you more than 1-2 C at 100 watts radiated heat."
I have since written the author in question yet again, asking him to please either remove both of these grossly innaccurate articles from publication, or to at least revisit his testing methodologies and then see the resultant differences for himself.
Here is that email:
Hello once again Joe,
Normally I don't bother writing people in regards to published articles. Yet I feel compelled to again inform you now of the continued discrediting of your site from the publishing of the follow-up article listed below:
http://overclockers.com/articles662/
The original article I wrote you about is still posted here:
http://overclockers.com/articles660/
There's a distinct difference between figures that exist solely on paper, and reality - where the proverbial rubber actually hits the road.
What is particularly disturbing is this quote:
"The "ultimate" thermal grease will reduce CPU core temps by about 1 C."
This is clearly a gross misstatement that is in fact self-contradicted later in the same article(!) by this statement;
"don't expect any quality grease to cost you more than 1-2 C at 100 watts radiated heat."
The main point I'm making here is that your 'conclusions' contradict every properly conducted scientific measurement process (and the results thereof) that exist on the subject of thermal compounds, in addition to the findings of practically every other professional overclocker, reviewer, and user all over the world.
To address this more in detail, the scientific paper quoted here;
http://www.electronics-cooling.com/Resources/EC_Articles/SEP96/sep96_01.htm
is an (older), more generalized, theoretical paper on the overall benefits of using thermal interface materials as opposed to not using any. It was published in September '96, over 6 years ago, using socket 5 CPU's and what can easily now be called inferior thermal interface materials. Technology has changed greatly since then.
This article should not be used out of context with substituted (and suspect) values gathered from separate outside sources. The formula listed there may indeed be a decent (if yet theoretical) formula, but it isn't encompassing enough to be considered the "end-all, be-all" of how well thermal compounds will actually perform in general. In fact, it doesn't appear that the author ever intended for this article to be used in that specific, comparative regard. So perhaps it should not be misinterpreted and used thusly. In all honestly, I doubt a comprehensive "formula" truly exists, even now - or that one can actually exist, for that matter. There are always more variables in reality than there are on paper, much like friction is to the perpetual motion machine.
For example, if we were to examine chipset specs of say, different video cards, we would find that certain cards with higher actual hardware specs simply don't perform as well as certain other cards with lower specs. The potential reason(s)? Drivers and differing test environments and methodologies, obviously. But you can't factor these into any math equation. Likewise, so should the same attitude be taken when approaching thermal compounds. You cannot rely upon what is an older, potentially incomplete formula to justify what are results generated by inadequate testing methods, sorry.
Scientifically controlled tests should normally include using the same HSF, temp monitoring equipment and processes for gaining properly comparative results. So your statements regarding these having an affect on temperatures, while decidedly are both true and valid, are also somewhat of a moot point - I can't help but feel that they are included as a "make-up" statement for the lack of any extended articulation regarding this subject in the last article (article660). In fact, myself and other professionals feel that this entire article (article662) is pretty much an attempt at self-justification for the last article (article600) - specifically the testing results posted.
Here's a list of some missing and/or erroneous information found in both of these two articles:
#1. You either didn't use Shin Etsu G-751 (their latest version) in your tests - and/or if you did, you initially linked to the wrong product.
#2. The follow-up review does actually specify G-751, but whether that product was actually tested in the first article (instead of G-749) remains in question.
#3. The reported W/mK figures from each thermal compound were gathered from different sources, and should be considered "representative values" only - the actual values of each are dependent upon the specific testing methods used. Being that these values were derived from different sources, this subsequently invalidates their potential usage. In other words, using these values is unfortunately baseless for any sort of measurement comparison.
#4. You unfortunately didn't mention the ambient temperature of the testing room. The consistency of this factor is critical for any kind of air-cooled testing, as well as the all-important factors of consistency of HSF/CPU and case airflow, among other things.
#5. You only measured initial temperatures within an 8 hour period, unfortunately and completely missing the boat on settling/curing time factors for each type of compound.
#6. Instead of either removing the first article from publication, or posting a follow-up admission that you were initially incorrect or perhaps lacking information, you followed-up with a second article which actually attempts to justify your firsts' findings, hiding behind an older, potentially incomplete and basically non-applicable mathematical formula, using relative, substituted values which weren't determined by any uniform testing method.
Your statement,
"I repeat: The "ultimate" thermal grease will reduce CPU core temps by about 1 C."
contradicts the reported results of anyone who has ever used different thermal compounds, anywhere else in the world, and again is even self-contradicted within the same article.
It is from misinformative articles like these that people do actually carry the misperception that all thermal compounds are basically the same - when proper testing clearly shows that they are not.
Therefore, I would urge that you either remove these articles in question from publication immediately, and/or revisit your testing methodologies and then see the differing results for yourself. As it stands now however, these articles do seriously discredit yourself and www.overclockers.com (despite the valiant, yet vain effort of justification via math equations). Both articles misinform readers of the potential benefits of using a higher quality thermal compound.
I do hope that you will give my recommendations your serious and immediate consideration.
Wizard
www.insideproject.com
www.tweakers.com.au
www.chillblast.com
I (again) have received no response as of yet, unfortunately.
It is therefore from careful deliberation and consideration that I have made the following official statements on several technical forum websites: Thermal compound reviews at www.overclockers.com should be disregarded as being misinformative, and completely inaccurate.
I do hope that the author reads this, and realizes that amongst the professional community, asking for more information or even admitting ignorance on a particular subject isn't a crime or weakness. It is actually considered a strength, and welcomed instead - for we're all learning. <|;^)>
I'm going to start off this show with a bang - be warned, for this is a very long read:
First, I was recently informed that an article was published on www.overclockers.com regarding the comparison of thermal compounds. In this short, one-page review, several thermal compounds were reportedly tested. However, the results didn't seem to be congruent with any other professional reviewers findings (or even user reports for that matter), anywhere else in the world. I found this to be quite odd, considering that www.overclockers.com was generally considered as a very reputable site amongst the overclocking enthusiast community.
Upon further investigating the review's test results, and the methods used to gain those results, I subsequently found them lacking both consistency and thoroughness. Now normally, I wouldn't be inclined to personally write the author of any review - even a bad review, such as this one obviously was. However, this review was simply too insulting to anyone who knew anything about thermal compounds.
So I emailed the author directly, commenting politely on the obvious mistakes made in the article which really should be corrected (on an immediate basis), and which really wouldn't have been too difficult to correct - with no tangible "loss of face" suffered. Here's the email I sent:
"Hello Joe,
Recently I learned that overclockers.com had made a thermal compound review, posted here:
http://www.overclockers.com/articles660/
Now please compare yours with the review that I conducted for Inside Project:
http://www.insideproject.com/showreview.cfm?reviewid=81
While I've always enjoyed your site, I unfortunately have distinct issues with your review. First, you should already know that you must allow the compounds to settle and cure for at least 3 days while on medium to high heat, and then re-take your temperatures again. The additional fact that there was no control mechanism in place for verifying temperatures also could lend skepticism towards your reported results.
There are also several versions of Shin Etsu, with the G-751 being the best performer. Your review unfortunately didn't clearly specify what version was tested, yet pointed to an older version (G-749) via a hyper-link.
Frankly put, Shin Etsu G-751 blew away everything else in its class in our tests, and in similar tests conducted by www.meetthegeeks.org. Shin Etsu G-751 is available at www.chillblast.com, and a few other places around the world.
As a recommendation, perhaps you should've used the most recent version of Shin Etsu, and conducted further tests after a period of at least 72 hours, and you may have found the same results that we did.
In summary, I found this review pretty short-sighted, and potentially misleading, unfortunately.
I am VERY surprised to have seen this actually posted on overclockers.com... You guys should know better - you have a very good reputation amongst the community.
My friendly .02,
Wizard
[email protected] "
I however, received no response.
A second "follow-up" article appeared shortly thereafter, which apparently (and vainly) attempted to support the first article's findings by using a potentially incomplete and basically non-applicable mathematical formula from over 6 years ago, out of context, with substituted and relative W/mK values for compounds which weren't obtained by any uniform testing method.
Also included in the review was the statement:
"The "ultimate" thermal grease will reduce CPU core temps by about 1 C."
Yet later in the same article, this was self-contradicted by the following statement:
"Don't expect any quality grease to cost you more than 1-2 C at 100 watts radiated heat."
I have since written the author in question yet again, asking him to please either remove both of these grossly innaccurate articles from publication, or to at least revisit his testing methodologies and then see the resultant differences for himself.
Here is that email:
Hello once again Joe,
Normally I don't bother writing people in regards to published articles. Yet I feel compelled to again inform you now of the continued discrediting of your site from the publishing of the follow-up article listed below:
http://overclockers.com/articles662/
The original article I wrote you about is still posted here:
http://overclockers.com/articles660/
There's a distinct difference between figures that exist solely on paper, and reality - where the proverbial rubber actually hits the road.
What is particularly disturbing is this quote:
"The "ultimate" thermal grease will reduce CPU core temps by about 1 C."
This is clearly a gross misstatement that is in fact self-contradicted later in the same article(!) by this statement;
"don't expect any quality grease to cost you more than 1-2 C at 100 watts radiated heat."
The main point I'm making here is that your 'conclusions' contradict every properly conducted scientific measurement process (and the results thereof) that exist on the subject of thermal compounds, in addition to the findings of practically every other professional overclocker, reviewer, and user all over the world.
To address this more in detail, the scientific paper quoted here;
http://www.electronics-cooling.com/Resources/EC_Articles/SEP96/sep96_01.htm
is an (older), more generalized, theoretical paper on the overall benefits of using thermal interface materials as opposed to not using any. It was published in September '96, over 6 years ago, using socket 5 CPU's and what can easily now be called inferior thermal interface materials. Technology has changed greatly since then.
This article should not be used out of context with substituted (and suspect) values gathered from separate outside sources. The formula listed there may indeed be a decent (if yet theoretical) formula, but it isn't encompassing enough to be considered the "end-all, be-all" of how well thermal compounds will actually perform in general. In fact, it doesn't appear that the author ever intended for this article to be used in that specific, comparative regard. So perhaps it should not be misinterpreted and used thusly. In all honestly, I doubt a comprehensive "formula" truly exists, even now - or that one can actually exist, for that matter. There are always more variables in reality than there are on paper, much like friction is to the perpetual motion machine.
For example, if we were to examine chipset specs of say, different video cards, we would find that certain cards with higher actual hardware specs simply don't perform as well as certain other cards with lower specs. The potential reason(s)? Drivers and differing test environments and methodologies, obviously. But you can't factor these into any math equation. Likewise, so should the same attitude be taken when approaching thermal compounds. You cannot rely upon what is an older, potentially incomplete formula to justify what are results generated by inadequate testing methods, sorry.
Scientifically controlled tests should normally include using the same HSF, temp monitoring equipment and processes for gaining properly comparative results. So your statements regarding these having an affect on temperatures, while decidedly are both true and valid, are also somewhat of a moot point - I can't help but feel that they are included as a "make-up" statement for the lack of any extended articulation regarding this subject in the last article (article660). In fact, myself and other professionals feel that this entire article (article662) is pretty much an attempt at self-justification for the last article (article600) - specifically the testing results posted.
Here's a list of some missing and/or erroneous information found in both of these two articles:
#1. You either didn't use Shin Etsu G-751 (their latest version) in your tests - and/or if you did, you initially linked to the wrong product.
#2. The follow-up review does actually specify G-751, but whether that product was actually tested in the first article (instead of G-749) remains in question.
#3. The reported W/mK figures from each thermal compound were gathered from different sources, and should be considered "representative values" only - the actual values of each are dependent upon the specific testing methods used. Being that these values were derived from different sources, this subsequently invalidates their potential usage. In other words, using these values is unfortunately baseless for any sort of measurement comparison.
#4. You unfortunately didn't mention the ambient temperature of the testing room. The consistency of this factor is critical for any kind of air-cooled testing, as well as the all-important factors of consistency of HSF/CPU and case airflow, among other things.
#5. You only measured initial temperatures within an 8 hour period, unfortunately and completely missing the boat on settling/curing time factors for each type of compound.
#6. Instead of either removing the first article from publication, or posting a follow-up admission that you were initially incorrect or perhaps lacking information, you followed-up with a second article which actually attempts to justify your firsts' findings, hiding behind an older, potentially incomplete and basically non-applicable mathematical formula, using relative, substituted values which weren't determined by any uniform testing method.
Your statement,
"I repeat: The "ultimate" thermal grease will reduce CPU core temps by about 1 C."
contradicts the reported results of anyone who has ever used different thermal compounds, anywhere else in the world, and again is even self-contradicted within the same article.
It is from misinformative articles like these that people do actually carry the misperception that all thermal compounds are basically the same - when proper testing clearly shows that they are not.
Therefore, I would urge that you either remove these articles in question from publication immediately, and/or revisit your testing methodologies and then see the differing results for yourself. As it stands now however, these articles do seriously discredit yourself and www.overclockers.com (despite the valiant, yet vain effort of justification via math equations). Both articles misinform readers of the potential benefits of using a higher quality thermal compound.
I do hope that you will give my recommendations your serious and immediate consideration.
Wizard
www.insideproject.com
www.tweakers.com.au
www.chillblast.com
I (again) have received no response as of yet, unfortunately.
It is therefore from careful deliberation and consideration that I have made the following official statements on several technical forum websites: Thermal compound reviews at www.overclockers.com should be disregarded as being misinformative, and completely inaccurate.
I do hope that the author reads this, and realizes that amongst the professional community, asking for more information or even admitting ignorance on a particular subject isn't a crime or weakness. It is actually considered a strength, and welcomed instead - for we're all learning. <|;^)>