Zalman CNPS 9700 replaced by TRUE

[LennyRhys]

Yes, I bought a black TRUE and a white fan


First things first: let's void the warranty...


After 2500 grit - I know it's unnecessary, but it came in the lapping kit so I used it anyway!


Silverstone FM122, 800rpm - 2400rpm... not everybody's favourite but it moves a lot of air


Side-by side with an old Thermaltake fan


Overhead view showing 32mm depth vs standard 25mm depth


Motherboard and Q9650 ready for installation of TRUE


First installation without thermal paste, solid as a rock, no swaying or twisting, so no need for washer modification etc.


Installed and ready to go into the PC - this combo means business


TRUE + FM122 installed and running:


Integrated fan controller, which conveniently slots inside a floppy drive bay, unlike the crappy Zalman Fanmate 2 (no, I'm not bitter ):



Enough pictures...how does a lapped TRUE compare with the Zalman CNPS 9700 when cooling a Q9650 @ 4.3GHz, 1.368v?

When dealing with a very intense CPU workload at moderately high vcore, the TRUE is more efficient by a margin of

greater than* 15°C.

Zalman small FFTs load: 82°C and climbing, so I had to stop it
TRUE small FFTs load: 67°C

*I say "greater than" because the ambient temperature was significantly higher when I tested the TRUE, so in all probability the TRUE can shear nearly 20°C off the load temps yielded by the Zalman in this scenario.

Note: the performance of the TRUE will vary between CPUs, as will the performance of the Zalman. Bear in mind that with a dual core CPU the difference in load temperatures may not be so pronounced. The intensity of the workload also affects the heat generated by the CPU even at the same voltage, so with larger FFTs (a lighter workload) the difference was closer to 10°C (Zalman 65°C, TRUE 55°C). Not surprisingly, the idle temps are virtually the same with both cooling solutions.

Ironically, the TRUE cost less than the Zalman that it replaced (even though the Zalman was a gift).

[Demo]

What RPMS are you running the FM122 with, anything above 1,200 on that thing for me makes it annoying, i switched to a zalman F3 cooler and my temps dropped by another 2c with reduced noise levels also.

[LennyRhys]

Whilst stress testing I have the fan maxed out at 2400rpm, but for everything else it doesn't need to be above the lowest speed, not really.

I may buy another fan somewhere down the line since a decent 25mm fan costs only about £5, but this one will do for now.

[celemine1Gig]

You really lapped that true on a wooden table? Did you at least check the table's flatness before doing so?
I mean the temperatures look great but that could've been plain luck.

[LennyRhys]

No I didn't lap the cooler on wood lol it's a laminate kitchen unit so it is very flat and also rock hard - this is what I used when I lapped my old Conroe CPU.

I made every effort to get the best temps - I carefully checked the base after lapping, checked the rigidity of the mounting before applying TIM, and I'm happy that there's no movement. I also used hardly any thermal paste (Zalman STG-1) and I'm certain that the effort I put into this setup is what's paid off. I hate to think what the temps would be like if I hadn't lapped it...

Unfortunately I can't further lower my vcore because my system becomes unstable, but at least now I have appropriate cooling.