DeepCool AK620

[crazyeyesreaper]

DeepCool looks to assert dominance with its latest dual-tower air cooler, the AK620. Featuring an accommodating height of just 160 mm, it fits where other large coolers won't—all while packing serious performance with clean and stylish looks.

Show full review

 

[The King]

Had my eye on this cooler for awhile. Performance seems slightly weaker compared to other reviews I have seen where this cooler outperformed the DCA3 and the D15 in several tests.

Maybe related to the different CPUs being used.

 

[luches]

Deepcool is creating a good line up of air coolers that challenges Noctua at every level. Back then i was assembling my 5900x setup, NH-D15 was out of stock in my country, so i was kinda forced to buy the Deepcool Dark Assassin III based on reviews and didn't look back. This one also looks like very solid choice for those who don't want a truck sized air cooler in their rig but still want top notch performance.

 

[JustBenching]

Was really excited about this review. Sad it performs great only on AMD. I guess ill stick with my u12a

Do you think it's possible to fit a 30mm fan between the heatsinks? Like a t30 120

 

[ixi]

Had my eye on this cooler for awhile. Performance seems slightly weaker compared to other reviews I have seen where this cooler outperformed the DCA3 and the D15 in several tests.

Maybe related to the different CPUs being used.

I guess I'm blind cause in TPU provided graphs it eats NH-D15 on AMD while on intel wins and loses somewhere.

 

[JustBenching]

I guess I'm blind cause in TPU provided graphs it eats NH-D15 on AMD while on intel wins and loses somewhere.

I guess you are, cause it only wins in 1 test on AMD. Dunno about idle, didn't check those. On Intel it gets destroyed.

 

[80-watt Hamster]

Anyone else feel like we've basically reached the "flat" of the optimization curve in CPU cooling? From the noise-normalized chart:

Seems one could just say, "Pick a tower cooler you like the looks of and call it a day. Don't spend more than USD70 (or less than USD20)."

 

[crazyeyesreaper]

Anyone else feel like we've basically reached the "flat" of the optimization curve in CPU cooling? From the noise-normalized chart:

View attachment 233725

Seems one could just say, "Pick a tower cooler you like the looks of and call it a day. Don't spend more than USD70 (or less than USD20)."

Keep in mind the charts include the data for stock and overclocked settings. Thus it tends to skew a bit more conservatively.

 

[80-watt Hamster]

Keep in mind the charts include the data for stock and overclocked settings. Thus it tends to skew a bit more conservatively.

Right; I'd zeroed in on that one because (and I could be wrong here) it aggregates the stock and OC noise-normalized values for both AMD and Intel, plus provides a common constraint to the performance envelope. In my mind, it's kind of like if one ran a CPU test with common power limits to see which one does a particular job most efficiently. You know, if one could actually assume said limits were accurate and consistent across models.

Anyway, hypothetical BS and erroneous conclusions on my part aside, what I mean by the flat of the curve is that advancements are getting harder and harder to come by, kind of like MPG in autos. We're already up to six heat pipes, two towers and multiple huge fans for top performers (sticking with air cooling here). We're not really getting better mousetraps, but simply differentiated ones, so to speak.

 

[Chrispy_]

Why would I spend almost three times the price of a Fera5? For $26-35 there are several coolers in the chart that perform similarly enough that it doesn't matter.

$69 is Noctua territory, and I'm still using my original NH-U12 purchased for a Core2 Q6600 about 15 years ago. They still honour the free socket upgrade kit and I'm getting my money's worth!

 

[JustBenching]

Why would I spend almost three times the price of a Fera5? For $26-35 there are several coolers in the chart that perform similarly enough that it doesn't matter.

What? Which 25 to 35 dollars cooler perforns similarly? Im going to but it tomorrow morning. Most ek cheapos including the fera5 throttled in a bunch of tests

 

[Chrispy_]

What? Which 25 to 35 dollars cooler perforns similarly? Im going to but it tomorrow morning. Most ek cheapos including the fera5 throttled in a bunch of tests

The tests that are literally as described by the tester as being "an unrealistic workload" with the fans slowed down to just 45dBA? Yeah. The AK620 is uncomfortably roasty in some of those tests too. 91C and 93C are't exactly far off throttling either.

TBF I was looking at the relative performance charts where the ID 224XT for $29 is within 5%, noise-normalised. That's close enough for me to start wondering about why one is significantly more than double the cost.

If you're buying a Z-series board and a $500 CPU with 250W PL2 at stock, you're basically asking for trouble if you want to start adding voltage for OC with an air cooler. Not even the NH-D15 will save you under the "unrealistic workload" there.

I'm with @80-watt Hamster here; If you're not doing anything crazy, almost any tower will do just fine.
Likewise, if you're doing crazy stuff, no tower is enough, you need liquid cooling.

 

[Mussels]

This actually looks like a really solid option.

Pity two of my three systems run 51mm RGB RAM

 

[JustBenching]

The tests that are literally as described by the tester as being "an unrealistic workload" with the fans slowed down to just 45dBA? Yeah. The AK620 is uncomfortably roasty in some of those tests too. 91C and 93C are't exactly far off throttling either.

TBF I was looking at the relative performance charts where the ID 224XT for $29 is within 5%, noise-normalised. That's close enough for me to start wondering about why one is significantly more than double the cost.

If you're buying a Z-series board and a $500 CPU with 250W PL2 at stock, you're basically asking for trouble if you want to start adding voltage for OC with an air cooler. Not even the NH-D15 will save you under the "unrealistic workload" there.

I'm with @80-watt Hamster here; If you're not doing anything crazy, almost any tower will do just fine.
Likewise, if you're doing crazy stuff, no tower is enough, you need liquid cooling.

Not really, I have a u12a on a 12900k. At stock it peaks at 77 running cbr23. That's with the fan at whisper quiet. If I max the fans sometimes it stays below 70. Overclocked I can run 5.3 to 5.2 ghz depending on the temperature (I have octvb dropping bins at 85 and 90). Air coolers are fine even for high loads, its all about the contact pressure. From what ive seen online all AIOs get worse results than my u12a, only some custom WC with moras on the balcony are competing. Dunno why, maybe its not the cooler but a better quality ihs or whatever.

You should always look at the unrealistic tests when you are buying a cooler. I mean do you really buy it for idle temps or for gaming? In that case, probably every cooler besides something tiny like intel stock will do the job.

 

[Mussels]

I look at those unrealistic tests because they help compensate for things normal reviews miss - the heat from a high wattage GPU being added into the mix, high ambients, and anything else that could negatively affect the cooling.


You want a cooler that does the job AND has room to spare - so if you know it can do 5.1GHz on your air fryer intel at 93C, you can be confident it'll handle 4.8Ghz even in poor conditions

I've used a lot of coolers (and cases) out there that had glowing reviews, only to turn around and find they sucked ass in real life because reviews often test with CPU load only, in an open test bench (this also makes me think of the many NZXT case complaints out there about their beautiful systems that overheat like mad because NZXT decided front air intake was for losers)

 

[MachineLearning]

Thank you for this review, excellent as always.

I'm sure you guys already have a long backlog of coolers to review, but if it piques your interest, check out the Thermalright Peerless Assassin PA120. In my own experience it's a great cooler for ~$40USD but I'd love to see it go through your testing gauntlet. I'll seriously buy one for TPU if they won't send you a sample, not enough TR reviews around these days!

 

[LlF]

why would it perform significantly worse on intel than amd? on many amd disucssion people are saying ryzen runs hot, learn to accept it. are you saying intel is even warmer?

 

[JustBenching]

why would it perform significantly worse on intel than amd? on many amd disucssion people are saying ryzen runs hot, learn to accept it. are you saying intel is even warmer?

Νope, AMD does run warmer, it's just that the part of the cooler that makes contact with the CPU might have been designed for better compatibility with Ryzens CPU shape. After all, CPUs are not flat, especially after inserted into the socket

 

[Mussels]

why would it perform significantly worse on intel than amd? on many amd disucssion people are saying ryzen runs hot, learn to accept it. are you saying intel is even warmer?

mounting has to prefer one design over the other, for every cooler. Few work equally on intel and AMD, except for the high end premium products where they put a lot more R&D into the mounting

Ryzen doesnt run hot, ryzen is hard to cool.
Intel runs hot, but is easier to cool.

 

[Space Lynx]

@crazyeyesreaper I really wish you had a review for the popular Vetroo V5 cpu cooler, I am curious to see how it does versus this cooler... this one is currently on lighting sale on amazon for $55... im super tempted...

 

[JustBenching]

mounting has to prefer one design over the other, for every cooler. Few work equally on intel and AMD, except for the high end premium products where they put a lot more R&D into the mounting

Ryzen doesnt run hot, ryzen is hard to cool.
Intel runs hot, but is easier to cool.

Xcuse me? So hot much wow

 

[Mussels]

Xcuse me? So hot much wow

Whatever you're trying to say, you forgot to actually say it.