Thermalright Showdown: TRUE Cu vs IFX-14

[LennyRhys]

As some of you know, I was looking for a cooler to replace my hefty Copper TRUE (Thermalright Ultra 120 Extreme): I wanted something that was not only lighter but also something that could at least match the performance of the mighty TRUE. A few days ago whilst conducting my daily "thermalright" search on ebay D) I noticed that somebody was selling a used IFX-14 at a very, very good price, so I took the plunge without needing to persuade myself and endured four days of agnoizing anticipation until the cooler finally arrived.

The coolers meet before the showdown:

One thing I wanted to draw particular attention to is the base of the IFX-14, which was so curved that it could be seen very clearly without the aid of a razor blade:

Despite the state of the IFX's base, I proceeded with the comparison to see how a "vanilla" IFX-14 would compare with a lapped TRUE Cu.

Here is the procedure I observed for the comparative tests:

• Prime 95 Blend was run for 20 mins using each cooler and each fan configuration. After 20 mins the max temperature reading for each core was taken from Real Temp and averaged out to give me a final value.
• A temperature threshold of 90°C was imposed and all tests were aborted if and when this threshold was reached or exceeded
• The Delta AFC1212E was placed between the two towers of the IFX-14; for dual fan tests, the IFX was configured as push-push whilst the TRUE Cu was configured as push-pull
• The same mounting hardware was used for both coolers (TR 1366 bolt through kit Rev. B)
• The same thermal compound was used for both coolers (Arctic MX-2)
• The ambient temperature was maintained at 19°C for all tests; there were no drafts or cold air currents which could interfere with the results

Speaking of results:

As you can probably understand, I was sorely disappointed in the IFX-14, both in terms of silent performance and voltage scaling performance: with the CPU at 1.4v (BIOS), the temperature reached 93°C even with the Delta at full throttle. At first I thought that the cooler might be faulty, but the other results clearly show that it can cope with a high heat load and that it scales well with increased airflow/static pressure - even to the extent that it matched the performance of the TRUE.

I also thought that the IFX-14 might benefit from having a 38mm fan on each tower rather than a single 38mm fan sandwiched between them, so I tested again with two powerful 38mm fans and the results were the same: temps rising to 90°C and beyond.

Not wanting to give in so easily, I lapped the IFX-14 and reinstalled it with a pressure mod to compensate for the copper that had been removed from the base.

Although lapping to a "mirror" finish doesn't do a great deal for temps, it certainly helps verify the flatness of the base!

And here is evidence of what appears to be very good contact with the IHS:

The result after lapping and pressure modding? Load temps were EVEN WORSE than before! Load temperatures at 4.1GHz with two Xilence fans rose to 90°C!!

I'm afraid I have no explanation for this - I fully expected lapping the IFX-14 to have a pronounced impact on its performance, but unfortunately it was not meant to be.

Despite the poor performance of the IFX-14 I really enjoyed conducting this comparative test, and I hope you enjoyed reading my findings.

 

[bing]

Dave, 1st, thanks for nice report, definitely a precious reference, but feel sorry for you since the ifx doesn't perform as expected.

Now questions, have you ever used prolimatech's product like megahalem ? Friend of mine managed to squeeze 3-4 degrees better temp from his silver arrow by using a modded megahalem's mounting. Imo, prolimatech's cpu mounting mechanism is the best compared to thermalright or noctua's like at D14. If you never use it before, I can post few shots to show you some details on the mounting.

I'm just speculating that the weight of this big heatsink + heavy fans might actually tilted it self against the cpu when the mobo was positioned vertically, making the contact point between the cpu ihs weaker. Just for curiousity, try do two runs with the mobo positioned horizontally versus vertical to see any significant different.

 

[capttripppp]

Awesome write up

 

[muddocktor]

bing, have you ever used the 1366 rev B TR mount? While it's not manufactured out of aluminum bar stock like the Prolimatech mount, it is a very good mounting system. And if the IFX14 is anything like a Silver Arrow in the weight department, then the vertical placement making for bad contact is a non-issue, because the Cu TRUE is almost 3X the mass of the IFX14 and that mount has no problems handling a Cu TRUE (using that mount with one right now myself).

Lenny, I am a bit surprised at the performance of the IFX14. I just put a Silver Arrow on my test rig and at 4.0, temps are looking pretty decent with it; nothing like you are seeing with the IFX14. I'm not making a test run with it, but am running a mock run and the highest temp I've seen so far with the stock fans is 70 C. I'll tell you what though; I don't like the new style fan clips that TR has gone with. Instead of holding the fan by the flange side closest to the heatsink, it reaches around to the far side to clip it and that way it won't hold a 38 mm fan. But I think the SA will be able to use the Ven-X fan clips. I'll test that out later.

EDIT: I see what you mean about the original base too. That sucker was really bowed a lot.

 

[LennyRhys]

@ bing,

Mud is right - the IFX-14 is very light, and what's more I was testing with a pressure mod which placed a lot more stress on the cooler's base. But thanks for the suggestion.

The Rev. B mount is easily one of my best purchases this year - SUCH a nice mount... but seriously, the IFX-14 is THE biggest pain in the *** to install because it's so hard to get access to the screws. It's even worse than the original TRUE by quite some margin. I had to use a pair of pliers to hold the first screw in place because there isn't nearly enough room for hands/fingers between the two cooling towers, and the tension is very high from the outset so it's difficult to get the screw in, not to mention difficult to see what you're doing LOL. I should have made a video of me installing it, complete with all my shouting every time I dropped a screw, hurt myself, or became snagged on the evil spikey fins

Installing the TRUE with the Rev. B mount is a joy and a pleasure because there's nothing in the way of the screws. LOVE IT!

@ Mud,

I should mention that I have a great chip for testing heatsinks - it's a sucker for overclocking, needing very high QPI voltage, but it runs really hot so gives heatsinks a very good run for their money. For 4GHz and with mem/uncore at 1530/3060, the system needs 1.4v QPI for stability. The 4.1GHz test was done with 1.425v QPI, and the 4.2GHz test 1.45v QPI.

Having thought about it for some time I think the bowed base is something of a paradox - it must have been making better contact with the center of the IHS with that ridiculous bow, which would explain why load temps were better before lapping. My CPU is also lapped which is why I thought I'd get better and more even contact with a lapped cooler, but this goes to show that it doesn't appear to be so simple. The TRUE is just far superior when it comes to handling higher heat loads, and my inkling is that six smaller heatpipes still trumps four larger ones.

Here's a picture of my mod which consisted of two pieces of 1.4mm mount board glued together and placed under the cross piece. There was a hole cut in the board to keep it centered. I dare you to try installing one of TR's coolers with this much added resistance LOL!

 

[SupaMonkey]

Nice review and interesting thread. Subscribed.

 

[muddocktor]

Lenny, did you try running it without the shims on the crossbar? What you could do for added pressure would be to get some washers you could install above the springs on the mounting screws. I'm wondering if maybe the shims were interfering with getting a good, even pressure across the top of the base. I know they look perfectly flat, but it is a possibility.

BTW, the mount on the SA is exactly like the one that came with the original HR-02 I reviewed a year ago. It's a little easier to mount with the SA than a rev. B would be, due to the mounting nuts being permanently captured by the crossbar. So you don't have to try to feed a couple of spring loaded bolts through the middle of the SA.

 

[LennyRhys]

I actually tried it without the pressure mod first, and I blamed the poor performance on insufficient pressure because it definitely installed more easily after lapping; however with the screws tightened right dow it was still very solid... these springs are extremely tough, as is the flanged crossbar!

Using the shims improved temps by 1C at best, which was still several degrees below the pre-lapping test with the same fans.

At the time of testing the coolers I had only one of my graphics cards installed, however with SLI there isn't nearly enough room for the IFX-14 because I have to have the HR-03GT on the top graphics card mounted on top of the PCB which eats into the CPU cooling space.

I'm still tempted to pull the trigger on a TRUE Rev.C purely to reduce the strain on the CPU socket, as this is the only way I can mount my TRUE at the moment:

 

[muddocktor]

A TRUE Rev. C is still a good choice, especially if you want to run big, bad-azz fans on it, because it scales so well with cfm and static pressure. You will also see some slightly lower temps than with the Cu TRUE but not anything that will let you overclock any higher. But it will drastically reduce the mass hanging off the mobo. Have you thought about bracing the top of the Cu TRUE to the case? Someone had posted a pic of 1 done like that a year or so ago; might have been burebista or bing that posted it.

 

[LennyRhys]

For the first few months I had it supported, but have been running it unsupported for at least six months (not in the configuration shown in that picture, however).

The more I thought atbout it, the more neurotic I became, so recently I've been using a rubber band stretched from the top of the case and holding onto the top of the heatpipes. I actually think the 8-layer PCBs of modern motherboards are easily strong enough to support it, but I'm still neurotic!!

You're not going to believe this, but there's another IFX-14 on ebay which I'm very tempted to buy.... call it morbid curiosity, but I would love to know for certain if the unit I currently have is faulty. Every review of the IFX-14 I can find puts it ahead of the TRUE/TRUE Cu by quite some margin. Annoyingly, if I install the IFX-14 I have to lose the SLI config.

 

[Bobnova]

The IFX-14 is supposed to beat the TRUE. Very strange that one doesn't IMO.

 

[LennyRhys]

I agree Bob - Xbitlabs crowned it as a clear winner among all the top coolers.

Well I said I was tempted to buy another IFX-14 to draw more comparisons... so I did.

I'm totally on a roll - this time I got it for £15 ($22) including postage, and what's more the seller is claiming it's new/unused. LOL!

I'll not have it for a while as I'm going to be away for the next week but I will run the same tests again in the new year. Very much looking forward to that!

 

[RGE]

Vapor had tested the IFX-14 vs True as well, and found True was much better, by ~8C or so on quad with 1.39 vcore.

Like Bobnova said, xbit labs on this graph had IFX-14 at ~5C better than True with 1500 rpm fans on i7920.

Will be interesting to see what your other one does...wonder if its a QC problem with heatpipes?

 

[muddocktor]

I seem to remember reading the tests on the IFX14 done by HardOCP and wasn't real impressed with the results then. Not nearly enough to spend the kind of money they were asking for it new back in the day.

 

[LennyRhys]

I agree that the IFX was overpriced, much as I think the SA and Archon also are. TRUE Rev.C might not come with fans but here in the UK it costs considerably less than the SA or Archon and is easily a contender where performance is concerned.

Anyways, I'm looking forward to running some tests and although I do not expect the IFX to beat the TRUE Cu (nor even to come close), it would be nice to see it fare better than its lapped sibling. I'll be home at the weekend so might be able to do something on Saturday, depending how tired I am and how likely it is that my wife will let me make a noise in the living room!

I stumbled upon This thread on a Romanian PC enthusiast forum in which one of the mods extensively (and I mean exhaustively) compares the TRUE to a number of different coolers. Whilst the tests do not feature any newer coolers because the thread is 2-3 years old, the conclusions do give a very good insight into the capabilities of "high performance" heatsinks which really shine when the TDP of the CPU is taken to extreme levels and when extremely high CFM fans are used.

His findings (and, seemingly, the general consensus) with the IFX-14 were similar to mine, although not nearly as damning - it was approx 5C behind the TRUE in his tests, and actually survived the extreme TDP torture, but in either case was not popular lol. If only I had found this before buying the first one... I might have gone instead for the Zalman 10X Extreme which purportedly beats the TRUE LOL.

 

[Tech Tweaker]

I just have to say it, as good as a TRUE looks in the regular finish, it looks even better in copper. I've never said this about a heatsink before, but that thing is beautiful in a copper finish.

Probably weighs a bloody ton though.

Also, I have a suggestion that may help you to get better temps. I would look into trying to find one of those bolt-through kits that have an X-bracket style mount. I think that was the revision 1.0 version for mounting the heatsink before they switched over to rev. 2 where it's just a metal bar that goes across the heatsink and is secured on both sides of the bar.

I had a mounting bracket similar to the one you show there on my 775 rig, but found it left my heatsink a little too loose and would swivel if accidentally bumped or nudged, then I got my hands on one of the other style of mounts and the mounting pressure of that one was much higher and my heatsink would no longer swivel around when bumped.

 

[inVain]

Sorry to hear the IFX-14 was on the losing side

At the time of testing the coolers I had only one of my graphics cards installed, however with SLI there isn't nearly enough room for the IFX-14 because I have to have the HR-03GT on the top graphics card mounted on top of the PCB which eats into the CPU cooling space.

I'm still tempted to pull the trigger on a TRUE Rev.C purely to reduce the strain on the CPU socket, as this is the only way I can mount my TRUE at the moment:

I really love to sandwich a 38mm fan with the Ven-X and the HR-03GT.

the idea was to exhaust the hot air straight to the top of the case

during my daily PC usage, temps was never hit 60 (Q6600 @3.2) and my GFX (old 8800GTX) was never hit 65 .... but never tried this setup on prime and furmark though

I'm still obsessed to mount an intake 120mm fan on the 5.25 bays which should improve airflow.

 

[muddocktor]

Lenny, poparamiro is a member here and posts occasionally. I've seen burebista link to their forum occasionally too and he's friends with paparamiro. He has done some pretty extreme testing, but haven't seen anything recently.

I'll try PM'ing him to see if he'll drop in and say something about his testing of the IFX14.

 

[LennyRhys]

@ Tech Tweaker, yeah the TRUE Cu is beautiful and polishes very nicely indeed. Looks great when the top of it has a mirror finish! It is very heavy too, nearly 2kg (4.4lbs)! As for mounting brackets, I have the X-style bracket for the TRUE and the crossbar style is actually better, believe it or not - it gives better mount pressure and provides better stability. Besides, I wanted to make this comparison apples-to-apples where possible, and since both coolers can use the Rev. B mount then I thought that'd be a good start. Thanks for the suggestion though

@ InVain, my GTX460 is too hot for a pulling fan haha I have a Delta 92mm push on it and it works perfectly on low speed.

@ Mud, that's excellent news - would be great to have him post here.

 

[LennyRhys]

Well I got back home yesterday and none of the stuff I bought has arrived yet - must be the holiday period slowing everything down. I had hoped the new IFX would have arrived today but alas nothing came in the post.

I'll be mega busy back to work this coming week but I'll try to find some time to continue/conclude this review.