Nine Big Air Coolers For Intel's Haswell CPUs, Reviewed

[Crashman]

I don't understand the results, because the Phanteks PH-TC14PE had the lowest Delta T (64* above ambient) and yet the Noctura was selected as the best of the coolers tested? What am I missing?

Yogi

It also had higher chipset temperature, and by a larger delta. And it had worse noise. And it had a worse cooling-to-noise ratio, which is the true measure of total performance. And it had a worse value score. Did you read the article or just pick the one measurement you wanted to see?

 

[Y0GI]

I don't understand the results, because the Phanteks PH-TC14PE had the lowest Delta T (64* above ambient) and yet the Noctura was selected as the best of the coolers tested? What am I missing?

Yogi

It also had higher chipset temperature, and by a larger delta. And it had worse noise. And it had a worse cooling-to-noise ratio, which is the true measure of total performance. And it had a worse value score. Did you read the article or just pick the one measurement you wanted to see?

Yes, I read the article - more than once - trying to figure out the results. I didn't understand the terminology for "Chipset Temperature", was it MB PWM?

Thanks for the clarification, Crashman!

Yogi

 

[Crashman]

I don't understand the results, because the Phanteks PH-TC14PE had the lowest Delta T (64* above ambient) and yet the Noctura was selected as the best of the coolers tested? What am I missing?

Yogi

It also had higher chipset temperature, and by a larger delta. And it had worse noise. And it had a worse cooling-to-noise ratio, which is the true measure of total performance. And it had a worse value score. Did you read the article or just pick the one measurement you wanted to see?

Yes, I read the article - more than once - trying to figure out the results. I didn't understand the terminology for "Chipset Temperature", was it MB PWM?

Thanks for the clarification, Crashman!

Yogi

Sorry, I used the onboard thermal sensor for chipset temperature back then. I'm using a Thermistor these days placed directly at the PWM. The first method requires a heat pipe to tie temperature readings to a single place, the second method is better since it measures only the most important onboard component to be cooled.

 

[Y0GI]

I don't understand the results, because the Phanteks PH-TC14PE had the lowest Delta T (64* above ambient) and yet the Noctura was selected as the best of the coolers tested? What am I missing?

Yogi

It also had higher chipset temperature, and by a larger delta. And it had worse noise. And it had a worse cooling-to-noise ratio, which is the true measure of total performance. And it had a worse value score. Did you read the article or just pick the one measurement you wanted to see?

Yes, I read the article - more than once - trying to figure out the results. I didn't understand the terminology for "Chipset Temperature", was it MB PWM?

Thanks for the clarification, Crashman!

Yogi

Sorry, I used the onboard thermal sensor for chipset temperature back then. I'm using a Thermistor these days placed directly at the PWM. The first method requires a heat pipe to tie temperature readings to a single place, the second method is better since it measures only the most important onboard component to be cooled.

From what I know, the term "PWM" stands for "Pulse Width Modulation" which is a description of a voltage variation method. However, you are using it to denote a piece of hardware on the MB. I'm not trying to be picky here - just trying to understand. Are you referring to the MOSFETs which are parts of the VRMs or what? I count 17 MOSFETS in the photo of the MB and they are on 2 sides of the CPU socket, so which one does the MB monitor and where do you place your sensor to use instead?

 

[Crashman]

where do you place your sensor to use instead?

I use the term PWM rather than VRM because voltage regulators are no longer modular (the M in VRM). I stick the probe at the base of the main sink, behind a choke. The "main" part of the voltage regulator depends on the board, it's usually behind the socket except on LGA 2011, where it's instead usually above the socket.

 

[Y0GI]

where do you place your sensor to use instead?

I use the term PWM rather than VRM because voltage regulators are no longer modular (the M in VRM). I stick the probe at the base of the main sink, behind a choke. The "main" part of the voltage regulator depends on the board, it's usually behind the socket except on LGA 2011, where it's instead usually above the socket.

Thank you for the additional clarification, Crashman! Keep up the great work!

Yogi

 

[jeffredo]

For this sort of a test the Silverstone AR03 would have been more appropriate than the AR01. Its a given its going to be better with serious overclocking than the mainstream AR01 (while still being under $50).

 

[WebWalker55]

Running AC "for hours" so a cooler can stay in the competition sounds dumb... Flunk the cooler for not handling Real Life(tm).
My AC stays at 77F so the house stays 80F or lower... Inside the case is a few degrees warmer: 87F Which cooler gets the CPU down to what temp now?

 

[Crashman]

Running AC "for hours" so a cooler can stay in the competition sounds dumb... Flunk the cooler for not handling Real Life(tm).
My AC stays at 77F so the house stays 80F or lower... Inside the case is a few degrees warmer: 87F Which cooler gets the CPU down to what temp now?

You mean flunk the CPU. Because none of the coolers would have been sufficient at an 80° F room temperature

 

[WebWalker55]

Running AC "for hours" so a cooler can stay in the competition sounds dumb... Flunk the cooler for not handling Real Life(tm).
My AC stays at 77F so the house stays 80F or lower... Inside the case is a few degrees warmer: 87F Which cooler gets the CPU down to what temp now?

You mean flunk the CPU.

I'll go with that! This is the conclusion the article carefully /almost/ shouts.

Because none of the coolers would have been sufficient at an 80° F room temperature

Hows about Tom & Co pay attention to this aspect of Real Life(tm)? Temperature and humidity affect overclock ability of *any* hardware rig. (except maybe the LN crowd, but they're psychowhacked to start!) Phase Change rigs have to dump the heat *somewhere*, water-cooled rigs have to dump the heat *somewhere*, air-cooled ... we just covered that.

 

[Crashman]

Running AC "for hours" so a cooler can stay in the competition sounds dumb... Flunk the cooler for not handling Real Life(tm).
My AC stays at 77F so the house stays 80F or lower... Inside the case is a few degrees warmer: 87F Which cooler gets the CPU down to what temp now?

You mean flunk the CPU.

I'll go with that! This is the conclusion the article carefully /almost/ shouts.

Because none of the coolers would have been sufficient at an 80° F room temperature

Hows about Tom & Co pay attention to this aspect of Real Life(tm)? Temperature and humidity affect overclock ability of *any* hardware rig. (except maybe the LN crowd, but they're psychowhacked to start!) Phase Change rigs have to dump the heat *somewhere*, water-cooled rigs have to dump the heat *somewhere*, air-cooled ... we just covered that.

It's the problem of unrealistic expectations. People know these can support high overclocks without crashing. People know they run hot. But most people DON'T test coolers with full load on the AVX pipeline and across all cores, even virtual cores. And most people DON'T use proper temperature monitoring software, so they WOULDN'T even look for thermal throttling.

We try to get the word out. We try to max out the CPU at achievable room temperatures and a low-enough voltage level that we don't need to worry about electromigration, at least not over the 1-3 years that most users expect out of a minor overclock.

This little experiment proved that no cooler is really good enough for Haswell, because Haswell has trouble getting its heat to the heat spreader. That means this isn't the proper CPU for testing coolers, which is why the upcoming sealed-liquid update reverts back to SB-E.

 

[Y0GI]

Running AC "for hours" so a cooler can stay in the competition sounds dumb... Flunk the cooler for not handling Real Life(tm).
My AC stays at 77F so the house stays 80F or lower... Inside the case is a few degrees warmer: 87F Which cooler gets the CPU down to what temp now?

You mean flunk the CPU.

I'll go with that! This is the conclusion the article carefully /almost/ shouts.

Because none of the coolers would have been sufficient at an 80° F room temperature

Hows about Tom & Co pay attention to this aspect of Real Life(tm)? Temperature and humidity affect overclock ability of *any* hardware rig. (except maybe the LN crowd, but they're psychowhacked to start!) Phase Change rigs have to dump the heat *somewhere*, water-cooled rigs have to dump the heat *somewhere*, air-cooled ... we just covered that.

It's the problem of unrealistic expectations. People know these can support high overclocks without crashing. People know they run hot. But most people DON'T test coolers with full load on the AVX pipeline and across all cores, even virtual cores. And most people DON'T use proper temperature monitoring software, so they WOULDN'T even look for thermal throttling.

We try to get the word out. We try to max out the CPU at achievable room temperatures and a low-enough voltage level that we don't need to worry about electromigration, at least not over the 1-3 years that most users expect out of a minor overclock.

This little experiment proved that no cooler is really good enough for Haswell, because Haswell has trouble getting its heat to the heat spreader. That means this isn't the proper CPU for testing coolers, which is why the upcoming sealed-liquid update reverts back to SB-E.

So, overclocking, to the degree used in this review, can result in CPU life of only 1-3 years?

Yogi

 

[Crashman]

Running AC "for hours" so a cooler can stay in the competition sounds dumb... Flunk the cooler for not handling Real Life(tm).
My AC stays at 77F so the house stays 80F or lower... Inside the case is a few degrees warmer: 87F Which cooler gets the CPU down to what temp now?

You mean flunk the CPU.

I'll go with that! This is the conclusion the article carefully /almost/ shouts.

Because none of the coolers would have been sufficient at an 80° F room temperature

Hows about Tom & Co pay attention to this aspect of Real Life(tm)? Temperature and humidity affect overclock ability of *any* hardware rig. (except maybe the LN crowd, but they're psychowhacked to start!) Phase Change rigs have to dump the heat *somewhere*, water-cooled rigs have to dump the heat *somewhere*, air-cooled ... we just covered that.

It's the problem of unrealistic expectations. People know these can support high overclocks without crashing. People know they run hot. But most people DON'T test coolers with full load on the AVX pipeline and across all cores, even virtual cores. And most people DON'T use proper temperature monitoring software, so they WOULDN'T even look for thermal throttling.

We try to get the word out. We try to max out the CPU at achievable room temperatures and a low-enough voltage level that we don't need to worry about electromigration, at least not over the 1-3 years that most users expect out of a minor overclock.

This little experiment proved that no cooler is really good enough for Haswell, because Haswell has trouble getting its heat to the heat spreader. That means this isn't the proper CPU for testing coolers, which is why the upcoming sealed-liquid update reverts back to SB-E.

So, overclocking, to the degree used in this review, can result in CPU life of only 1-3 years?

Yogi

I think 1.30V would get you into the 1-3 year range. Unfortunately, the CPU will thermal-throttle at anything more than 1.25V. Fortunately, that should give you the high range (3-years or more).

Since we're guessing based on prior experience and CPU behavior, you can't expect those guesses to be perfect until someone has actually ran one at 1.25V for at least 3 years.

 

[Y0GI]

Running AC "for hours" so a cooler can stay in the competition sounds dumb... Flunk the cooler for not handling Real Life(tm).
My AC stays at 77F so the house stays 80F or lower... Inside the case is a few degrees warmer: 87F Which cooler gets the CPU down to what temp now?

You mean flunk the CPU.

I'll go with that! This is the conclusion the article carefully /almost/ shouts.

Because none of the coolers would have been sufficient at an 80° F room temperature

Hows about Tom & Co pay attention to this aspect of Real Life(tm)? Temperature and humidity affect overclock ability of *any* hardware rig. (except maybe the LN crowd, but they're psychowhacked to start!) Phase Change rigs have to dump the heat *somewhere*, water-cooled rigs have to dump the heat *somewhere*, air-cooled ... we just covered that.

It's the problem of unrealistic expectations. People know these can support high overclocks without crashing. People know they run hot. But most people DON'T test coolers with full load on the AVX pipeline and across all cores, even virtual cores. And most people DON'T use proper temperature monitoring software, so they WOULDN'T even look for thermal throttling.

We try to get the word out. We try to max out the CPU at achievable room temperatures and a low-enough voltage level that we don't need to worry about electromigration, at least not over the 1-3 years that most users expect out of a minor overclock.

This little experiment proved that no cooler is really good enough for Haswell, because Haswell has trouble getting its heat to the heat spreader. That means this isn't the proper CPU for testing coolers, which is why the upcoming sealed-liquid update reverts back to SB-E.

So, overclocking, to the degree used in this review, can result in CPU life of only 1-3 years?

Yogi

I think 1.30V would get you into the 1-3 year range. Unfortunately, the CPU will thermal-throttle at anything more than 1.25V. Fortunately, that should give you the high range (3-years or more).

Since we're guessing based on prior experience and CPU behavior, you can't expect those guesses to be perfect until someone has actually ran one at 1.25V for at least 3 years.

WOW!! first time that I've seen anybody predict such a drastically reduced CPU lifespan due to OC'ing!

Thank you, Crashman!

 

[Crashman]

WOW!! first time that I've seen anybody predict such a drastically reduced CPU lifespan due to OC'ing!

Thank you, Crashman!

I'm hearing from a motherboard rep that a few people have reported failures at 1.35V already. From there you look at what's worked in the past and what's available now and try to narrow your range.

 

[Y0GI]

WOW!! first time that I've seen anybody predict such a drastically reduced CPU lifespan due to OC'ing!

Thank you, Crashman!

I'm hearing from a motherboard rep that a few people have reported failures at 1.35V already. From there you look at what's worked in the past and what's available now and try to narrow your range.

Very interesting! I very much appreciate your informed opinion!

Yogi

 

[Roger Rogers-1382392]

I installed the ZALMAN CNPS14X cooler onto the Asus Z78 Sabertooth.

Now there is a nearly impossible set of bolts to entertain yourself with for a few frustrating hours; with a magnifying glass, a torch and the patience of a saint.

 

[Y0GI]

I installed the ZALMAN CNPS14X cooler onto the Asus Z78 Sabertooth.

Now there is a nearly impossible set of bolts to entertain yourself with for a few frustrating hours; with a magnifying glass, a torch and the patience of a saint.

That is the main reason that I eliminated the Zalman cooler from consideration when I was buying. Almost all of the customer reviews reported great difficulty with installation.

Yogi

 

[Crashman]

I installed the ZALMAN CNPS14X cooler onto the Asus Z78 Sabertooth.

Now there is a nearly impossible set of bolts to entertain yourself with for a few frustrating hours; with a magnifying glass, a torch and the patience of a saint.

Maybe that's why they never sent me one to review. I've installed most of its other products with great ease...but maybe I've been doing this too long...

 

[Roger Rogers-1382392]

I installed the ZALMAN CNPS14X cooler onto the Asus Z78 Sabertooth.

Now there is a nearly impossible set of bolts to entertain yourself with for a few frustrating hours; with a magnifying glass, a torch and the patience of a saint.

Maybe that's why they never sent me one to review. I've installed most of its other products with great ease...but maybe I've been doing this too long...

Good on....
And I have shimmed a V12 Jag engine (hydraulic tappet set), cleaned and re-built assault rifles in the pitch dark, and swoped the clutch out of a Rover 820 in the garden with just two adjustable spanners and a stick.
But the Zalman14X and the Z87 Sabertooth are quite the tougher combination.

I almost got the mini-grinder out.

 

[Novuake]

Good on....
And I have shimmed a V12 Jag engine (hydraulic tappet set), cleaned and re-built assault rifles in the pitch dark, and swoped the clutch out of a Rover 820 in the garden with just two adjustable spanners and a stick.
But the Zalman14X and the Z87 Sabertooth are quite the tougher combination.

I almost got the mini-grinder out.

Pfft gimme a Fairlane 4.2l V8 or even a Winsor 3.6l above your Jag V12 screamer anyday!

 

[Crashman]

I installed the ZALMAN CNPS14X cooler onto the Asus Z78 Sabertooth.

Now there is a nearly impossible set of bolts to entertain yourself with for a few frustrating hours; with a magnifying glass, a torch and the patience of a saint.

Maybe that's why they never sent me one to review. I've installed most of its other products with great ease...but maybe I've been doing this too long...

Good on....
And I have shimmed a V12 Jag engine (hydraulic tappet set), cleaned and re-built assault rifles in the pitch dark, and swoped the clutch out of a Rover 820 in the garden with just two adjustable spanners and a stick.
But the Zalman14X and the Z87 Sabertooth are quite the tougher combination.

I almost got the mini-grinder out.

Oh yeh? Well I've TUNED A QUADRAJET, so everything seams easy to me 😛

 

[warrenhenery]

Government and Municipal spaces such as Airports, Courtrooms, and Hospitals, Shooting ranges and indoor stadiums and gymnasiums all need soundproofing and using acoustic panels is the easiest solution. Buy <a href="http://www.mixmasteredacoustics.com/" rel="dofollow">Acoustic Panels</a> at Mix Mastered Acoustics.

 

[Lima Lcr]

Leaving D14 sense like commercializing bullcrap compitators. D14 is still the air king lol

 

[mapesdhs]

There are various air coolers which can beat the D14, eg. the Phanteks PH-TC14PE, and at
the time the latter launched it was quite a bit cheaper.

I'm switching over to water cooling though, H80i costs less than either of the above air
coolers, works better, etc.

Ian.