Reviewer Tryouts: Thermalright Ultra-120 Extreme
Thermalright Ultra 120 Extreme Review
The Thermalright Ultra-120 Extreme is widely regarded as being one of the top performing air coolers available today. Released in the late spring of 2007, it has already been reviewed numerous times by websites and users. Despite this, certain aspects of the heatsink’s performance and quality were still unclear to me, and my purchase of this cooler was not with 100% confidence.
I hope to share and clarify a few things for people who are considering making the same purchase.
First, I wanted to compare the heatsink’s performance with dual fans in push-pull setup versus a single fan setup on an overclocked setting. Second, I wanted to investigate how the performance of this cooler is affected by fan speed - being somewhat sensitive to noise, I wanted to avoid running a loud fan at 12V on this cooler. Third, I wanted to know what effect varying fan speeds had on push-pull configurations - would running two fans on this cooler at 7V be a comparable (and far quieter) alternative?
Finally, I wanted to examine its quality. Some users have complained of a poorly machined base, of finding it to be not flat. This could result in poorer performance from the cooler, due to worse contact with the heatspreader of the processor. Also, the "scissor" mounting clip has been reported as not providing enough mounting pressure, with the heatsink being able to rotate a bit underneath it.
These are things I looked out for in this review. I hope help clear up a few things.
Full specifications of the Ultra-120 Extreme can be found at Thermalright's website.
The Ultra-120 Extreme comes in a plain cardboard box, which is the norm for Thermalright. It comes packaged with mounting hardware for Socket AM2 and LGA775. Those with Socket 753/940/939 can purchase a separate mounting kit. Note that this heatsink does not come with a fan, and that mounting clips for only one fan are provided. To mount an additional fan, clips can be bought for around $2-3 CAD (look up part number FWC-120-U120), or users can use a DIY method like zipties.
The Ultra-120 Extreme is a large upright design, measuring 63.44 x 132 x 160.5 (LxWxH, in cm). Note the height - it is very tall and might not fit some cases, or may interfere with side-mounted fans. The six heat pipes are copper and the fins are aluminium; everything is nickel plated. It weighs a hefty 790g - with a fan or two, which weigh roughly 180g each, this is a heavy load for the mounting mechanism / motherboard.
I only went through the Intel socket 775 installation here. Those curious about the differences for an AM2 install can check the directions on Thermalright's website.
Note that you must remove the motherboard from your case to install this heatsink.
The installation overall was pretty simple. Once complete I checked to see if the heatsink would swivel. For me the heatsink did not rotate very much at all and required some force to actually make it do so. The Zalman cooler that I had installed previously would swivel just as much. The heatsink was mounted solidly.
Intel Core 2 Duo e6400
ASUS P5B Deluxe, BIOS 1004
2GB Crucial Ballistix PC2-6400
Seagate 7200.10 HDDs
Enermax Liberty 400w
Windows XP Pro SP2 32bit
All tests were done in open air to remove the airflow characteristics of my case from testing.
Thermal paste used was Arctic Silver 5.
The Ultra 120 Extreme was mounted twice, the better results were used.
C1E / EIST (Speedstep) were disabled in the BIOS.
Any noise results were subjective only.
Ambient temperature at the time of testing was 20C.
The official specifications for the TriCools are 2000 / 1600 / 1200rpm (high / medium / low settings). In reality I think that their specs are closer to these results (2000/1500/900rpm) ... they move very little air on the "low" setting.
Speedfan, Core Temp v0.95 and Orthos applications were used for testing.
Orthos (Small FFTs) was run for 5-10 minutes and the maximum temperature observed in Speedfan's graphing mode on both cores was averaged and recorded. Idle tests were done after load tests, with temperatures allowed to stabilize for around 10 to 15 minutes and then maximum temperature on both cores averaged.
Speedfan was used to record voltages as detected in Windows.
Both the CNPS9500 and the Intel stock cooler were run at 100% fan speed for all tests.
Cooling at stock speeds
Voltage was 1.3000V in BIOS, 1.26v in Windows.
The Ultra-120 Extreme beats both the Zalman and Intel by a solid amount at stock speed. Note that at this speed, there is no difference between single and dual fans.
Only 7C above ambient at idle is pretty impressive, though.
Cooling at overclocked speeds
Voltage was 1.4750 in BIOS, 1.42v in Windows.
The Thermalright starts to show its benefits under a higher voltage load. It beats the Zalman by roughly 7-10C at load, which is a great result. Note that the benefit of dual fans starts to show at this speed - the dual fan setup is cooler by 3C.
Cooling at various fan speeds, push and push-pull
Voltage was 1.4750 in BIOS, 1.42v in Windows.
There are many interesting results in this graph. Note that there is barely any difference between the 1500 / 2000 rpm temperatures. Also, two fans at 1500rpm cool better at load than one at 2000rpm. Finally, a dual fan setup at 900rpm is 4C better than a single fan setup at the same speed.
Noise at these speeds varied widely. All of the fan speeds were quieter than both the Intel stock cooler and the Zalman, which were both obnoxiously loud. With the Antec fans, 2000rpm was loud, too much for me. 1500rpm was reasonable, but still somewhat loud. 900rpm was barely audible.
Comparing this to the Zalman results, note that the Thermalright can be run pretty much inaudibly with two fans at 900rpm, yet still beat the very loud Zalman by 4C.
From these results the best noise to performance ratio, I hypothesize, is roughly 1500rpm. Anything faster than this will probably result in diminishing returns, especially considering how loud the fans will be.
Also, very low airflow fans might not work best with this heatsink, judging from the large drop in performance between 900 and 1500 rpm compared with 1500 and 2000. Low speed fan performance improved with two fans, however.
Maximum overclock temperatures
Voltage was 1.5750 in BIOS, 1.52v in Windows.
I quickly ran a last test at a higher voltage load. Due to lack of time only the Thermalright was tested with a dual fan setup at 2000rpm.
The Thermalright holds the temperature at a great (considering how high the voltage was) temperature.
60C at 1.52v really is an excellent result.
Can be run very quietly
Dual fan setups possible
Base isn't flat
The fact that I found that this cooler performed great is nothing groundbreaking. The performance was something I more expected than hoped for, I guess.
The more important results from the fan speed and dual fan testing were overall pretty good. At stock speed, two fans on the cooler had no affect on performance. However, once the voltage and heat load rose the benefits showed up. 3C was taken off of the load temperature at 1.42v, making me wonder if a dual fan setup would show even more benefit at something like 1.5v. I'm starting to wish I had done more testing at 1.52v...
The fact that the base was not flat was somewhat of a concern, but the cooler still managed to perform very well. When I removed the Ultra-120 Extreme to check its contact with my processor, I found it to be fine. I can't say to what degree the flatness of the base ended up affecting the results.
In Canada, this cooler can be found for less than $50– for its performance, it really can't be beat. Consider that I paid something like 80$ for the Zalman CNPS9500 last year.
In conclusion: the Thermalright Ultra-120 Extreme is a great cooler. Unless its huge size is of concern, I see little reason to buy anything else.
Note: At a later time I might try re-testing with better thermal paste such as Arctic Cooling's MX-2. I also might revisit 1.5+ volts for testing. :thumb:
Nice review, looks like that cooler is a monster!
Yes its big
At least its tall and has a more narrow profile c.w. the sythe ninja or some other sinks for example.
That means you dont have to rearrange memory modules or have fans sitting on top of the memory.
nice Review !
considering this heatsink, thanks for the review.:clap:
Good review gof... but a few issues.
1.Testing cooling solutions really should be done in a case, as open-air bench doesn't duplicate airflow issues that do matter in a closed-case environment(the majority of readers).
2. This one's not "directly related" to the testing protocols, or results.. just a sort of unnerving observation. Laying computer parts all over a carpet doesn't illustrate safe handling(of ESD sensitive parts) practices to the readers that don't realize the possible dangers of doing so.
Sorry to sound so picky(i hope no hard feelings :whistle: ), they were the two standout(negatively) things that i couldn't help but question...the rest of the review was great.
However this was probably less of an unrealistic benefit than testing outside a case was :doh:... I guess I wanted results that were at least repeatable or reproduceable by users.
If I do any future reviews, I guess I will add some in-case testing results. Maybe block off or turn off my top fan. Because I agree that testing outside a case gives a slight temperature benefit to all testing.
Computer parts and carpets without precautions don't mix. Static in the carpet could kill your electronic components if you are not careful. If you absolutely have to handle your computer parts on a carpet, there are a few things I would recommend you do.
1) Do not have your components in direct contact with the carpet. If you can, do what I did... keep them on anti-static bags that come with motherboards or some other parts.
2) Keep your case nearby, so you can ground yourself as a precaution before handling the parts.
3) Don't wear socks? :bleh:
Thanks for the comment :thumb:
Great work, GOF- nice pics- are you using a light filter? almost looks like sepia
|All times are GMT -7. The time now is 05:58 AM.|