G
Guest
Guest
I was rather skeptical as to how one could do a testing of such massive number of coolers in a relatively short time.. and my worries have surfaced now. Not only that, the whole article seems to put down Zalman coolers more than enough, going so far as to put misinformation. Here is a copy of the e-mail I sent to the author.
--Start--
Hello!
I came across your cooler benchmark today, and I'm rather disappointed by the results posted. First things first, I have connections to Zalman Tech's labs, and I was involved in decisions regarding shipping sample products to review sites, including yours. Therefore, I may be partial to the company. However, I would have to question as to why the article seem to virtually dismiss Zalman products as 'shoddy' in general, going so far as to misinform the readers.
For one, I doubt your testing methodology was correct. Your benchmark shows Swifttech product showing at an incredible result of 30C while Zalman's best goes for 44C, but this does not correlate with some other benchmarks I've seen. FrostyTech, with their synthetic testing method, for one. http://www.frostytech.com/articleview.cfm?articleid=680&page=8 They are only around 1 1/2 degrees better in the 'small block' test, which should be analogous to today's flip-chip design. In this, Zalman products hold up very well, not a middle-of-the-road performer. On a related note, testing from Tokyo University put Zalman's CNPS3100 on top. http://www.zalman.co.kr/news/20010514.htm
I do believe the problem stems from the fact that the location of the temperature probe, which is inevitably externally attached because of using SocketA, would produce inconsistent results. A certain wind turbulence could throw off the results substantially. No matter how good a Lab you used may be, bad testbed / methodology would just ruin it.
Also, the noise level measuring leaves something to be desired. Zalman coolers are nowhere near 46dB you're quoting, as far as recent testing at Zalman labs are concerned. Then again, other coolers were measured to be much higher, too. It seems that your results are somewhat 15dB farther off in general. It seems that you did not remove other noise generating factors such as conventional power supplies and hard disks that interfere with measurements. Again, I'm suspecting bad testbed here. What's more, you failed to note what noise level it is WITHOUT a cooler/fan attachment. Casually noting in dB levels is far from being accurate and contains less meaning. Did you know, that adding two noise sources do not yield a linear increase in dB level because dB is a logarithmic scale? With a separate sound source other than the cooler/fan, you cannot accurately compare the noise.
Speaking of noise levels, you should've provided the noiseless mode of the Zalman coolers, too, if you were so concerned about noise level insofar as to test the levels separately.
And, it seems strange that you mentioned the fans on Zalman coolers to be 'no name'. Zalman ships its name-branded fans with the heatsink, so you should've mentioned it as Zalman, like what you did with GlobalWin. Depending on the model, some fans are OEM'ed from Adda or Sunon, so you could've asked Zalman directly for this information if you were going to make a chart on the fan names like that.
"We were somewhat surprised to discover that Zalman, a company based in South Korea, uses a PC case fan from the People's Republic of China. This continues the claim of passive cooling ad absurdum. The fan is included with the models sporting fan-shaped heat sinks. "
I have no idea why you should dismiss Chinese fans so easily here, though. Unless you have a good proof that PRC fans have a substantial problem in performance or lifespan, you should post it along with the claim. Otherwise, it's just wrong and should be removed.
It's also ridiculous that you devoted a whole page to 'Zalman Coolers: Poor Manufacturing Quality'. You are claiming that the fan is of 'low' quality, yet it gives off the impression that the whole package in general is of poor quality, which is quite contrary to truth. Have you taken a careful look at the milling on the bottom of the heatsink? Did you not notice how the fan bracket is made so that it won't cut your finger when installing, etc.
That's not all.
"One thing that all the Zalman coolers have in common is a chintzy-looking CPU clamp, which proves to be not very durable. Not only that, but an incorrectly designed spring keeps this cooler from pressing down firmly enough. To add insult to injury, the burrs on the CPU clamp weren't shaved off properly. The result? Very sharp edges and the inevitable injury. "
Really? I had no such problems with the clamp my CNPS3100-G. Is it so 'undurable' to you? I never saw it break. The pressure is firm enough to hold down the CPU, contrary to your claims. If you really want to prove that it's not enough, get a force measurement unit and check if it's under AMD's recommended parameters. It's not 'incorrect', as it should fit a normal socket fine without installation troubles as it is designed to be so. It also puts down the force in a single-contact manner, which should be suitable to today's small core CPUs in contrast to dual-contact type, which shipped with old versions of Alpha PAL6035, which is better for flat ceramic surface old CPUs had. My clamp also did not have sharp edges. And it's something I bought for full price in retail!
You also have mis-information on the Zalman products' introduction as well. CNPS3100 is NOT a Cu-Au alloy; it's made of pure copper. CNPS3100-G does NOT have 0.3mm thick gold coating, as the FINS THEMSELVES are 0.3mm thick. you could've used a micrometer to check that yourself, or caculate the thickness by measuring the base that stems off the fins, then dividing by number of fins.
All in all, it's a very disappointing and misleading information that I normally don't expect from a 'respectable' site like Tom's Hardware Guide. I hope you have a very good explanation for this, for you'd need an apology to the readers as well as Zalman's executives unless you do.
Wesley Chung / WDSoft.com (http://www.wdsoft.com)
Order/Support Assistance for Zalman Tech.
--End--
--Start--
Hello!
I came across your cooler benchmark today, and I'm rather disappointed by the results posted. First things first, I have connections to Zalman Tech's labs, and I was involved in decisions regarding shipping sample products to review sites, including yours. Therefore, I may be partial to the company. However, I would have to question as to why the article seem to virtually dismiss Zalman products as 'shoddy' in general, going so far as to misinform the readers.
For one, I doubt your testing methodology was correct. Your benchmark shows Swifttech product showing at an incredible result of 30C while Zalman's best goes for 44C, but this does not correlate with some other benchmarks I've seen. FrostyTech, with their synthetic testing method, for one. http://www.frostytech.com/articleview.cfm?articleid=680&page=8 They are only around 1 1/2 degrees better in the 'small block' test, which should be analogous to today's flip-chip design. In this, Zalman products hold up very well, not a middle-of-the-road performer. On a related note, testing from Tokyo University put Zalman's CNPS3100 on top. http://www.zalman.co.kr/news/20010514.htm
I do believe the problem stems from the fact that the location of the temperature probe, which is inevitably externally attached because of using SocketA, would produce inconsistent results. A certain wind turbulence could throw off the results substantially. No matter how good a Lab you used may be, bad testbed / methodology would just ruin it.
Also, the noise level measuring leaves something to be desired. Zalman coolers are nowhere near 46dB you're quoting, as far as recent testing at Zalman labs are concerned. Then again, other coolers were measured to be much higher, too. It seems that your results are somewhat 15dB farther off in general. It seems that you did not remove other noise generating factors such as conventional power supplies and hard disks that interfere with measurements. Again, I'm suspecting bad testbed here. What's more, you failed to note what noise level it is WITHOUT a cooler/fan attachment. Casually noting in dB levels is far from being accurate and contains less meaning. Did you know, that adding two noise sources do not yield a linear increase in dB level because dB is a logarithmic scale? With a separate sound source other than the cooler/fan, you cannot accurately compare the noise.
Speaking of noise levels, you should've provided the noiseless mode of the Zalman coolers, too, if you were so concerned about noise level insofar as to test the levels separately.
And, it seems strange that you mentioned the fans on Zalman coolers to be 'no name'. Zalman ships its name-branded fans with the heatsink, so you should've mentioned it as Zalman, like what you did with GlobalWin. Depending on the model, some fans are OEM'ed from Adda or Sunon, so you could've asked Zalman directly for this information if you were going to make a chart on the fan names like that.
"We were somewhat surprised to discover that Zalman, a company based in South Korea, uses a PC case fan from the People's Republic of China. This continues the claim of passive cooling ad absurdum. The fan is included with the models sporting fan-shaped heat sinks. "
I have no idea why you should dismiss Chinese fans so easily here, though. Unless you have a good proof that PRC fans have a substantial problem in performance or lifespan, you should post it along with the claim. Otherwise, it's just wrong and should be removed.
It's also ridiculous that you devoted a whole page to 'Zalman Coolers: Poor Manufacturing Quality'. You are claiming that the fan is of 'low' quality, yet it gives off the impression that the whole package in general is of poor quality, which is quite contrary to truth. Have you taken a careful look at the milling on the bottom of the heatsink? Did you not notice how the fan bracket is made so that it won't cut your finger when installing, etc.
That's not all.
"One thing that all the Zalman coolers have in common is a chintzy-looking CPU clamp, which proves to be not very durable. Not only that, but an incorrectly designed spring keeps this cooler from pressing down firmly enough. To add insult to injury, the burrs on the CPU clamp weren't shaved off properly. The result? Very sharp edges and the inevitable injury. "
Really? I had no such problems with the clamp my CNPS3100-G. Is it so 'undurable' to you? I never saw it break. The pressure is firm enough to hold down the CPU, contrary to your claims. If you really want to prove that it's not enough, get a force measurement unit and check if it's under AMD's recommended parameters. It's not 'incorrect', as it should fit a normal socket fine without installation troubles as it is designed to be so. It also puts down the force in a single-contact manner, which should be suitable to today's small core CPUs in contrast to dual-contact type, which shipped with old versions of Alpha PAL6035, which is better for flat ceramic surface old CPUs had. My clamp also did not have sharp edges. And it's something I bought for full price in retail!
You also have mis-information on the Zalman products' introduction as well. CNPS3100 is NOT a Cu-Au alloy; it's made of pure copper. CNPS3100-G does NOT have 0.3mm thick gold coating, as the FINS THEMSELVES are 0.3mm thick. you could've used a micrometer to check that yourself, or caculate the thickness by measuring the base that stems off the fins, then dividing by number of fins.
All in all, it's a very disappointing and misleading information that I normally don't expect from a 'respectable' site like Tom's Hardware Guide. I hope you have a very good explanation for this, for you'd need an apology to the readers as well as Zalman's executives unless you do.
Wesley Chung / WDSoft.com (http://www.wdsoft.com)
Order/Support Assistance for Zalman Tech.
--End--