Thermalright Ultima-90: Small Wonder?
by Wesley Fink on August 20, 2007 2:00 AM EST- Posted in
- Cases/Cooling/PSUs
Overclocking
As cooling solutions do a better job of keeping the CPU at a lower temperature, 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: 2.00V
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 memory speed of DDR2-800. As the 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 Ultima-90 reaches a stable 3.90 GHz, matching the top performers we have tested like the Tuniq 120, Thermalright Ultra-120, Scythe Infinity with push-pull fans, and the Scythe Ninja Plus B/OCZ Vindicator running a high output SilenX IXTREME fan. The only air cooler that outperforms the Ultima-90 in overclocking is its big brother - the Ulltra-120 eXtreme.
These results are with a high output 120mm fan. Results with a 92mm fan are slightly lower, with a top overclock of 3.87GHz. As seen in the scaling charts the 120mm and 92mm fan results are almost the same to the top of the speed charts. Cooling performance is also about the same with either a 120mm or 92mm driving. However, the 120mm fan allows a slightly higher overclock.
If wattage is considered, a stock X6800 consumes about 75 watts, where the 3.83GHz OC pulls about 150 watts - double the amount required at stock speed. At the voltage required for stability at 3.90 GHz with a 120mm fan, the CPU wattage is around 161W. The highest air result ever with this CPU at 3.94GHz translates into about 165W, where the 92mm top of 3.87GHz translates into around 158W. Considering the smaller size and reduced weight of the Ultima-90, this performance can only be considered outstanding.
Note that a C2D pushed to the incredible overclocks they can reach draws a lot of power, and our cooling test is designed to find where the top coolers fail. Keep this in mind when looking at our cooling results. Many of the middle coolers in our roundup have done very well when tested with AMD processors or with less demanding cooling test procedures.
As cooling solutions do a better job of keeping the CPU at a lower temperature, 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: 2.00V
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 memory speed of DDR2-800. As the 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 Ultima-90 reaches a stable 3.90 GHz, matching the top performers we have tested like the Tuniq 120, Thermalright Ultra-120, Scythe Infinity with push-pull fans, and the Scythe Ninja Plus B/OCZ Vindicator running a high output SilenX IXTREME fan. The only air cooler that outperforms the Ultima-90 in overclocking is its big brother - the Ulltra-120 eXtreme.
These results are with a high output 120mm fan. Results with a 92mm fan are slightly lower, with a top overclock of 3.87GHz. As seen in the scaling charts the 120mm and 92mm fan results are almost the same to the top of the speed charts. Cooling performance is also about the same with either a 120mm or 92mm driving. However, the 120mm fan allows a slightly higher overclock.
If wattage is considered, a stock X6800 consumes about 75 watts, where the 3.83GHz OC pulls about 150 watts - double the amount required at stock speed. At the voltage required for stability at 3.90 GHz with a 120mm fan, the CPU wattage is around 161W. The highest air result ever with this CPU at 3.94GHz translates into about 165W, where the 92mm top of 3.87GHz translates into around 158W. Considering the smaller size and reduced weight of the Ultima-90, this performance can only be considered outstanding.
Note that a C2D pushed to the incredible overclocks they can reach draws a lot of power, and our cooling test is designed to find where the top coolers fail. Keep this in mind when looking at our cooling results. Many of the middle coolers in our roundup have done very well when tested with AMD processors or with less demanding cooling test procedures.
Facebook
Twitter
RSS
38 Comments
View All Comments
Rick1 - Sunday, August 26, 2007 - link
In a couple of post above the questions are asked Why run one of these coolersMy answer is simple
Quiet and cool running system
The only fan I hear is the One installed in the hard drive compartment of this P182B case
( cooling 4 drives. 2x raptors and 2x Barracuda 7200.10 SATA 3.0Gb/s 500-GB )
With 2 fans S-Flex blowing in and the stock two exhaust fans
This Q6600 runs at 32Cto35C and has never gone over 48C under a full load
I was never able to get the stock H/S below 50C on warm days
jnk - Thursday, October 18, 2007 - link
question for the reviewer:when you reviewed the ultima-90, when mounted were you able to twist it? I recently bought one and i can twist it even while its locked and the screws are tight. I emailed thermalright about it and they that its normal.
Patvs - Saturday, August 25, 2007 - link
This is the best CPU cooler review on the net! I have one question though.In the Noise Level test, some coolers are tested with low and high RPM settings.
For example the Tuniq Tower 120 @1000 RPM and @2100 RPM. However the Thermaltake Big Typhoon VX is only tested @2000 RPM.
Is the HIGH RPM setting always used for the Temperature Tests (IDLE and LOAD)?? Or do you use the LOW RPM setting as default for the temperature tests? *confused* If HIGH: it shows the Tuniq is really quiet at LOW RPM, but you show its cooler potential in temperature in HIGH RPM? How does it cool at LOW RPM? (or if LOW: how much does the temperature decrease if the cooler is at HIGH RPM) Also I would love to see a test with TWO fans hooked up to some of these coolers in the future.
Patvs - Saturday, August 25, 2007 - link
Edit: You state you use stock speed RPM settings for the temperature tests.So why use 2000 RPM for the Big Typhoon VX? Isn't its stock speed 1300 RPM. (it is for the Big Typhoon non-VX version without the fan controller)
muddocktor - Wednesday, August 22, 2007 - link
As always, a good heatsink review by you, Wesley. But I have a question about the Thermalright samples that you all get for review. Do you receive these directly from Thermalright or are they procured from an authorized reseller such as Sidewinder Computers or Newegg from actual shipping production? The reason I ask is that while I find the engineering and design of Thermalright's heatsinks to be top-notch, I have personally found that their base finish to be spotty. I own or have owned 2 XP90's, an XP90-C, SI 120, Ultra 120, and an Ultra 120 eXtreme (all bought at retail except the SI 120, which was bought used) and of those the only ones that had a half decent base finish that was usable as-is were the XP90-C and SI 120. The others either had a fairly poor base finish with visible machining marks left in them and in the case of the U-120, an absolutely horrible base finish with a ridge left on one side of the base. The XP-90's also had very concave bases too. All saw improved to much improved performance after giving the base a lap job. If you are getting your review samples directly from Thermalright instead of from a vendor that handles their heatsinks, I am sure that the samples you receive are thoroughly checked for finish before sending them out to you and aren't truly representative of what is actually going out for sale through normal channels. It would be interesting to see if you could get some samples from someone online that didn't know they were going to Anandtech (to minimize the chance of cherry picking the heatsinks) and do a consistency review showing differences (or lack of) in performance of production line Thermalright heatsinks.Also, I have heard the thing Thermalright has put out about the concaveness of their bases being engineered into their design and don't buy that a bit. I think that is just something the salesmen have thought up to cover inconsistencies in the base finish from their manufacturer and my experience (so far) has proved out for me at least that a lapped, flat base works better on both LGA775 and socket 939 systems. Other than the base finish though, I find that Thermalright makes the best overall line of premium aircooling equipment on the market.
Wesley Fink - Wednesday, August 22, 2007 - link
We first tested the Ultima-90 with a pre-production cooler. Results were then verified with a Retail cooler from a stocking retailer. Where there have been questions about items being "hand-picked" we often verify results with a retail sample.In the case of the Ultima-90 the performance of the Thermalright-supplied Ultima-90 and the one off the Retail shelf were exactly the same.
muddocktor - Sunday, September 2, 2007 - link
Thank you for the answer answer on your samples you test, Wesley. I guess I just have bad luck in the base finishes I get on the Thermalright heatsinks I buy then. But since I know how to lap the base anyways it's no big problem for me. Thermalright does make some truly excellent products for cooling highly overclocked cpus.Wesley Fink - Monday, September 3, 2007 - link
Thermalright is not alone in advocating convex bases. Swiftech has moved from promoting flat, mirror-finish bases to convex bases with a finish not as good. They call them their bow base and claim 2 to 4C better performance with the "bow".Engineers have found that the convex base (fat center) mates tighter in the area of the CPU under the cap. Also Intel manufactures caps that are not flat by design - conncave, convex, and wave designs are all used in certain processors. Research shows the convex cooler base mates best with any of these 3 Intel cap types.
The point is this is not a marketing story, as you will see as we go into water-cooling in more detail. Lapping a convex base Thermalright can DROP performance by 2 to 4C.
customcoms - Friday, August 24, 2007 - link
Any chance you guys will be reviewing an Ultra-90? The reason I ask is because silicon valley computers often has these heatsinks in stock and on sale for $15...that seems like a great bargain to me especially when the bigger brothers, the Ultra-120 and 120 eXtreme, retail for upwards of $50. I would really like to know if spending the extra money for an Ultima-90 or Ultra-120 is even worth the performance increase...dm0r - Tuesday, August 21, 2007 - link
Great review Wesley, as aways.