avengerx

This is a long-time answer but maybe this helps other owners of H45 corsair cooler. I have one of this and, maybe it is another revision, because no matter how I change the PWM, the pump jack always reads around 4300 RPM. So the pump probably got designed differently to allow (or ignore) it to always run at the same pumping speed -- or maybe it just always inform the same rpm although it could be running slower, just to mask the fact it works where it shouldn't be run behind 100% power. This result and the feedback from this post suggest the pump should always go at 100%. But I have been running the pump at "silent" and "normal" BIOS power policies for a long, long time, without signs of it dying or performing badly. No matter at 0% workload (minimal pwm input) or 100%, the noise coming from the assembly is the exact same (and I can't tell there's noise coming from the pump) and the RPM reading is still around 4300RPM. Notice though my H45 informs a ~4300RPM while the report here states 4500RPM, so there are most likely "revision changes" between mine and thread OP's H45s.

Hi! After years owning a H45 I noticed my mistake with the cooler. Actually I had a two-fan-mount behind (stock) and in front of the radiator, so never really had issues with that. Thing is, I always thought the pump would pump more or less with the workload, what is not the case. Well, in short, if you mean to reduce noise and run the fans at lowest speeds while temps are also low, and accordingly pick up as the CPU heats, you should do this: - pump power plug: connect o a CHASSIS_FAN (or SYS_FAN) and set it to full speed, always. According to post #7 in H45 pump Noisy thread, corsair support recommends it to be always at 100%. - 12" fan power plug: connect to the CPU_FAN it should go. This way you will get the fan blowing air according to the CPU temperature dissipation and will be in-check with corsair (reportedly) recommendation for our hardware. If you just set all fans to 100% in BIOS then this information is irrelevant to you -- as long as your fan and pump are at 100% always, there's not improvement (other than noise level and maybe fan life span). This reported recommendation from corsair is congruent to the RPM values I get from the pump as I calibrate the workload-per-rpm values in gigabyte's System Information Viewer Smart Fan Advanced feature. I believe I worked with the wrong set up for a loong time! Thought the "pump picked up" with the workload input to the pump power jack until recently I removed an extra 12" fan I had in the other side of the radiator. :)

That's interesting what you discuss here, I also got an H45 one for a long time. Also on a gigabyte board (GA-Z97M-D3H) and just installed the SIV to check some fan speed issues I had (different subject). Tried then changing the RPM forcibly pulling it to low RPM and then to high RPM. No noise difference (I don't really notice any annoying noise from the pump!). Interesting thing is the result when I run its calibration, I get the following table of RPM-per workload: No code has to be inserted here. The pump is plugged at CPU_FAN jack (yes, I'm going to fix that) and the fan (and other fans) are plugged to the different SYS_FAN jacks throughout the motherboard, and they all obey the workload-versus-rpm settings; I'm not sure whats up, but it looks as if the H45 has some "masking" RPM display and somehow maintains its pumping no matter what PWM (workload) we apply to the pump jack. EDIT: After taking out the cooler, giving the old radiator and fan a good vacuum and cleanup, then plugging all back to the right places, then re-running gigabyte's smart fan another calibration, I get progressive RPM rates from the pump. Hopefully this translates into fluid flow in the pump somehow. So, updated table became: No code has to be inserted here. Yet, corsair recommendation should be followed, I believe. Taking the pump to lower values (or turning it off) corretly triggered motherboard alarm (for 0rpm) and also temperature picked up at low pump rates, so the pump really goes "slower" as we push the gauge down. I don't know, but I really feel compelled to let the pump work at least at half power when the computer is idle -- yet, must bear in mind pump power will be tied to system temperature, not CPU's.

I have been running the fans in the wrong jacks for a long looong time!.. I only noticed it today cause CPU hit 80C and the alarm noise was up while fan was sitting at its puny 20% power. Only then I noticed I had it plugged wrong; how was sys_fan2 to scale to cpu temp? lol. But the missing bit was then, where should I plug the pump? And this thread confirmed it. As I ran gigabyte's "smart fan" bit of its "SIV" tool (system information viewer), where it "calibrates" the fans, showing rpm by workload%, I found the pump's speed informed RPM ranged about 4300RPM no matter what the workload was. I.e.: The calibration gave a reading of 4299 RPM in 0, 10, 30, 50, 60, 70 and 90% workload settings, and 4327RPM at the remaining 20%, 40%, 80% and 100%. This means the reading the pump sends is "faked" to just indicate the pump is operational or not, no matter how the workload is input to the power jack. Now shutting down the system to invert the plugs -- and set full speed to the connector I place the pump in. :) I have set up alarms for stopped fans and overburdened CPUs but they never really sounded unless in extreme cases like EDIT: in the past I didn't have issues with overheating (mostly) because I had another 12" fan in front of the radiator (so, the original fan in one side, and another fan at the other side of the radiator), and the air pull was enough for most situations. I always thought the pump picked up the pace according to workload (PWM) applied to its jack. Even if it that's the case, it seems it is not projected to that, as corsair "masks" the RPM reading to pretend it is always at maximum. Manually changing PWM/workload to the pump did not alter noise levels.