Dual tower coolers. Worth it?

[sakete]

Do you guys think dual tower coolers (such as Noctua D15 or Cryorig R1) are worth the extra money vs. their single tower counterparts? Is the extra cooling performance that much better than single towers? This is if you will overclock your cpu. Currently debating whether I should invest the extra money in a dual tower cooler (will be getting a 4970K).

 

[jji7skyline]

Dual tower coolers like the Noctua D15 perform close to mid-range AIO liquid coolers when configured properly. Depending on how much you want to overclock, it may be worth it.

 

[lehtv]

Given that 4790K already runs at 4.2-4.4GHz without OC, there are very diminishing returns to buying expensive overclocking gear. I would go with

(1) basic cooling to keep it quiet (92-120mm tower) and a quality H97 motherboard (H97-D3H, Fatal1ty H97 Performance / Killer, Asus H97 Pro Gamer, MSI H97 Gaming-3), possibly an entry level SLI board if SLI interests you
(2) 4690K which has more OC headroom than 4790K, and a 6-8 phase OC motherboard (Z97S SLI Krait, Z97 Pro Gamer, Z97 Gaming-5 etc) and a single tower cooler such as Mugen 4, HR-02 Macho, U14S, or perhaps D14 since it's quite cheap compared to D15

 

[BonzaiDuck]

I can most easily recommend that which I've used and tested myself against (a) a well-known cooler in a near-identical system and (b) review conclusions such as available at sites like Hardware Secrets.

I've used an NH-D14 on one system since July, 2011 -- with the 32nm Sandy Bridge 2600K OC'd to 4.7 Ghz.

Last year, I began using an EVGA ACX cooler on an identical system with a 2700K OC'd to 4.7 Ghz. The thermal power on these two systems was essentially the same under IBT "Maximum" stressing, same room-ambient, same cases, same or very similar fan choices, and same ducting-modifications (despite D14 dual-tower and the single-tower ACX).

Without any ducting or other special treatment, the Hardware Secrets review shows the ACX cooler outperforming the D14 by 6C lower temperatures. With my two systems ducted to rear-exhaust, the difference is about 5.5C degrees -- or an overall improvement for the ACX by ~12C below the stock review performance of the D14.

The EVGA ACX has a single-tower size no bigger than any single tower cooler -- most of which are larger than a CM Hyper 212 EVO. But the difference in size from the EVO isn't really material. Most people complain about the size of the D14, D15, IB-E TR SilverArrow and similar dual-towers.

The D15 outperforms the D14 by 3 to 4C in reviews I've seen, so the ACX would still outperform the D15.

The only shortcoming of the ACX is its noisier LED fan, which can be eliminated entirely by a TR blue accordion duct to rear exhaust (~$5.00), or just a better fan.

http://www.newegg.com/Product/Produ...re=EVGA_ACX_CPU_cooler-_-35-288-004-_-Product

I would like anyone with knowledge of another single-tower heatpipe cooler that outperforms the ACX to post the link and a review comparison. I had no knowledge of the ACX (or its early name -- "SuperClock") until last year, so it escaped my scrutiny for some time.

FOOTNOTE: How this all works with the 4790K, which will have lower thermal power under overclocks than a Sandy Bridge, I cannot tell. Because of the smaller die size of the Haswell, cores can reach higher temperatures with the same cooler in comparison to the Sandy cores. The Haswell/Devils Canyon uses a polymer TIM instead of indium solder between IHS and processor die. It would depend on the degree of overclocking. But using a popular AiO dual-fan cooler's performance as reference, all the coolers mentioned would be more than adequate since the AiO cooler's performance is more than adequate.

 

[jji7skyline]

I can most easily recommend that which I've used and tested myself against (a) a well-known cooler in a near-identical system and (b) review conclusions such as available at sites like Hardware Secrets.

I've used an NH-D14 on one system since July, 2011 -- with the 32nm Sandy Bridge 2600K OC'd to 4.7 Ghz.

Last year, I began using an EVGA ACX cooler on an identical system with a 2700K OC'd to 4.7 Ghz. The thermal power on these two systems was essentially the same under IBT "Maximum" stressing, same room-ambient, same cases, same or very similar fan choices, and same ducting-modifications (despite D14 dual-tower and the single-tower ACX).

Without any ducting or other special treatment, the Hardware Secrets review shows the ACX cooler outperforming the D14 by 6C lower temperatures. With my two systems ducted to rear-exhaust, the difference is about 5.5C degrees -- or an overall improvement for the ACX by ~12C below the stock review performance of the D14.

The EVGA ACX has a single-tower size no bigger than any single tower cooler -- most of which are larger than a CM Hyper 212 EVO. But the difference in size from the EVO isn't really material. Most people complain about the size of the D14, D15, IB-E TR SilverArrow and similar dual-towers.

The D15 outperforms the D14 by 3 to 4C in reviews I've seen, so the ACX would still outperform the D15.

The only shortcoming of the ACX is its noisier LED fan, which can be eliminated entirely by a TR blue accordion duct to rear exhaust (~$5.00), or just a better fan.

http://www.newegg.com/Product/Produ...re=EVGA_ACX_CPU_cooler-_-35-288-004-_-Product

I would like anyone with knowledge of another single-tower heatpipe cooler that outperforms the ACX to post the link and a review comparison. I had no knowledge of the ACX (or its early name -- "SuperClock") until last year, so it escaped my scrutiny for some time.

FOOTNOTE: How this all works with the 4790K, which will have lower thermal power under overclocks than a Sandy Bridge, I cannot tell. Because of the smaller die size of the Haswell, cores can reach higher temperatures with the same cooler in comparison to the Sandy cores. The Haswell/Devils Canyon uses a polymer TIM instead of indium solder between IHS and processor die. It would depend on the degree of overclocking. But using a popular AiO dual-fan cooler's performance as reference, all the coolers mentioned would be more than adequate since the AiO cooler's performance is more than adequate.

Doesn't seem fair to compare two heatsinks with different fans. For example my heatsink performs vastly differently depending on what fan is on it, and what speed the fans run at.

 

[BonzaiDuck]

Doesn't seem fair to compare two heatsinks with different fans. For example my heatsink performs vastly differently depending on what fan is on it, and what speed the fans run at.

I really didn't compare MY OWN system with different fans -- or it was only a PARTIAL difference.

The 2600K and 2700K both have GT AP-30 PWM exhaust fans controlled by the CPU_FAN ports.

The D14 on the 2600K is fitted with a 140mm "R" or circular Akasa Viper -- rated at ~110 CFM at top speed of ~1,650 RPM. The ACX cooler on the 2700K has a Noctua iPPC 3000 (3,000 RPM top-end) rated at approximately 110 CFM. The Akasa is mounted between the towers, so it "pulls" from the innermost tower and pushes through the tower closest to the AP-30 exhaust.

But I think you really mean that the different coolers are tested with their respective and different fans, so maybe you support my idea of a brute-force "CFM" test for maximum cooling on any given cooler to test their full cooling power.

I always throw up references to the Anthony Hopkins portrayal of Burt Munro in "World's Fastest Indian." If it's an accurate depiction of his adventures and exploits at the Bonneville Salt Flats, then there's an analog to the flap dramatized in the movie about his tires and "safety equipment."

I raise that vague an imperfect thought as an imperfect analogy to cooler testing: There are probably rules, and the rules probably require agreement of the manufacturers who submit their products for the tests. Since they bundle their fans with the coolers, they probably insist on testing their respective "whole enchiladas."

Maybe not. I can't say for sure. But I'd like to see tests with 120mm/140mm equivalents to the Vantec Tornado -- noise be damned -- to see what those coolers can really do for a deaf person.

:biggrin::biggrin:

 

[soccerballtux]

I can most easily recommend that which I've used and tested myself against (a) a well-known cooler in a near-identical system and (b) review conclusions such as available at sites like Hardware Secrets.

I've used an NH-D14 on one system since July, 2011 -- with the 32nm Sandy Bridge 2600K OC'd to 4.7 Ghz.

Last year, I began using an EVGA ACX cooler on an identical system with a 2700K OC'd to 4.7 Ghz. The thermal power on these two systems was essentially the same under IBT "Maximum" stressing, same room-ambient, same cases, same or very similar fan choices, and same ducting-modifications (despite D14 dual-tower and the single-tower ACX).

Without any ducting or other special treatment, the Hardware Secrets review shows the ACX cooler outperforming the D14 by 6C lower temperatures. With my two systems ducted to rear-exhaust, the difference is about 5.5C degrees -- or an overall improvement for the ACX by ~12C below the stock review performance of the D14.

The EVGA ACX has a single-tower size no bigger than any single tower cooler -- most of which are larger than a CM Hyper 212 EVO. But the difference in size from the EVO isn't really material. Most people complain about the size of the D14, D15, IB-E TR SilverArrow and similar dual-towers.

The D15 outperforms the D14 by 3 to 4C in reviews I've seen, so the ACX would still outperform the D15.

The only shortcoming of the ACX is its noisier LED fan, which can be eliminated entirely by a TR blue accordion duct to rear exhaust (~$5.00), or just a better fan.

http://www.newegg.com/Product/Produ...re=EVGA_ACX_CPU_cooler-_-35-288-004-_-Product

I would like anyone with knowledge of another single-tower heatpipe cooler that outperforms the ACX to post the link and a review comparison. I had no knowledge of the ACX (or its early name -- "SuperClock") until last year, so it escaped my scrutiny for some time.

FOOTNOTE: How this all works with the 4790K, which will have lower thermal power under overclocks than a Sandy Bridge, I cannot tell. Because of the smaller die size of the Haswell, cores can reach higher temperatures with the same cooler in comparison to the Sandy cores. The Haswell/Devils Canyon uses a polymer TIM instead of indium solder between IHS and processor die. It would depend on the degree of overclocking. But using a popular AiO dual-fan cooler's performance as reference, all the coolers mentioned would be more than adequate since the AiO cooler's performance is more than adequate.

did you verify the voltage and LLC settings you were supplying to both CPUs were the same? Did you try swapping coolers between the machines? Swapping not necessary to convince me, but if you check the voltages shown in CPU-z to verify they were the same, then at the same frequency (4.7ghz), yes the conclusion that they had similar TDP would be acceptable. The problem though is I've never seen software properly report the power consumption.

I'm just really hoping I didn't make my EVGA ACX purchase by blind faith in someone who made a silly mistake!!!! <3

 

[BonzaiDuck]

did you verify the voltage and LLC settings you were supplying to both CPUs were the same? Did you try swapping coolers between the machines? Swapping not necessary to convince me, but if you check the voltages shown in CPU-z to verify they were the same, then at the same frequency (4.7ghz), yes the conclusion that they had similar TDP would be acceptable. The problem though is I've never seen software properly report the power consumption.

I'm just really hoping I didn't make my EVGA ACX purchase by blind faith in someone who made a silly mistake!!!! <3

I split fewer hairs than that, so to speak. Everything was identical except the coolers themselves, and both coolers were similarly ducted. Fan profiles, number of fans, type of fans were all identical between the identical cases. Same basic motherboard. LLC and voltage settings the same. Thermal power measured in HWMonitor -- the same.

It still proved a 6C improvement for the ACX over the D14.

NOW it's different though! Both systems had graphics cards of similar type, cooling and other aspects -- single cards, one per computer. When I added a second 970 GTX to the 2700 box, temperatures increased by a few degrees C.

 

[BonzaiDuck]

Here's also a thought in reasonable speculation.

Today's Anandtech home page features a "roundup" of air coolers. One of them you will notice has two features similar to the ACX: closed sides on at least one heatpipe tower's fins, and the rubber fan-mounting parts and design.

So you could suspect that EVGA contracted to have the ACX made, and the Anand article may (may!) indicate where that trail ends.