Zalman’s CNPS9500 LED CPU cooler

Manufacturer Zalman
Model CNPS9500 LED
Price (Street)
Availability Now
IT’S RARE THAT WE review CPU coolers at TR, but every so often one comes along that we can’t resist. OK, every so often one comes along that I can’t resist. It seems I’ve developed something of a fetish for funky, silent CPU coolers, and although my therapist assures me that there’s nothing wrong with lusting after artfully sculpted heat pipes and cooling fins, I do my best to suppress those urges. I’m only human, though, and when we spotted Zalman’s new CNPS9500 LED at Computex, I couldn’t resist the impulse to procure one for testing.

Zalman has been producing funky CPU coolers for some time, and while some of the company’s designs are outlandish to the point of impracticality, the CNPS9500 LED has been designed with broad compatibility in mind. It’s also been designed to do more with less, and Zalman claims that the cooler offers better performance with a 92mm fan than solutions equipped with 120mm fans. Does the CNPS9500 LED deliver? Join us as we explore the cooler’s exquisite array of heat pipes and cooling fins to find out.

The cooler
I’m not really sure what’s art and what isn’t, but the CNPS9500 has to come close. The cooler is delicately gorgeous, with 90 (yes, I counted) paper-thin copper fins radiating around a translucent 92mm fan. With these looks, it’s almost a shame to bury it inside a case.

Of course, the CNPS9500 has function to back up its form. That function differs slightly from other Zalman coolers, though. Zalman has traditionally mounted CPU cooling fans parallel to the CPU socket, resulting in air flow that’s perpendicular to the processor; however, the CNPS9500 turns that design on its ear, mounting the fan perpendicular to the motherboard to provide airflow parallel to the board. This orientation is intended to more effectively vent warm air from around the CPU through a case’s rear exhaust port.

To best exploit a case’s rear exhaust port, Zalman suggests orienting the CNPS9500 with its fanless side toward the back of the case. That allows the fan to push air from inside the system over the array of cooling fins and out of the case.

While we’re looking at the CNPS9500 from behind, note that its fins radiate from a series of heat pipes that loop up from the cooler’s base in a figure eight. There are three heat pipes in total, and each is independent of the others. Each pipe is linear, so it doesn’t loop back into itself at the base of the cooler.

The CNPS9500’s trio of heat pipes converges in a surprisingly minimal copper base. Zalman does a decent job of polishing the base to remove imperfections that might otherwise impede heat transfer between the cooler and the CPU, but the finish isn’t as mirror-perfect as we’ve seen on other coolers, including some of Zalman’s own. Still, the base is smooth and scratch-free, and Zalman provides thermal compound with the cooler to fill any micro-grooves.

Clearance and mounting hardware
The CNPS9500 looks huge, and it is, but fortunately not in a way that creates clearance problems.


CNPS9500 LED, CPNS7700-AlCu, Stock AMD cooler

Measuring 85mm wide, 112mm long, and 125mm tall, much of the CNPS9500’s size is vertical. It towers over Zalman’s CNPS7700, which is only 67mm tall, and AMD’s stock Athlon 64 cooler is even shorter still.


CNPS9500 LED, CPNS7700-AlCu, Stock AMD cooler

Distributing its surface area vertically gives the CNPS9500 a smaller footprint than the CNPS7700, which measures a whopping 136mm in diameter. This allows the CNPS9500 to avoid the clearance issues that hamper the CNPS7700, making Zalman’s latest creation compatible with a much broader array of motherboards.

We installed the CNPS9500 on a handful of Socket 939 and LGA775 motherboards and didn’t encounter any clearance problems. In all cases, the cooler’s design allowed DIMMs to be installed and removed with ease, and the fins didn’t conflict with north bridge coolers or graphics cards. In fact, the CNPS9500’s motherboard compatibility is so broad that Zalman hasn’t identified any boards that won’t work with this cooler. The same can’t be said for the CNPS7700, whose wide stance creates issues with a number of boards.

While the CNPS9500’s smaller footprint is handy, the cooler’s relatively low weight is perhaps more impressive. Despite its all-copper design, the CNPS9500 weighs in at just 530g—nearly half the weight of the CNPS7700-Cu. Low weight is especially important for the CNPS9500 given its taller profile; one wouldn’t want too much weight tugging on the motherboard when it’s mounted vertically in a tower enclosure.

A relatively small footprint may be responsible for the CNPS9500’s compatibility with a wide range of motherboards, but it’s Zalman’s impressive collection of mounting brackets that allows the cooler to work with a variety of CPU sockets. Zalman ships the CNPS9500 with mounting hardware for sockets 754/939/940, LGA775, and even Socket 478—one size fits all. The mounting brackets are easy to use, and installation is quick and painless for all sockets. LGA775 installation does require that the motherboard be removed from the system so that the bracket’s back plate can be screwed on, though. Previous Zalman coolers have also required back plate installation for sockets 754/939/940, but the CNPS9500 screws right into AMD’s retention bracket.

In addition to mounting hardware, the CNPS9500 also comes with a variable-speed fan controller. The fan controller is optional, and users with motherboards that feature temperature-based fan speed control probably won’t need it. However, manual fan speed control is still a handy feature to have, especially for older systems and Athlon 64 motherboards where temperature-based fan speed control isn’t always a given.

Zalman’s fan speed controller uses an analog knob to dial fan speeds between 1350 and 2600RPM, allowing for plenty of fine-tuning to balance noise levels with cooling performance. Unfortunately, the fan controller uses a three-pin connector, making the CNPS9500 incompatible with the linear fan speed control on some LGA775 motherboards. Many newer LGA775 motherboards provide provisions for fan speed control with both three- and four-pin fans, though.

Before we dive into the CNPS9500 LED’s performance, we should note that the cooler comes with a bright blue LED. Everyone and his mother is rocking blue LEDs these days, and while the novelty wore off long ago for some, only those with case windows will notice the CNPS9500’s glow.

Our testing methods
Today we’ll be comparing the CNPS9500 LED’s performance to that of a stock AMD cooler and to Zalman’s own CNPS7700-AlCu. We’ve confined the bulk of our testing to a Socket 939 platform with an early Athlon 64 FX-53 sample that runs nice and hot. Extended load tests were also conducted with a Pentium 4 660 3.6GHz to confirm that the CNPS9500 LED could indeed handle one of Intel’s hottest Prescott chips without throttling. The CNPS9500 had no problem handling the Prescott processor, and even after hours of a combined Folding@home and Prime95 load with the cooler’s lowest fan speed setting, the CPU temperature refused to budge over 65C.

We used the following test system.

Processor AMD Athlon 64 FX-53 2.4GHz
System bus HyperTransport 16-bit/1GHz
Motherboard DFI LANParty UT NF4 Ultra-D
BIOS revision NF4LD209
North bridge NVIDIA nForce4 Ultra
South bridge
Chipset drivers ForceWare 6.66
Memory size 1GB (2 DIMMs)
Memory type OCZ PC3200 EL Platinum Rev 2 DDR SDRAM at 400MHz
CAS latency (CL) 2
RAS to CAS delay (tRCD) 2
RAS precharge (tRP) 2
Cycle time (tRAS) 5
Hard drives Seagate Barracuda 7200.7 NCQ SATA
Audio nForce4/ALC850
Audio driver Realtek HD 1.24
Graphics NVIDIA GeForce 6200 with ForceWare 77.77 drivers
OS Microsoft Windows XP Professional
OS updates Service Pack 2, DirectX 9.0c

Our test system was powered by OCZ PowerStream power supply units. The PowerStream was one of our Editor’s Choice winners in our latest PSU round-up.

We used the following versions of our test applications:

The test systems’ Windows desktop was set at 1280×1024 in 32-bit color at an 85Hz screen refresh rate. Vertical refresh sync (vsync) was disabled for all tests. Most of the 3D gaming tests used the Medium detail image quality settings, with the exception that the resolution was set to 640×480 in 32-bit color.

All the tests and methods we employed are publicly available and reproducible. If you have questions about our methods, hit our forums to talk with us about them.

Noise levels
Noise levels were measured with an Extech 407727 digital sound level meter placed one inch from the top of the motherboard and out of the path of direct air flow. We tested the Zalman coolers with their highest and lowest fan speed settings to illustrate the range of noise levels one can expect. To ensure consistent results, the motherboard’s temperature-based fan speed control was disabled.

The CNPS9500 is about a decibel quieter than Zalman’s CNPS7700 with both the high and low fan speed settings. Interestingly, though, both Zalman coolers are louder than a stock AMD cooler at their highest fan speeds.

CPU temperatures
To test the CNPS9500’s cooling performance, we subjected our test system to a 30-minute test that consisted of ten minutes of idling, followed by ten minutes of Prime95’s maximum heat torture test, followed by ten minutes of idle cooldown. CPU temperatures were logged every second through the entire test.

To make things more difficult for the coolers, we tested with Cool’n’Quiet disabled. We also conducted extended load tests to ensure that our ten-minute Prime95 torture test was long enough for the processor to hit its peak temperature. Tests were conducted on an open test bed with an ambient room temperature of 24C.

We’re graphing geeks, so we couldn’t resist posting a graph of CPU temperatures over the length of our test. The graph isn’t all that easy to read, though, so we’ve busted it in two to isolate cooling performance with Zalman’s high and low fan speed settings. Note that the following graphs don’t have a baseline of zero degrees on the Y axis; we’ve just tried to make them a little easier to read.

Interestingly, while the CNPS9500 is quicker to cool down with both fan speed settings, it heats up faster than the CNPS7700 on the low fan speed setting. The graphs are still a little muddled, though, so we’ve whipped up some average results for the last minute of the idle, load, and cooldown periods.

Although the CNPS9500 can’t quite match the CNPS7700’s temperatures at the highest fan speed setting, the 9500 yields lower temps during the idle, load, and cooldown periods at the lower fan speed setting. The CNPS9500 also beats the stock AMD cooler across the board.

Conclusions
The CNPS9500 has exotic lines and a striking look, but this cooler’s compatibility and performance are its most impressive assets. With the lowest fan speed setting, the CNPS9500 is quieter and cooler than Zalman’s CNPS7700-AlCu, and it’s practically silent compared to a stock AMD cooler. This thing should be able to handle the hottest CPUs around, too. Zalman has validated the cooler for all of AMD’s and Intel’s dual-core processors, and we couldn’t even get a 3.6GHz Prescott to break 65C. But Zalman coolers have traditionally offered good performance and lower noise levels. What makes the CNPS9500 really stand out is the fact that its design allows for broad motherboard compatibility. The cooler’s extremely low weight is a nice touch, too, as is the inclusion of mounting hardware for all modern desktop CPU sockets.

Unfortunately, the CNPS9500 doesn’t come cheap. The cooler is currently selling for between $62 and $80 online, which is significantly more expensive than Zalman’s CNPS7700 series. Of course, the CNPS7700 series weighs more, is a little louder, and has some compatibility issues with a number of motherboards. You get what you pay for.

Whether the CNPS9500 is worth the extra scratch will ultimately depend on how much you value weight and compatibility. The CNPS9500 excels on both fronts, and it’s certainly a superior design to Zalman’s previous offerings. I’m just not sure if my fetish for funky, silent CPU coolers runs deep enough to compel me to drop over $60 on one.

Comments closed
    • Craig P.
    • 14 years ago

    Zeroing the graphs to 0° C doesn’t really make sense — the zero point for temperature is actually at -273.15° C.

      • aerst2
      • 14 years ago

      Actually….zeroing it to room temperature would make the most sense, since that would be perfect performance.

    • Eckre
    • 14 years ago

    Though the ALCU should be the same noise… but not the same temp…humm.

    • Eckre
    • 14 years ago

    Why Why WHY! Why doesn’t anybody review this cooler compared to the 100% copper version 7700? I think it’s a conspiricy. I think the 100% copper version is cooler and just as quiet. Anyone seen a review comparing the 100% copper 7700 to this one at quietest settings?

      • continuum
      • 14 years ago

      IIRC, it’s not much different in performance than to the AlCu…

      $62 though… I’ll stick with a Thermalright SI-120 if I’m going to spend that much money…

    • firestorm02
    • 14 years ago

    Check out my results in the cooling forum.
    §[< https://techreport.com/forums/viewtopic.php?t=34740<]§

    • AmishRakeFight
    • 14 years ago

    Nice review, but it’s definitely not art.

    • Vertigo
    • 14 years ago

    First page of the review instantly crashes Firefox 1.0.7 on Ubuntu Linux, even with the flash plugin removed.

    Edit: working now – thanks!

    • d0g_p00p
    • 14 years ago

    I leave my machine running 24/7. It seems to me like more wear and tear powering it up every time i want to use it.

      • sbarash
      • 14 years ago

      I have a Sonic Tower. It does come with clips allowing a 120mm fan, if you want. I can overclock the hell out of my Athlon without adding a fan. Needless to say, I love it. With my fanless 6800, and only a couple of 120mm case fans, I’m running pretty silent.

      -Stephen

      • ludi
      • 14 years ago

      I’ve also got one, which I purchased after seeing it in a friend’s Athlon64 box; as long as you have decent case ventilation, it works. In his computer he has two very quiet 120mm fans (one front, one back), and a fan on the chipset. PSU is passive and the SonicTower is passive. You can’t hear anything except the faint singing of four SATA hard disks.

      In my computer, the SonicTower holds a Barton 2800+ at around 50C with three 80mm case fans (one front, two back), a slot-vent fan (mostly to keep the 9800XT from overheating) and the fan in the PSU all running. Not bad, considering how much heat being produced in there. The only issue I ran into is that my chipset has a passive heatsink, and once I removed the CPU fan from the equation, the loss of ventilation around the chipset was just enough to cause ocassional lockups. I worked around that by creating a duct from a piece of polycarbonate plastic.

    • flip-mode
    • 14 years ago

    First: Thank you for this review. Like Usa, I appreciate the HSF reviews. And I’ve been curious about this HSF since Zalman started talking about it.

    Second: Like FlyingFox, I was surprised to see a K8 get so hot, but you did remark that the FX53 was one of the hottest.

    Third: I was also suprised that the cooler didn’t do better! It bested the stock cooler by only 5deg at load, and that on high. And that was it’s largest margin. Like muntjac I saw the Xbit piece a while back and held this cooler in higher regard until your review. I feel like Vrock – it simply isn’t worth the price, and if you actually want a temp drop over the stock HSF, then you have to deal with *[

      • ludi
      • 14 years ago

      Keep in mind that there can be other benefits to such a configuration, though, such as improved heat evacuation by the case/PSU fans and reduced component heating around the CPU socket.

    • Flying Fox
    • 14 years ago

    I thought the K8 CPUs never get over 70 degrees. And yet the benchmarks had some that approached that?

      • MagerValp
      • 14 years ago

      They disabled Cool’n’Quiet, making the CPU run a lot hotter than usual. 65 degrees is fine though, as long as it doesn’t go much higher.

        • Dissonance
        • 14 years ago

        Cool’n’Quiet isn’t going to lower temperatures under a sustained load. We just happen to have an early, toasty, FX-53.

    • Koly
    • 14 years ago

    Geoff, I ‘d like to know how the RPM varied on the stock AMD cooler (it’s temperature controlled, right?) during testing and especially what was the RPM when noise level was measured. Thanks.

    • Vrock
    • 14 years ago

    The real question is this: how much money is a few db of quiet worth to you? Not $62, at least not for me.

    The stock cooler does just fine, it comes with your CPU (assuming you bought retail), and it is pretty quiet to boot, too.

    • Usacomp2k3
    • 14 years ago

    BTW, I really love the HS reviews. It is what caused me to buy the vf700-cu for my video card. Keep up the good work.

      • Shintai
      • 14 years ago

      I love that cooler too on my x800XL 😀

      But why all the damn LEDs on new coolers! They might be cool if I was 10years old….

        • Vrock
        • 14 years ago

        My ex-wife used to call computers with excessive LEDs “Mexican computers”…if you’ve ever seen Latinos cruise by in a 1977 Caprice Classic, you know why.

          • sbarash
          • 14 years ago

          uh, maybe in the states, but go to Mexico and you’ll see that this is not a Mexican thing.

          Better say ‘Mexican American computers’…

          • CasbahBoy
          • 14 years ago

          I guess thats an equivalent of the whole “ricer” thing – its hardly confined to latinos or asians here in northern Ohio, believe me. Mostly upper-middle class suburbanite white kids that like looking and sounding fast better than actually /[

            • Vrock
            • 14 years ago

            Yeah, ricing is something totally different, but equally stupid.

      • CasbahBoy
      • 14 years ago

      TR’s VF7-Cu review is exactly what convinced me to buy mine, too! I know the review put more (?) emphasis on the less expensive and identical performing AlCu, but I had to have it match my CNPS7000A-Cu…

    • Usacomp2k3
    • 14 years ago

    /me hugs my $80 watercooler setup running @ 40 load (read: F@H)

      • Logan[TeamX]
      • 14 years ago

      But my TT Venus 12 and 500MHz OC are also @ 40-42C running F@H… for $30 CDN!

      I’d still love to try watercooling one day.

        • KorruptioN
        • 14 years ago

        With significantly more noise, I would assume. Of all the times I’ve had to play with the Venus 12, I found it excessively noisy.

          • Logan[TeamX]
          • 14 years ago

          Actually it’s not that bad. No doubt the water cooler is much quieter, but the cost and the risk don’t outweigh the noise for me.

        • Usacomp2k3
        • 14 years ago

        My ambient temperature is almost always at least 80F, usually in the high 80’s. At least until winter hits. 8)

Pin It on Pinterest

Share This