Thermalright Ultra 120 Extreme: Is More Better?
by Wesley Fink on March 7, 2007 12:05 AM EST- Posted in
- Cases/Cooling/PSUs
Overclocking
In the review of the Thermalright Ultra 120, which matched the 3.90GHz highest overclock of the best tested coolers, we concluded that the maximum stable overclock of this X6800 processor on air was likely 3.90GHz. This looked reasonable considering the dual-fan Scythe Infinity, the Tuniq Tower 120, and the Thermalright Ultra 120 all topped out at 3.90GHz. Other heatpipe towers all topped out below this threshold, with the better coolers very close to the 3.90GHz plateau.
We were wrong. Better air cooling was possible and we are seeing what that cooling can do for overclocking in the performance of the Ultra 120 Extreme. It is all about cooling, and as cooling solutions do a better job of keeping the CPU at a lower temperature, then it is reasonable to expect the overclocking capabilities of the CPU will increase.
In each test of a cooler we measure the highest stable overclock of a standard X6800 processor under the following conditions:
CPU Multiplier: 14x (Stock 11x)
CPU voltage: 1.5875V
FSB Voltage: 1.30V
Memory Voltage: 1.90V
nForce SPP Voltage: 1.35V
nForce MCP Voltage: 1.7V
HT nForce SPP <-> MCP: Auto
Memory is set to Auto timings on the 680i and memory speed is linked to the FSB for the overclocking tests. This removes memory as any kind of impediment to the maximum stable overclock. Linked settings on the 680i are a 1066FSB to a DDR2 memory speed of DDR2-800. As FSB is raised the linked memory speed increases in proportion. The same processor is used in all cooling tests to ensure comparable results.
The Thermalright Ultra 120 Extreme with the Scythe S-Flex fan set a new overclocking record by maintaining a stable 3.94GHz. This is the highest stable overclock we have reached with air cooling with this CPU. We could boot at speeds as high as 4.0GHz, but gaming performance was not stable at speeds higher than 3.94GHz. Increasing the CPU voltage also did not increase stability above 3.94GHz, as 1.60V was the maximum voltage usable without generating random reboots of the system.
The best coolers prior to the Ultra 120 Extreme have topped out at 3.90GHz. The 3.94GHz with the Ultra Extreme was something of a surprise, but it proves again that if you move enough air and use enough heatpipes in a good design you can further improve air cooling. Eventually the limit will be reached with air cooling, since it is not an active cooling solution. Air cooling cannot cool lower than the ambient temperature of the room, and it will therefore be outperformed by some active cooling solution like TEC or phase-change cooling. Water cooling is also limited by ambient temperature, but liquid is a better medium for absorbing and releasing the heat produced in CPU cooling. This is also part of the principal used in the design of heatpipes that are proving effective in the current crop of heatpipe towers.
At a 3.94GHz maximum overclock the Thermalright Ultra 120 Extreme is the best air cooler we have tested at AnandTech. It is clear Thermalright knew what they were doing in modifying the Ultra 120 design with additional heatpipes.
In the review of the Thermalright Ultra 120, which matched the 3.90GHz highest overclock of the best tested coolers, we concluded that the maximum stable overclock of this X6800 processor on air was likely 3.90GHz. This looked reasonable considering the dual-fan Scythe Infinity, the Tuniq Tower 120, and the Thermalright Ultra 120 all topped out at 3.90GHz. Other heatpipe towers all topped out below this threshold, with the better coolers very close to the 3.90GHz plateau.
We were wrong. Better air cooling was possible and we are seeing what that cooling can do for overclocking in the performance of the Ultra 120 Extreme. It is all about cooling, and as cooling solutions do a better job of keeping the CPU at a lower temperature, then it is reasonable to expect the overclocking capabilities of the CPU will increase.
In each test of a cooler we measure the highest stable overclock of a standard X6800 processor under the following conditions:
CPU Multiplier: 14x (Stock 11x)
CPU voltage: 1.5875V
FSB Voltage: 1.30V
Memory Voltage: 1.90V
nForce SPP Voltage: 1.35V
nForce MCP Voltage: 1.7V
HT nForce SPP <-> MCP: Auto
Memory is set to Auto timings on the 680i and memory speed is linked to the FSB for the overclocking tests. This removes memory as any kind of impediment to the maximum stable overclock. Linked settings on the 680i are a 1066FSB to a DDR2 memory speed of DDR2-800. As FSB is raised the linked memory speed increases in proportion. The same processor is used in all cooling tests to ensure comparable results.
The Thermalright Ultra 120 Extreme with the Scythe S-Flex fan set a new overclocking record by maintaining a stable 3.94GHz. This is the highest stable overclock we have reached with air cooling with this CPU. We could boot at speeds as high as 4.0GHz, but gaming performance was not stable at speeds higher than 3.94GHz. Increasing the CPU voltage also did not increase stability above 3.94GHz, as 1.60V was the maximum voltage usable without generating random reboots of the system.
The best coolers prior to the Ultra 120 Extreme have topped out at 3.90GHz. The 3.94GHz with the Ultra Extreme was something of a surprise, but it proves again that if you move enough air and use enough heatpipes in a good design you can further improve air cooling. Eventually the limit will be reached with air cooling, since it is not an active cooling solution. Air cooling cannot cool lower than the ambient temperature of the room, and it will therefore be outperformed by some active cooling solution like TEC or phase-change cooling. Water cooling is also limited by ambient temperature, but liquid is a better medium for absorbing and releasing the heat produced in CPU cooling. This is also part of the principal used in the design of heatpipes that are proving effective in the current crop of heatpipe towers.
At a 3.94GHz maximum overclock the Thermalright Ultra 120 Extreme is the best air cooler we have tested at AnandTech. It is clear Thermalright knew what they were doing in modifying the Ultra 120 design with additional heatpipes.
Facebook
Twitter
RSS
54 Comments
View All Comments
Ender17 - Wednesday, March 7, 2007 - link
I'm basing my statements off of SPCR's results.You can read their testing methodology here:
http://www.silentpcreview.com/article687-page1.htm...">http://www.silentpcreview.com/article687-page1.htm...
Wesley Fink - Wednesday, March 7, 2007 - link
While we respect the quest for a silent PC, testing fans in a foam block isolated from power sources is not real world. It does allow isolation of the lowest possible noise that can be emitted by a component. However, in a system power supplies do generate noise, cases do vibrate with mounted fans, and the video card does have a fan.You can minimize all these variables in a specialized PC that is not overclocked,but many users want a system that is very competent, reasonably quiet, but still uses a power supply with a fan. That makes the PS the noise floor. The configuration (open/closed cases), measurement distance, and measurement method determines the dbA level. Our noise measuremtns aim at measuring a real world computer enviromment and they do not isolate the PS in another room for noise measurement. They should also be considered worst case noise in the cooler being tested.
Our test room has all other equipment turned off and only incandescent lights.
PCTechNow - Wednesday, March 7, 2007 - link
If you do not isolate the components for testing then why measure it all? There are so many variables within the case and your room that any measurements provided are worthless.It would be nice to see how these air coolers compare to water systems. Why is there not a review or at least a comparison in your results?Wesley Fink - Wednesday, March 7, 2007 - link
We measure and report total system noise with each cooler using a standardized test system that is typical of an enthusiast system. We have not evaluated individual fans.PCTechNow - Wednesday, March 7, 2007 - link
Is the noise level in your room and case always the same between test measurements? If you do not isolate the test parts and ensure the room noise level is equal then how are your results valid? I am not a gamer so I expect to have a quiet system. I do not think you are providing numbers that cater to the majority of people who use computers. It is hard to tell from any of your results if a unit is really quiet or not. The power supply is already at 38db so any cooler that is quieter than this will not be reported. Is there anyway to tell us if a cooler is quieter than 38db?Wesley Fink - Wednesday, March 7, 2007 - link
The noise level in my test room is always below the noise floor of the single standardized test system with the case side open. I measure the room noise each time before I begin system/cooler noise measurements.There is already a web site for users who are obsessed with system noise, and we are not a substitute. We are trying to address users who care about system noise and want to build a quiet PC on a relative basis - not those who make noise their only concern in choosing computer components.
Manufacturers are definitely doing a better job of addressing noise these days - mainly because users like our reders complain about high noise components and stop buying them. There are not many 62db fans around like the screamers that used to be common on Socket A builds.
I wish the northbrdge coolers were quieter - that is where the highest noise is these days. The Northbridge fan on the test EVGA 680i is the noisiest component in our system. It is so gbad we have to disconnect it before making any noise measurements on coolers.
PCTechNow - Wednesday, March 7, 2007 - link
What is your problem with not using quieter components in the test PC or even changing rooms? Fine with not catering to quiet pc users but at least get your noise floor down to something reasonable like 30db. I am sure I am not in the minority here but most of the boards used now are passively cooled so the two main noise issues in a pc are the cooler and power supply. I am not after total silence but having something that is not louder than the ambient noise in the room is important.gramboh - Thursday, March 8, 2007 - link
The review is fine for users like me, those that are going to build a PC, want it to be REASONABLY quiet, but also overclock and have a high end video card for gaming. If you browse the AT forums, a lot of the users fall into this bucket. Yes there are also silence enthusaists, HTPC etc. If I were building one of those systems I would be reading up on SPCR.The review is gauging an overall end user experience.
kmmatney - Wednesday, March 7, 2007 - link
"I am not after total silence but having something that is not louder than the ambient noise in the room is important."That doesn't really make sense - anything lower than the ambient noise will be pretty much silent. If you want a silent cpu, then go fanless. Most people have a power supply that makes noise, as well as video cards that make noise, and these reviews make sense.
bigpow - Wednesday, March 7, 2007 - link
I was absent for a one year and now I never see it being compared anymore..How is it compared to today's top coolers?
Is Scythe Infinity the new Ninja/p?