Madshrimps Forum Madness

Madshrimps Forum Madness (http://www.madshrimps.be/vbulletin/)
-   Air - Water - Peltier - Phase Change - LN2 Cooling Projects (http://www.madshrimps.be/vbulletin/f9/)
-   -   Second look at IFX-10 (http://www.madshrimps.be/vbulletin/f9/second-look-ifx-10-a-63795/)

photonblaster 17th May 2009 15:25

Second look at IFX-10
 
The I7 runs very hot when in extreme OC so a 5C temp drop in HS performance starts to become very critical, which got me thinking some more about the IFX-10 and IFX-14.
The IFX-10 is a small heat pipe HS that mounts its block on the back of the CPU. It is available either with the IFX-14 or separately (at least by the dimensions on Thermalright's WWW they seem the same but I am not sure IFX-14 version has fan tabs). Most IFX-14 reviews I have seen pay little attention to the IFX-10. I even saw one review where wiring was bunched up above the IFX-10 impeding convection flow. The IFX-10 review I liked the most is at Hardware Canucks: ....hardwarecanucks.com/forum/hardware-canucks-reviews/5600-thermalright-ifx-10-motherboard-backside-cooler-review-8.html.. (Sorry, I do not have enough posts to be allowed to do the url short form). This review shows that at higher temps the IFX-10 plays a more important role in the cooling. At 70C with the IFX-14 it gives more than 4degC temp drop with a fan. The reviews for the IFX-14 do not seem to have a fan on the IFX-10.

It would be great to see the IFX-10 revisited for extremely hot CPUs:
1) IFX-14 with and without fan on the IFX-10
2) Take the closest competitors to the IFX-14 and see if they could beat it if combined with the IFX-10.

I am sure I am not the first to consider these tests...anyone have firsthand experience with either of these? If not, are these ideas worthy of some MadShrimps tests?

jmke 17th May 2009 18:38

in my humble opinion only viable if you plan to benchmark your system outside a case and want the best possible cooling on air; the IFX-10 is the only product available that cools the backside of the motherboard, so it becomes an easy choice.

inside a case you'll much faster run into compatibility issues regarding to size; and using it passively is the only way to do it if your aim is a balance between performance & noise. If you starting adding a 80mm fan on the IFX-10 you can throw "low noise" computing out the door.

I tested the IFX-14 with IFX-10 backplate, it claimed top performance spot, most likely also thanks to the IFX-10 doing its magic. Even if the case used was a compact mid-tower and no active cooling was used, the IFX-10 does work, shaving off a few degrees under load.
http://www.madshrimps.be/?action=getarticle&articID=673


Now if your system's stability and usability depend on a few degrees C, there are other things to look at first ;)

photonblaster 17th May 2009 20:49

Thanks for your response, jmke. I did read the linked article as well as many others.

As you mentioned, if you are considering using the IFX-10, you have to make sure it fits. Same if you want to use one of the tall towers. But lets assume it fits.

You also mention 80 mm fan as not compatible with low noise. Let's assume we use a low noise fan running at moderate RPM so the added noise is in the 20db range. This would raise the db of a 30db system to 30.4 db. Even ramping the RPM up to 25db, the system would only be at 31.2db. And adding 30db to 40db system is also essentially no change. So let's assume we use low noise 80 mm fan. 30-40 cfm is not super flow, but a lot better than convection cooling of the IFX-10. One key item to consider is the temp of the fins...they are running hot which increases the temp drop the -10 can provide. If your CPU temps are already well under 70C then don't bother even thinking about my post because the gain in using the -10 would indeed be small by almost anyone's standards.

You mention that there are other things to look at. I absolutely agree, but you have to look at them one at a time. I just happened to pick this one to post first. I am also looking at air flow patterns, especially MB backside cooling. I have a carreer in R&D, so this is the type of thing I do for fun. What else do you think has room for improvement or where conventional wisdom should be re-thought?

Al that being said, my estimate is that an heavy OC of I7 on air with a good HS could run cooler by 5 degC if coupled with the IFX-10 plus fan. Wouldn't it be neat if a new "top performer" could be crowned? Besides this in my humble opinion 5degC is much more than "shaving off a few degrees under load" if your CPU is running near 70C. For me, it can make the difference in reaching my goal of building an I7 system OC to >=4GHZ for my son's graduation present. I can't afford to buy the components and check things out myself, so I am hoping for real life data from all the pro's here on the board.

I think there are several reasons for continuing to look at the IFX-10. First, the IFX-14 reviews did not have air cooling for the -10 so its full potential was not examined. Second, wouldn't it be neat if -10 could be combined with a more compact tower and actually beat the IFX-14? Some will say that this is cheating adding an extra cooler to a tower HS, but in that case Thermalrite could be accused of "cheating" too!

SO:
Anyone have experience putting a fan on the IFX-10 to see the difference in performance?
Coupling it with another tower?
Anyone intrigued by possibility of making a new "king of the hill" and want to try it with all those HS's they have lying around?

jmke 18th May 2009 09:01

Quote:

this in my humble opinion 5degC is much more than "shaving off a few degrees under load" if your CPU is running near 70C.
if 5C CPU load difference is crash or stable then you have other things to look at first; room temp can easily fluctuate 5-10C; you don't want a PC that crashes on your in the summer; or in the winter when you heat your room.

Quote:

Let's assume we use a low noise fan running at moderate RPM so the added noise is in the 20db range. This would raise the db of a 30db system to 30.4 db
when talking dB (not dBA) a 3dB increase results in a sound pressure TWICE as loud. if you're talking dBA, it takes 10dBA to make the system twice as loud. I have not encountered a 80mm fan yet which can considered "quiet"; rather place some ducting around the IFX-10 which helps guide air over the HSF from your in-take/out-take fan.

photonblaster 18th May 2009 14:18

Quote:

Originally Posted by jmke (Post 238090)
when talking dB (not dBA) a 3dB increase results in a sound pressure TWICE as loud. if you're talking dBA, it takes 10dBA to make the system twice as loud. I have not encountered a 80mm fan yet which can considered "quiet"; rather place some ducting around the IFX-10 which helps guide air over the HSF from your in-take/out-take fan.

My bad, I dropped the A like most people. You might enjoy this reference I found a while back at SPRC Re 10 dBA doubling of sound perception: "Yes, 10dB is the rule of thumb often used in engineering. How ever, psychoacoustics is the science that measures this. "Half-Loudness corresponds to a 6-dB attenuation rather than the 10 dB suggested..." (ref, Psychology of Hearing, 5th ed., Brian Moore, 2003). There is considerable variance based on spectral distribution, preceding sounds, masking and other factors." (Post by Felger Carbon on this thread http://www.silentpcreview.com/forums...=6dba&start=30 near the end) So I use 6dBA when considering noise level changes to be on the safe side. Doesn't really matter though if adding a fan calculates to 1dBA increase. Here is a link to fan roundup that shows several 80 mm fans with low dBA (and low CFM of course compared to a 80 mm screamer or a 120 mm low noise fan): http://www.pcper.com/article.php?aid=-232 Low is subjective, suffice it to say I consider adding a <1dBA increase in system noise fan as adding a low noise fan to my system.

Back to the 5degC drop as being significant in operating a hot CPU. Can we at least agree that it is not nice to operate your I7 CPU above 70degC or so even though it has a max operating temp of 90C? And if it is operating at 70C range, a 5degC drop in temp starts to be big enough to extend its expected lifetime or at least make the user feel a little more comfortable about the operating temp?

Or, if not this agreement, can we agree that it would be neat to see a new record air cooling option? If we agree here, then we hopefully we can agree on the main reason for my original post....I think the IFX-10 deserves to be revisited, and hopefully someone can share some real world experience with it combined with different towers and the effects of a fan.

jmke 18th May 2009 14:25

70C is well within operating spec for the Core i7; so I wouldn't worry about that. I would worry if the CPU passes over the 90C mark at regular intervals.

The "increase life expectancy" argument could be used if you're running the CPU at 120C all the time. I still have a working Pentium 1 200MMX system that has been running for 24/7 as Linux router without signs of failing; so life expectancy won't be noticeably reduced because you run the CPU 20C under max operating temp :)

Quote:

Or, if not this agreement, can we agree that it would be neat to see a new record air cooling option? If we agree here, then we hopefully we can agree on the main reason for my original post....I think the IFX-10 deserves to be revisited, and hopefully someone can share some real world experience with it combined with different towers and the effects of a fan.
I agree on this; but max overclocks on AIR is on an open bench, at which point the addition of high CFM fans and what not all equate to top clocks; for a 24/7 stable OC inside a case, the system should never be stable/unstable depending on <10C temp swings:)

ps:
Quote:

Originally Posted by photonblaster (Post 238106)

they're using the factory provided dBA ratings... no dBA meter used; hence no real noise numbers in that old article;)
I did a similar test http://www.madshrimps.be/?action=get...&articID=19 2 5 years ago before I got a dBA meter, but added more noise info;)

photonblaster 18th May 2009 15:23

Thanks, jmke!
Very nice link you gave me on 80 mm fans. I definitley will make use of the info since I am probably going to buy Elementt S case for the build (my son loves color red and this one has nice cable management and looks elegent with red trim even though its air flow options are not great) which has two 80 mm fan option. Plus of course potentially buying one for an IFX-10 :) Re operating temp, other reads seem more sensitive to temp than your insights, but I probably trust your expertise more than others. Thanks again for the advice.


Now if only someone has some data on the IFX-10???


All times are GMT +1. The time now is 10:52.

Powered by vBulletin® - Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
Content Relevant URLs by vBSEO