Case/CPU cooling question (Corsair H90/Define R4)

[BlueBomber]

Hello all!

On Friday I built my new system and while I'm amazed at how quiet it is I've also noticed that CPU temps while gaming seem pretty high (to me anyway).

I'm using the following equipment:
Case: Fractal Design R4
CLC: Corsair H90
CPU: i5 4670K

The way I currently have things set up is 2x 140mm Fractal Design Silent Series R2 fans (that came w/ the case) at the front of the case, and with a single 140mm Noctua NF-P14 FLX attached to the H90's radiator. I'm not sure if I have the fans, # of fans, or direction of the airflow configured optimally, so I'd like some advice.

For reference, the Noctua runs at 1200RPM with an airflow of 64.96 CFM while the Fractals run at ~1000RPM with 66 CFM of airflow. Corsair recommends that the H90's fan blow cool air in, but I currently have the Noctua sucking air out because the two fans in front are (currently) intakes. I could buy another Noctua and create a push-pull config. Right now I think I have more air coming in w/ the Fractals than the Noctua is expelling. I don't know if that's a problem, since I don't know much about the effects of pressure in cooling a case.

Over the weekend I noticed while gaming that the CPU cores could hit the low 70s (Celsius) on occasion, which seems high to me. This is with everything currently at stock. I should say that the ambient temperature in the room was rather hot due to SoCal weather (90F/32C) so I'm sure that contributes, but if anyone has any advice on how to modify my current setup to increase cooling, or if you think the temps I'm seeing are normal, I'd appreciate any feedback.

Thanks!

 

[denis280]

this should bring the cpu down.http://www.newegg.ca/Product/Product...82E16835608040

 

[lehtv]

32C ambient definitely contributes. In more humane 23C ambient, those low 70's become low 60's. Still seems a bit high for a stock clocked CPU, since the H90 is no slouch... Where's the original H90 fan? Would it be more efficient?

Could you test extreme load temperatures with Prime95 small FFT's?

 

[BlueBomber]

Where's the original H90 fan? Would it be more efficient?

Could you test extreme load temperatures with Prime95 small FFT's?

I still have the original fan, but I never bothered with it since I've seen complaints in reviews about the fans that come with the various AIO water cooling units that Corsair sells, but I might as well give it a go. The specs for it are 1500RPM and 94CFM which is significantly more airflow than the Noctuas, but the Corsairs also have a decibel rating of 35dBa vs. the Noctuas 19.6. I think the decibel rating isn't linear either, so that's significantly more noise. Still, no harm in trying it out.

The Prime95 small FFT temps that I got maxed out at 72/75/72/68C (that's each core) with an ambient of 87F/30.5C. Linpack on the other hand resulted in temps of 86/87/86/80C. I swear the other day when I ran Linpack one of the cores hit 100C and AIDA reported throttling, but I can't seem to reproduce that now, although it's quite a bit cooler today.

One thing I noticed, CPU-Z seems to show the clockspeed topping out at 3600MHz when stress testing, but when the CPU load isn't at 100% anymore I'll see it jump up to 3800MHz. Is that something with how Turbo Boost works, if all cores are in use then it maxes at 3.6GHz, but a single core can max out at 3.8GHz?

Anyway, I may try and redo the application of thermal paste this weekend since I have a feeling I may have used too much when I was building the computer. I was a bit rushed during that part.

 

[lehtv]

Yeah, Turbo boost frequency is higher when fewer cores are loaded. I think it's 3.5Ghz for 4, 3.6 for 3, 3.7 for 2 and 3.8 for 1.