CLC cooling results.

[hunkeelin]

Setup
Case: Cosair Air 540
CPU: 5960x at 4.3Ghz.
CPU cooler: H110 + 2x cosair sp 140

Temp measurement software: Hw mon

Benchmarking software: Realbench v2 specially the h 264 video encoding test cause that test generate the most heat of all the four test.

idle temp: 24C ~ 27C (some core runs hotter)

Load temp: 67C to 73C

noise: 24db

I am going to have a x61 kraken(might) and h240x in a week and i'll do an update.
Oh, and all the fans will be replaced by noctua nf14 by then as well.
My rig
http://postimg.org/image/7c9ltp9wh/

Realbenchv2 results 5960x at 3.3ghz
image editing: 94895
h264 video encoding: 134471
heaviny multitasking : 129958

(During maximum load)
Lowest core temp at 39C, Highest core temp at 48C

@ 4.3Ghz
image editing: 122185
h264 video encoding: 171728
heaviny multitasking : 157222.

(During maximum load)
Lowest core temp at 67C, Highest core temp at 73C

 

[Essence_of_War]

noise: 24db

How did you measure this?

<30 dB is very hard to measure w/o an anechoic chamber.

 

[hunkeelin]

How did you measure this?

<30 dB is very hard to measure w/o an anechoic chamber.

^ Well, I used two method to double confirm my numbers both of them have a difference of 1~3db
1) my phone with a db app
and
2) logitech microphone with a db measurement software http://www.pinnaclemicro.com/comput...0sGn1iFdQ91W2LaLNvLpommsDNY1wYaXW4xoCvKnw_wcB

 

[Essence_of_War]

Ok, but do you understand what 30dB is?

30dB is a quiet room. Like, a bedroom at night in the country with no fans, no computers, no nothing. I do not believe that you're using your equipment correctly if you think that you're measuring 24dB from an H110 w/ two 140mm fans running on high. You shouldn't be able to get a reading of below 30-32dB or so with the set-up you're describing.

I think what's probably happening is that you're measuring +24dB over the ambient noise level. If you assume a quiet room, that's ~54dB for the H110 which is very roughly conversational speech level. That seems pretty plausible.

 

[aigomorla]

Ok, but do you understand what 30dB is?

30dB is a quiet room. Like, a bedroom at night in the country with no fans, no computers, no nothing. I do not believe that you're using your equipment correctly if you think that you're measuring 24dB from an H110 w/ two 140mm fans running on high. You shouldn't be able to get a reading of below 30-32dB or so with the set-up you're describing.

I think what's probably happening is that you're measuring +24dB over the ambient noise level. If you assume a quiet room, that's ~54dB for the H110 which is very roughly conversational speech level. That seems pretty plausible.

+1

there is no way any high speed fans can do 24db.

 

[hunkeelin]

Ok. So here's what I do I turn off my computer and measure the db via my phone (1 feet away from my computer). The noise fluctuate between 18 to 23db. Then I turn on my computer, the db fluctuate between 30~34. Mind that is broad daylight so it's nosier than at night.

 

[aigomorla]

that doesnt make sense...

u can see martin's dB testing here, and they all start at 30+dB for the low setting quite fans.
Corsair kits do not come with low spin quiet fans.

here is a dB chart:

i am gonna go with essence's assumption, that the software you are using is displaying the delta of dB.

 

[hunkeelin]

^Oh. I switched out the stock fans for sp140 at 1200rpm.

 

[ClockHound]

Skeptical...the self-noise of any cheap smartphone mic cannot be trusted below 30-35dB. And the Logitech mic is a bandwidth limited noise cancelling mic, not suitable for acoustic measurements.

Even the best 'cheap' electret mics have self-noise above 25dB. Have to agree that the software is playing the 'relative' quiet game.

Rent a calibrated B&K meter and get back to us with those results.

 

[BonzaiDuck]

I don't like fan-noise either. I just don't think I'd bother with "measurement."

I've freely touted my rigs with acoustically-muffled AP-30 fans: neither of them need to spin up beyond 3,200 RPM to meet near-maximum cooling effectiveness.

I figure if the most one can hear is the "whoosh" of air-turbulent white-noise, it wouldn't be worse than the room's AC vent in summer. Anything noticeable only occurs when the system is under severe stress-tests -- including the white-noise. I REALLY DO believe that my air-cooled rigs are no noisier than either an AiO or custom-water rig, and I'd probably beef up the fans for an AiO or custom water with some Noctua iPPC or Akasa Viper fans.

It really depends on what load-test you're using, but the temperatures look encouraging for that sort of clock-speed on a 5960X. I think the TCASE spec for those processors is around 66C, and the temperatures you can measure are likely at least a few degrees higher. That is, with that TCASE spec, I think you'd be right at that limit with core temperatures such as yours or even a tad higher.

If you're really set on trying the Swiftech H240X, I'd be very curious and eager to see your results.

 

[hunkeelin]

I don't like fan-noise either. I just don't think I'd bother with "measurement."

I've freely touted my rigs with acoustically-muffled AP-30 fans: neither of them need to spin up beyond 3,200 RPM to meet near-maximum cooling effectiveness.

I figure if the most one can hear is the "whoosh" of air-turbulent white-noise, it wouldn't be worse than the room's AC vent in summer. Anything noticeable only occurs when the system is under severe stress-tests -- including the white-noise. I REALLY DO believe that my air-cooled rigs are no noisier than either an AiO or custom-water rig, and I'd probably beef up the fans for an AiO or custom water with some Noctua iPPC or Akasa Viper fans.

It really depends on what load-test you're using, but the temperatures look encouraging for that sort of clock-speed on a 5960X. I think the TCASE spec for those processors is around 66C, and the temperatures you can measure are likely at least a few degrees higher. That is, with that TCASE spec, I think you'd be right at that limit with core temperatures such as yours or even a tad higher.

If you're really set on trying the Swiftech H240X, I'd be very curious and eager to see your results.

^ is it dangerous for my 5960x for my cores to go above 66C?

 

[hunkeelin]

I updated my post

 

[Simsme]

thanks for the updating

 

[BonzaiDuck]

^ is it dangerous for my 5960x for my cores to go above 66C?

No. There's an ongoing discussion about the general topic, but the consensus seems to be that you can run up the temperatures to the TJunction throttling limit, and the processor will merely shut down to save itself.

However, I believe that Intel still includes the TCASE spec for a reason -- to guide OEM system builders in case design and thermal solutions.

It's also generally accepted that higher temperatures create electrical noise, which means more instability for overclock settings.

Intel spec publications I'd read for earlier processors seemed to define TCASE as a temperature that would be measured if a sensor were planted dead-center between the heat-spreader and the die. Other "white-papers" I'd read at Tom's Hardware suggested that TCASE would correspond to average core-sensor values several degrees above the spec.

I simply use that spec as a target or objective for cooling my overclocked systems.