Containing a 4790k overclock with air...?

[GuruMeditationError]

I've seen a few videos showing air coolers beating AIO's and so have been thinking about converting to air...

 

...the thing is, I've got a liquid metal delid, a Rockit Cool copper heatspreader and an H115i Pro with four 140mm Akasa Viper R's in push pull and it still can't seem to contain the overclock when running Prime 95 (ambient temperature 30 to 35 degrees C).

 

We've had an unusually hot (record breaking) 30 to 35 degrees C ambient temperature where I am, so I thought I'd run a thermal test with Prime 95 and it only took a few minutes for the fans to ramp up and suddenly accelerate to maximum RPM.

 

The Viper R's are extremely high performance, spinning up to a maximum of 1600 RPM and pushing 109.55 CFM at a pressure of 2.91 mm-H2O

I stopped the test when the CPU temp hit 50 C (It was the first time I've seen it hit 50 C since I fitted the H115i Pro) and I'm assuming it would have slowly ticked up a degree at a time, since it seems the cooler was probably saturated. 

So...is Prime 95 just the wrong kind of test to throw at it...?...or can I assume that switching to a dual-tower air cooler with a couple of fans attached is just not going to be up to the job of containing the overclock?

I should probably mention I have two more Viper R's in the base of the case as intakes, also controlled by the CPU via a powered PWM fan splitter/adapter to help feed the radiator fans.

Given the above, is switching to air a bad idea, or would it be okay as long as I'm not maxing out all cores for minutes at a time with something like Prime? It's a gaming rig (no video processing etc.) that runs hottest when installing Steam games and when booting up.

Advice would be most appreciated,      

[Hans Christian | Teri]

I'd let it run till it hit 80, if the rise hadn't stopped before then (which I imagine it would have, depending on your overclock). An air cooler probably wouldn't help much, if anything.

[Hairless Monkey Boy]

5 minutes ago, GuruMeditationError said:

I stopped the test when the CPU temp hit 50 C

Why are you so concerned about 50C? Up to the high 80Cs low 90Cs would be reasonable in Prime95 with a decent OC.

[GuruMeditationError]

Just now, HairlessMonkeyBoy said:

Why are you so concerned about 50C? Up to the high 80Cs low 90Cs would be reasonable in Prime95.

Because it looks like it maxed out the cooler. I was assuming the temperature would just continue to rise until the CPU started thermal throttling. Isn't that what would happen?

[Hairless Monkey Boy]

Just now, GuruMeditationError said:

Because it looks like it maxed out the cooler. I was assuming the temperature would just continue to rise until the CPU started thermal throttling. Isn't that what would happen?

Almost certainly not. Thermal throttle doesn't occur until 100C.

[GuruMeditationError]

1 minute ago, HairlessMonkeyBoy said:

Almost certainly not. Thermal throttle doesn't occur until 100C.

 

5 minutes ago, Hans Christian | Teri said:

I'd let it run till it hit 80, if the rise hadn't stopped before then (which I imagine it would have, depending on your overclock). An air cooler probably wouldn't help much, if anything.

Okay, I'll give it a try and see if it stabilises.

[Hairless Monkey Boy]

Just now, GuruMeditationError said:

 

Okay, I'll give it a try and see if it stabilises.

What is your OC voltage and clock speed?

[Bombastinator]

2 minutes ago, GuruMeditationError said:

Because it looks like it maxed out the cooler. I was assuming the temperature would just continue to rise until the CPU started thermal throttling. Isn't that what would happen?

Not necessarily. it’s an albedo issue. As the albedo rises  a cooler gets more efficient. Humidity matters too.  If ambient air was 36c 50c is albedo of only 14c which isn’t much.

[GuruMeditationError]

2 minutes ago, HairlessMonkeyBoy said:

What is your OC voltage and clock speed?

I've no idea, I'm using Asus' OC'ing application in the BIOS.

1 minute ago, Bombastinator said:

Not necessarily. it’s an albedo issue. As the albedo rises  a cooler gets more efficient. Humidity matters too.  If ambient air was 36c 50c is albedo of only 14c which isn’t much.

Yeah, I kind of figured that but I just thought the temperature would keep rising and not stabilise.

[GuruMeditationError]

Okay, so...it seemed to stabilised at 53 C after 4 to 5 minutes; with the fans fluctuating at from around 97 to 100% their maximum RPM

How does that look for air? Could an NH-D15 contain it?   

[Bombastinator]

21 minutes ago, GuruMeditationError said:

I've no idea, I'm using Asus' OC'ing application in the BIOS.

Yeah, I kind of figured that but I just thought the temperature would keep rising and not stabilise.

It would stabilize it’s just a question of where it would stabilize.  I was told not to run a 4770k over 75c (which may or may not be true) but my system (4770k @4ghz) with silicon lottery loser chip stabilizes in the low 70’s most of the time on prime95. Ive got an old viperX  aircooler. If your pump is dead (which can happen with water coolers) you could have heat problems.  Checking for large heat differences in hose temperature can sometimes identify that.  Different problem than water vs air though. 

[Bombastinator]

3 minutes ago, GuruMeditationError said:

Okay, so...it seemed to stabilised at 53 C after 4 to 5 minutes; with the fans fluctuating at from around 97 to 100% their maximum RPM

How does that look for air? Could an NH-D15 contain it?   

Water takes time to heat up.  It’s got a lot of thermal mass.  It wouldn’t surprise me if it took an hour to totally stabilize.

[GuruMeditationError]

11 minutes ago, Bombastinator said:

Water takes time to heat up.  It’s got a lot of thermal mass.  It wouldn’t surprise me if it took an hour to totally stabilize.

Thanks, I thought the sudden increase in fan speed was the cooler being saturated, but it might have just been that Prime 95 varies its routines? After about ten minutes the temperature and the fan speed dropped considerably. I'm not sure why else that would happen unless the water in the loop momentarily sped up?

[Bombastinator]

15 minutes ago, GuruMeditationError said:

Thanks, I thought the sudden increase in fan speed was the cooler being saturated, but it might have just been that Prime 95 varies its routines? After about ten minutes the temperature and the fan speed dropped considerably. I'm not sure why else that would happen unless the water in the loop momentarily sped up?

I don’t know what heat sensor you have your fans tied to so it’s hard to know.

[GuruMeditationError]

1 minute ago, Bombastinator said:

I don’t know what heat sensor you have your fans tied to so it’s hard to know.

They're being controlled by the CPU fan controller via a powered PWM fan splitter.

But more importantly, do you think I'd be okay for switching to air?

[For Science!]

52 minutes ago, Bombastinator said:

Not necessarily. it’s an albedo issue. As the albedo rises  a cooler gets more efficient. Humidity matters too.  If ambient air was 36c 50c is albedo of only 14c which isn’t much.

I think you mean delta T (Δt), as in the difference of two temperatures? All searching of "albedo" comes to reflectivity and/or astronomy terms

 

Quote

albedo

/alˈbiːdəʊ/

noun

ASTRONOMY

noun: albedo; plural noun: albedos

1. the proportion of the incident light or radiation that is reflected by a surface, typically that of a planet or moon.

   "the lunar maria have a lower albedo than the surrounding terrain"

https://en.wikipedia.org/wiki/Albedo

[GuruMeditationError]

1 minute ago, For Science! said:

I think you mean delta T (Δt), as in the difference of two temperatures? All searching of "albedo" comes to reflectivity and/or astronomy terms

 

https://en.wikipedia.org/wiki/Albedo

Yeah, I assumed that's what he meant...

...but what do you think re. the original question? 

Do you think I'd be okay for air?

[For Science!]

2 minutes ago, GuruMeditationError said:

Yeah, I assumed that's what he meant...

...but what do you think re. the original question? 

Do you think I'd be okay for air?

I don't get why you think the cooler is not containing the OC, and I also don't get why you won't just let the temperatures actually stabilize over say 20 minutes to find out if you cooler is actually heat soaking or not. Your CPU is well in the safe zone until at least 75 degrees, so you have plenty of headroom. Side-grading (at most) to an aircooler will get you nothing in this case since your AIO sounds like its still working well to me.

[Bombastinator]

9 minutes ago, GuruMeditationError said:

They're being controlled by the CPU fan controller via a powered PWM fan splitter.

But more importantly, do you think I'd be okay for switching to air?

I think you would spend money and gain nothing of note unless your AIO is dead. AIOs eventually die though. 

[GuruMeditationError]

1 minute ago, For Science! said:

I don't get why you think the cooler is not containing the OC, and I also don't get why you won't just let the temperatures actually stabilize over say 20 minutes to find out if you cooler is actually heat soaking or not. Your CPU is well in the safe zone until at least 75 degrees, so you have plenty of headroom. Side-grading (at most) to an aircooler will get you nothing in this case since your AIO sounds like its still working well to me.

I want to switch to a dual-fan heatsink with quiet fans to reduce the noise in my system. I'm hoping to use the overhead to lower the noise.

I thought the info I've provided so far might be enough to go on; is that not the case?

How long should I let it run for to get a sense of where it's stabilising? Bombastinator suggested it could take up to an hour.

[Bombastinator]

22 minutes ago, For Science! said:

I think you mean delta T (Δt), as in the difference of two temperatures? All searching of "albedo" comes to reflectivity and/or astronomy terms

 

https://en.wikipedia.org/wiki/Albedo

I’m parroting stuff that was told to me.  The teller called it albedo.  If that is not the correct term it is not the correct term.  The person in question knew a great deal about computers though I have lately learned he apparently lied to me occasionally for his own amusement. It’s possible this was one of those times.  Albedo etymologically means “whiteness” apparently. he might have used it because he knew I knew about photography and the concept mimics that used to measure relative light sources.

[GuruMeditationError]

3 minutes ago, Bombastinator said:

I think you would spend money and gain nothing of note unless your AIO is dead. AIOs eventually die though. 

As per the above post: I want to use the overhead to go with quieter fans. 

[Demonic Donut]

8 minutes ago, GuruMeditationError said:

Yeah, I assumed that's what he meant...

...but what do you think re. the original question? 

Do you think I'd be okay for air?

I wouldn't spend the money unless you want to go air for some reason. You should be fine to use an air cooler. A D15 is about equal to a 240mm AIO. The H115 might perform slightly better due to it being a 280mm. You should try taming the fan speed first. IMO anything under 70C is fine for temps. As the water temp goes up, the efficiency of heat rejection increases. That's why there are major diminishing returns when fan speeds increase and water loop temps drop.

 

 

Prime95 is also not a realistic load, what are your temps while using the computer the way you do everyday? And 53C is quite cool.

[Bombastinator]

4 minutes ago, GuruMeditationError said:

As per the above post: I want to use the overhead to go with quieter fans. 

I’ve seen tests  that show it both going both ways.  I suspect the difference is small enough that make/model of cooler and fan and testing methodology start to come into play.  

[For Science!]

2 minutes ago, GuruMeditationError said:

I want to switch to a dual-fan heatsink with quiet fans to reduce the noise in my system. I'm hoping to use the overhead to lower the noise.

I thought the info I've provided so far might be enough to go on; is that not the case?

How long should I let it run for to get a sense of where it's stabilising? Bombastinator suggested it could take up to an hour.

Get a program that can log data, even something as simple as MSI afterburner can do it. You see below for a 240 mm AIO (H100i v2) that I used to use, starting from the far left it starts at 31 degrees, spikes up and then gradually reaches a flat line. That is when you know you are at steady state. I think 15-20 minutes should be plenty

 

If you want to reduce your noise, just dial your fan curve back.