Thermalright Ultra 120: Elegant, Fanless and Efficient Cooling
by Wesley Fink on March 5, 2007 12:05 AM EST- Posted in
- Cases/Cooling/PSUs
CPU Cooling Test Configuration
The standard test bed for cooling tests is the EVGA NVIDIA 680i SLI motherboard. This is primarily based on the consistent test results on this board and the excellent NVIDIA Monitor temperature measurement utility, which is part of the nTune program.
NVIDIA Monitor has a drop-down pane for temperature measurement which reports CPU, System, and GPU results. Reviews at this point will concentrate primarily on CPU temperature. In addition to the real-time temperature measurement, NVIDIA Monitor also has a logging feature which can record temperature to a file in standard increments (we selected every 4 seconds). This allows recording of temperatures during testing and play back, for example, of stress test results that can then be examined when the stress tests are completed. There is also the handy reference of speeds and voltages in the top pane to confirm the test setup.
Other components in the cooling test bed are generally the same as those used in our motherboard and memory test bed:
All cooling tests are run with the components mounted in a standard mid-tower case. The idle and stress temperature tests are run with the case closed and standing as it would in most home setups. We do not use auxiliary fans in the test cooling case, except for the north bridge fan attached to the 680i for overclocking.
Since Thermalright provided a syringe of their own thermal compound, the Thermalright compound was used in mounting the Ultra 120. If the manufacturer does not provide a premium thermal compound with the heatsink we use a standard high quality silver-colored (but no silver content) thermal paste for testing.
In our experience the thermal compound used makes little to no difference in cooling test results. This is particularly true now that processors ship with a large manufacturer-installed heatspreader. Our only control on thermal compound is that we use the manufacturer-supplied product, if they supply a premium product, or a standard high-quality thermal paste if a premium brand is not supplied.
We first tested the stock Intel cooler at standard X6800 speed, measuring the CPU temperature at idle and while the CPU was being stressed. We stressed the CPU by running continuous loops of the Far Cry River demo. The same tests were repeated at the highest stable overclock we could achieve with the stock cooler. Stable in this case meant the ability to handle our Far Cry looping for at least 30 minutes.
The same tests were then run on the cooler under test at stock, highest stock cooler OC speed (3.73GHz), and the highest OC that could be achieved in the same setup with the cooler being tested. This allows measurement of the cooling efficiency of the test unit compared to stock and the improvement in overclocking capabilities, if any, from using the test cooler.
The Thermalright Ultra 120 does not come with a fan, but it is designed to mount one 120mm fan. The fan selected for benchmarking with the Ultra 120 is the Scythe S-Flex SFF21F, a high output and low noise fan with a fan motor using the Sony Fluid Dynamic Bearing. The Ultra 120 was also tested without a cooling fan.
Noise Levels
In addition to cooling efficiency and overclocking abilities, users shopping for CPU cooling solutions may also be interested in the noise levels of the cooling devices they are considering. Noise levels are measured with the case open on its side and are measured using a C.E.M. DT-8850 Sound Level meter. This meter allows accurate sound level measurements from 35bdB to 130dB with a resolution of 0.1dB and an accuracy of 1.5dB. This is sufficient for our needs in these tests, as measurement starts at the level of a relatively quiet room. Our own test room, with all computers and fans turned off, has a room noise level of 36.4dB.
Our procedures for measuring cooling system noise are described on page seven along with noise results comparing the stock Intel cooler and recently tested CPU coolers to the Thermalright Ultra 120 coolers.
The standard test bed for cooling tests is the EVGA NVIDIA 680i SLI motherboard. This is primarily based on the consistent test results on this board and the excellent NVIDIA Monitor temperature measurement utility, which is part of the nTune program.
NVIDIA Monitor has a drop-down pane for temperature measurement which reports CPU, System, and GPU results. Reviews at this point will concentrate primarily on CPU temperature. In addition to the real-time temperature measurement, NVIDIA Monitor also has a logging feature which can record temperature to a file in standard increments (we selected every 4 seconds). This allows recording of temperatures during testing and play back, for example, of stress test results that can then be examined when the stress tests are completed. There is also the handy reference of speeds and voltages in the top pane to confirm the test setup.
Other components in the cooling test bed are generally the same as those used in our motherboard and memory test bed:
| Cooling Performance Test Configuration | |
| Processor | Intel Core 2 Duo X6800 (x2, 2.93GHz, 4MB Unified Cache) |
| RAM | 2x1GB Corsair Dominator PC2-8888 (DDR2-1111) |
| Hard Drive(s) | Hitachi 250GB SATA2 enabled (16MB Buffer) |
| Video Card: | 1 x EVGA 7900GTX - All Standard Tests |
| Platform Drivers: | NVIDIA 9.53 |
| NVIDIA nTune: | 5.05.22.00 (1/16/2007) |
| Video Drivers: | NVIDIA 93.71 |
| CPU Cooling: | Thermalright Ultra 120 Scythe Infinity Zalman CNS9700 Zalman CNS9500 Cooler Master Hyper 6+ Vigor Monsoon II Lite Thermalright MST-9775 Scythe Katana Tuniq Tower 120 Intel Stock HSF for X6800 |
| Power Supply: | OCZ PowerStream 520W |
| Motherboards: | EVGA nForce 680i SLI (NVIDIA 680i) |
| Operating System(s): | Windows XP Professional SP2 |
| BIOS | Award P24 (1/12/2007) |
All cooling tests are run with the components mounted in a standard mid-tower case. The idle and stress temperature tests are run with the case closed and standing as it would in most home setups. We do not use auxiliary fans in the test cooling case, except for the north bridge fan attached to the 680i for overclocking.
Since Thermalright provided a syringe of their own thermal compound, the Thermalright compound was used in mounting the Ultra 120. If the manufacturer does not provide a premium thermal compound with the heatsink we use a standard high quality silver-colored (but no silver content) thermal paste for testing.
In our experience the thermal compound used makes little to no difference in cooling test results. This is particularly true now that processors ship with a large manufacturer-installed heatspreader. Our only control on thermal compound is that we use the manufacturer-supplied product, if they supply a premium product, or a standard high-quality thermal paste if a premium brand is not supplied.
We first tested the stock Intel cooler at standard X6800 speed, measuring the CPU temperature at idle and while the CPU was being stressed. We stressed the CPU by running continuous loops of the Far Cry River demo. The same tests were repeated at the highest stable overclock we could achieve with the stock cooler. Stable in this case meant the ability to handle our Far Cry looping for at least 30 minutes.
The same tests were then run on the cooler under test at stock, highest stock cooler OC speed (3.73GHz), and the highest OC that could be achieved in the same setup with the cooler being tested. This allows measurement of the cooling efficiency of the test unit compared to stock and the improvement in overclocking capabilities, if any, from using the test cooler.
The Thermalright Ultra 120 does not come with a fan, but it is designed to mount one 120mm fan. The fan selected for benchmarking with the Ultra 120 is the Scythe S-Flex SFF21F, a high output and low noise fan with a fan motor using the Sony Fluid Dynamic Bearing. The Ultra 120 was also tested without a cooling fan.
Noise Levels
In addition to cooling efficiency and overclocking abilities, users shopping for CPU cooling solutions may also be interested in the noise levels of the cooling devices they are considering. Noise levels are measured with the case open on its side and are measured using a C.E.M. DT-8850 Sound Level meter. This meter allows accurate sound level measurements from 35bdB to 130dB with a resolution of 0.1dB and an accuracy of 1.5dB. This is sufficient for our needs in these tests, as measurement starts at the level of a relatively quiet room. Our own test room, with all computers and fans turned off, has a room noise level of 36.4dB.
Our procedures for measuring cooling system noise are described on page seven along with noise results comparing the stock Intel cooler and recently tested CPU coolers to the Thermalright Ultra 120 coolers.
Facebook
Twitter
RSS
30 Comments
View All Comments
hox - Thursday, March 8, 2007 - link
I would like to know what you have enabled or disabled in the bios that controls CPU automatic throttling. Typically that would include Enhanced C1 control (C1E), SpeedStep, and for my Asus MB, something called CPU internal thermal control.These settings affect how the cpu "handles" load and will affect the temps reported to the Ntune program. Programs like CpuZ and Coretemp and Ntune do not detect some of these rapid throttling instances, but the temps of the processor are reduced because of this throttling. The Righmtark CPU temp utility is one of the few programs I know that allows you to visualize this throttling even when you have several of the throttling paramters disabled.
This issue is important because the temps you are seeing, even at idle, are very different for several of your reviewed heatsinks then what end users will experience when they have altered these parameters in the bios.
I believe this will be helpful for end users to feel like the products your reviews are deeming as superior are performing in their hands in a manner similar to your review. Please provide the bios settings to help the end users decide if the equipment they have purchased is functioning properly.
I suspect by doing this, many RMAs and emails to the manufacturer would be reduced.
There is a clear 10C difference between the reported CPU temps at both idle and full load for my QX6700 cpu caused by changing these three settings. Enabling these parameters in the bios I routinely see temps of 33 to 35 C for the cores of my QX6700 at stock speeds, MB temp is 35C. Disabling these parameters, temps on all 4 cores rise to 44C. Thus discrepancy in temps could lead someone to think that the Monsoon II lite cooler I am using is malfunctioning, when in reality it is working properly.
Also it would be helpful if you also tested these devices with the quad core processors which have a higher heat output. By testing the quad core cpus the high end capacity of these heat sinks to move heat would be tested. Providing the thermal resistance Degrees C/Watt would also be a helpful guideling for these devices.
It would also be helpful to comment whether the side door will fit with the thermalright heat sinks. As far as I know this heat sink is taller than the Tuniq which caused end users to move side panel fans and vents.
Thank you
arswihart - Tuesday, March 6, 2007 - link
I think you should consider retesting the Monsoon II Lite. As the only heatsink to make it to 3.96ghz on your OC tests, and with a full 6C increase in temp at 3.90ghz compared to the Ultra 120, I question whether you can replicate your results with the Monsoon at this point. We all know transistors age with OC'ing, and it is possible that your CPU has actually degraded in performance since you tested it with the Monsoon.There's no reason to think that the Monsoon enabled your CPU to reach 3.96ghz by some magical effect apart from it's cooling ability, which is 6C worse that the Ultra 120 at 3.90ghz.
arswihart - Tuesday, March 6, 2007 - link
Sorry my fault I mis-stated the temp difference at 3.90ghz between the two heatsinks, it is actually a paltry 2C!I think it still might be worthwhile to retest the Monsoon, just because the Tuniq Tower and today's Ultra 120 were unable to reach 3.96ghz even with comparable or better cooling at 3.90ghz.
muddocktor - Monday, March 5, 2007 - link
This is a decent review of the Ultra 120 Wesley, good job. But like another person posted, I also had gotten an Ultra 120 with a bad base finish. In my case, there was a machining ridge left on one side of the base that interfered with getting a flush mount plus the base was fairly concave too. And my initial testing of it (with those defects) found the performance to be no better than the stock Intel heatsink. After I filed the ridge down on the edge, then lapped the base of the Ultra 120 to be flat, the performance was much like the performance that you had with yours. I also tested it with both the S-Flex fan you chose plus I tried the Tuniq Tower's fan too(I also have one of those heatsinks) and saw very little difference in cooling ability with either fan mounted on either heatsink. So for the folks that want to swap out the Tuniq Tower's stock fan with the 63 cfm S-Flex you will only lose a minimal amount of performance and the S-Flex is noticeably quieter.Also in testing my Ultra 120, I tested on a socket 939 AMD system too and ran into a mounting orientation problem with my motherboard. Since AMD doesn't use a square mounting pattern on socket 939 (or 754 or 940 or AM2 for that matter), you can't just change the mounting direction by simply rotating the heatsink 90 degrees. Thermalright does make an "S" clip for this though, which I also tried out. The "S" clip is a real PITA to mount with but does do the job effectively and gives you a decent mount.
Finally, with the Ultra 120 you have the ability to mount different thickness fans too, unlike the Tuniq Tower. You can easily use any of the Panaflo 120 X 38 mm fans and if you are brave enough you can even mount the Delta GFB1212VHW 120 X 76 mm fan if you think it would help in cooling (I don't think so as there seems to be a point of diminishing returns on added cooling efficiency with this heatsink past 60-70 cfm with a moderate max static pressure fan).
The biggest problem I see with Thermalright's products lately is the somewhat variable nature of their quality control on base finish. I've also gotten some XP90's in the last year or so with concave base finishes that required lapping too. But other than that, Thermalright has some of the best designs on the market in my opinion.
takumsawsherman - Monday, March 5, 2007 - link
I have been having good success with the HR-01 from Thermalright, so I am glad to see this review which more accurately measures the performance of *this* heatsink versus competitors, which looks even better than the HR-01 in terms of construction quality (the HR-01, despite its good performance, looks unfinished).As far as the stock Intel coolers go, I have been pleasantly surprised by the latest generation of both Intel and AMD stock coolers. Though I recently replaced a customer's late model P4 stock heatsink (LG775 3.04 Prescott) with an HR-01 with Scythe SFF21F because of crazy temps that were causing the system to power down. Tried adjusting the heatsink, and when cleaning and reapplying thermal paste to the stock cooler I noticed that as opposed to the copper core of the C2D coolers, this had a chrome-like metal. After reinstallation, the temperature at idle was still crazy hot, with no apparent reason (Intel board, voltages looked OK, no overclocking, etc.. After installing the HR-01, idle dropped to 32C, full load after 20 minutes was 47C-49C.
As an aside, I am really not digging the new Intel fasteners. There always seems to be one that doesn't want to go in easily, and/or you can't hear the click. Then I go nuts trying to verify whether or not it is really in. Hopefully, AMD will not follow that system.
gramboh - Monday, March 5, 2007 - link
Awesome review, I love the amount of detail and variables involved in the test, very comprehensive and well described.I was set on a Tuniq but after being frustrated by how hard they are to get in Canada, I am going to go with an Ultra 120 instead. My current PC (4 year old P4 2.6 Northwood @ 3.25) has been cooled with a Thermalright SLK-900U (copper) and has been stablewith a 625MHz OC for 4 years. I like the company.
Anyone know if there are any fit issues using the Ultra 120 with a Asus P5B-E in an Antec P180B/P182 case?
Thanks
AbRASiON - Monday, March 5, 2007 - link
Nice article but all it has convinced me is that the intel stock cooler is surprisingly pretty good for "free"It's not too noisy (for a change) it's not holding back overclocking (that much) and it costs 0$
If you're a budget guy, right now an E4300 with the stock cooler is pretty ok for cheap.
Baked - Monday, March 5, 2007 - link
Thank you Anandtech for finally reviewing this HSF. Thermalright is king baby.mforce - Monday, March 5, 2007 - link
I know you specified in your article that an adaptor for AM2 needs to be purchased separately but I don't find this quite normal. I think Thermalright should make an effort and support AM2 out of the box. After all AM2 isn't a new thing , it's been out for quite some time now and it's also here to stay. An extra adaptor costs more and might also be hard to find.Maybe you should ask Thermalright and have an official position as to why it's so hard to support AM2 out of the box. Socket 754 and 939 are cool but they're quite dead.
johnsonx - Monday, March 5, 2007 - link
Are you sure you're describing that correctly? I don't ever recall seeing such a power supply, unless someone screwed up and installed the fan backwards. I agree though that a power supply with a bottom fan would be better suited to cool a fanless heatsink, but because it draws in air FROM the CPU area not blows ONTO the cpu area.