Monday, November 16th 2020

Der8auer Tries Out Intel's TEC Cooler on an AMD Ryzen 9 5950X

Der8auer has picked up Intel's latest TEC cooler, built in conjunction with EKWB, and unceremoniously plopped it right into an AMD Ryzen 9 5950X CPU. Well, not completely unceremoniously - there were many accommodations that had to be made in order to achieve this. For one, as this is a solution designed specifically for Intel CPUs and sockets, Der8auer had to Frankenstein his way through a number of cooling parts to be able to adapt the TEC solution to the AM4 socket. Not only that, but Intel's TEC requires deep software control for it to work properly - software control which only works with Intel silicon, of course. Der8auer thus had to have a second machine running an Intel 10900K CPU to control the software on the Intel cryo cooler.

All in all, the results were interesting, to say the least. The 16-core, 32-thread Ryzen 9 5950X saw single-core-load CCD temperatures in the 90 °C department with the TEC solution disabled - which promptly dropped down to only 50 °C with the cryo cooler actually operating. With a game load, the 5959X achieved up to 5.050 GHz in single-cores on its automatic boost profile. The entire chip often boosted to 4.8 - 4.9 GHz on all cores at once (with variances between the CCDs) whilst under this cooling solution and workload. With the TEC operating in its unregulated mode - which means, with no considerations for CPU operating temperature and power usage for the cooling process - saw the Ryzen 9 5950X achieving 2 °C core temperature results, and boosted frequencies up to a staggering 5323 MHz on all cores - before crashing. An interesting piece of work which you catch on video after the break; one can rest assured that most PC cooling specialists are already working on their own TEC-based cooling solutions following Intel's achievement in this field.
Source: via Tom's Hardware
Add your own comment

26 Comments on Der8auer Tries Out Intel's TEC Cooler on an AMD Ryzen 9 5950X

#1
IbaChiba
jee I wonder why intel has partnered up for this kind of cooling (looks at rocket lake....) power limits be damned.
Posted on Reply
#2
Houd.ini
Haha, imagine Intel being relegated to act as surrogate machines now, ouch.
Posted on Reply
#3
kapone32
I can't wait to see what my 5600X will do once my Alphacool Aurora Block arrives. I have a 420 rad and am waiting to see what this CPU can do.
Posted on Reply
#4
Mescalamba
TEC is pretty old, not entirely sure why it never really picked up. I guess demand was never that high for it. Question is, if its now. Sincerely doubt it, cause apart hardcore OC niché, most folks is okay running stock or slightly above, depending on cooling quality.

Also given power consumption and power cost, doubt many want to run OCed CPUs for long time.

Its nice and cute, as much as is multi-stage phase change, but.. in the end, its good for record and nothing else.
Posted on Reply
#5
lynx29
kapone32
I can't wait to see what my 5600X will do once my Alphacool Aurora Block arrives. I have a 420 rad and am waiting to see what this CPU can do.
ram is really most important here, if you read TPU review, W1zz was able tog et 4.7 on water all core, possibly 4.8 according to gamersnexus, but that is the cap for the 5600x, for some reason regardless of cooling it just won't go higher
Posted on Reply
#6
dj-electric
kapone32
I can't wait to see what my 5600X will do once my Alphacool Aurora Block arrives. I have a 420 rad and am waiting to see what this CPU can do.
The answer is exactly what a mid-high air cooling tower solution would. 5600X poses no threat to such coolers.
Posted on Reply
#7
kapone32
lynx29
ram is really most important here, if you read TPU review, W1zz was able tog et 4.7 on water all core, possibly 4.8 according to gamersnexus, but that is the cap for the 5600x, for some reason regardless of cooling it just won't go higher
It's not maximum clock I am looking at but maximum sustained clock. As one of the interesting things about Ryzen is how the boost program responds to cooling. It's crazy that I cannot get my 2920X (replacement part) to do 4.2 GHZ all cores but under PBO I see my best core boosting to 4.358 GHz which is not too shabby. The only AM4 CPU I have had that rad on before were the 1700 and 2600 which were limited to 4.015 and 4.2 GHZ all cores but the temps would be mid 20s C under idle and 45 C (72 C for 1700 die of 45 C) under load. As the Aurora is a block that has the reservoir above the CPU (in my opinion) it naturally helps to dissipate some of the heat which improves performance further. My Cooler Master Nepton 280 is the only AIO I would rate against the Esibear. Before people bash AIOs you can buy the Eisbear as individual parts if you would like to build your own custom water loop.
Posted on Reply
#8
thesmokingman
lynx29
ram is really most important here, if you read TPU review, W1zz was able tog et 4.7 on water all core, possibly 4.8 according to gamersnexus, but that is the cap for the 5600x, for some reason regardless of cooling it just won't go higher
Nah, doubt it is the limit. The voltage scaling on boost clock means they can run it on all cores with relative ease compared to previous gen. A guy on hard is running 4.8ghz all core at a silly low 1.29v. However that's all moot cuz in late 2020 all core overclocks are slow compared to AMD's PBO or Curve Optimizer. Btw the same user using AMD's method clocks higher and as much better single thread boost.
Posted on Reply
#9
TheLostSwede
Mescalamba
TEC is pretty old, not entirely sure why it never really picked up. I guess demand was never that high for it. Question is, if its now. Sincerely doubt it, cause apart hardcore OC niché, most folks is okay running stock or slightly above, depending on cooling quality.

Also given power consumption and power cost, doubt many want to run OCed CPUs for long time.

Its nice and cute, as much as is multi-stage phase change, but.. in the end, its good for record and nothing else.
Because moisture. Due to the temperature difference, moisture buildup is a problem with TEC cooling, or even frost buildup on the cooler at times. This was obviously some 15-20 years ago and hopefully things have improved since then.
Posted on Reply
#10
DR4G00N
Mescalamba
TEC is pretty old, not entirely sure why it never really picked up. I guess demand was never that high for it. Question is, if its now. Sincerely doubt it, cause apart hardcore OC niché, most folks is okay running stock or slightly above, depending on cooling quality.

Also given power consumption and power cost, doubt many want to run OCed CPUs for long time.

Its nice and cute, as much as is multi-stage phase change, but.. in the end, its good for record and nothing else.
It's because TEC's are terrible as active cooling, not only is their efficiency awful but as soon as they're overwhelmed they become an insulator and thus no cooling is happening.
I played around with some on an 8700K, gave low idle temps as you would expect but load temps were horrible, the custom water loop alone gave better temps.
Posted on Reply
#11
Tomgang
I'm sure Intel is very happy about it. I mean, now that intel litterly has lots mostly any reason to buy Intel. They can no longer hold the single core crown, games is mostly neck and neck, power consumption is horrible, 14 nm is so last season and multi core yeah let's not talk about that. I will just end up making Intel cry like a little baby. What have Intel left that is above Zen 3? Higher Clock speed and AVX512, that is pretty much its i can think of. They must be happy, there will at least be a possibility to be usable as surrogate machines...:nutkick:
Posted on Reply
#12
Haile Selassie
DR4G00N
It's because TEC's are terrible as active cooling, not only is their efficiency awful but as soon as they're overwhelmed they become an insulator and thus no cooling is happening.
I played around with some on an 8700K, gave low idle temps as you would expect but load temps were horrible, the custom water loop alone gave better temps.
Ditto.
The amount of media exposure for these contraptions lately is kind of funny really. Just shows how technically illiterate the tech media 'journalists' are these days.
Posted on Reply
#13
Imsochobo
Haile Selassie
Ditto.
The amount of media exposure for these contraptions lately is kind of funny really. Just shows how technically illiterate the tech media 'journalists' are these days.
derbauer did really say this is a bit stupid, cool but stupid and asked intel, what do you expect of this contraption ?
Posted on Reply
#14
thesmokingman
Imsochobo
derbauer did really say this is a bit stupid, cool but stupid and asked intel, what do you expect of this contraption ?
Cheaper than a hidden under the table 1kw chiller? :D
Posted on Reply
#15
Hattu
Now i know why there are cases that support dual systems. One for main rig and the other for cooler control.

:roll: :roll: :roll:
Posted on Reply
#16
Jism
DR4G00N
It's because TEC's are terrible as active cooling, not only is their efficiency awful but as soon as they're overwhelmed they become an insulator and thus no cooling is happening.
I played around with some on an 8700K, gave low idle temps as you would expect but load temps were horrible, the custom water loop alone gave better temps.
Yep. it's a dangerous setup to run 24/7 to be honest. It's cool for attempting an OC record or going below ambient etc. But it's not really intented for this purpose.

On xtremesystems.org you can still find a category devoted to TECS. Some very old threads showed some nuked TEC's while running on a 24/7 setup. If you have a heat source, you could however extract free power from it. You simply reverse the proces where the hot side is being heated; and the cold side is being cooled. This generates a small voltage and when paired enough you got some form of free energy.

But tec's are very bad in that respect; very low efficiency as well. They power the Voyager I and II, which is a sattelite in space that has a power estimation for over 40 years due to nuclear reaction. The heat is being converted into power for the sattelite. Now thats putting a tec to good use.
Posted on Reply
#17
Athlonite
TheLostSwede
Because moisture. Due to the temperature difference, moisture buildup is a problem with TEC cooling, or even frost buildup on the cooler at times. This was obviously some 15-20 years ago and hopefully things have improved since then.
not just Frost/moisture build up the cost of running one of these was also huge compared to water or air cooling tec's just use power like it's going out of fashion not to mention the fact that you also need a good water cooling loop to dissipate the heat from the tec itself aswell
Posted on Reply
#18
Nephilim666
I much preferred my Lian Li V2000B with a phase change unit built in. -50C load temps on Conroe. Ran it daily for around a year until condensation killed the board. That was with the board coated in conformal spray, liquid electrical tape and the socket filled with dielectric grease. Sub-ambient is no joke.
Posted on Reply
#21
Crackong
That TEC would be using more power than the 5950x itself lols
Posted on Reply
#22
thesmokingman
Crackong
That TEC would be using more power than the 5950x itself lols
Probably two times more than the chip itself, lol.
Posted on Reply
#23
cyberloner
Crackong
That TEC would be using more power than the 5950x itself lols
true.... cooling use more power than cpu itself.... lol
Posted on Reply
#24
PanicLake
kapone32
I can't wait to see what my 5600X will do once my Alphacool Aurora Block arrives. I have a 420 rad and am waiting to see what this CPU can do.
You know what a 5600x and a very expensive cooling system can do? Probably give you the value of a 5900x with regular cooling...
Sorry, but I will never understand that logic. Like those who buy small and cheap cars and then spends all their money with spoilers, paint, tuning...
Posted on Reply
#25
Arcdar
Now Der8auer banned from all Intel related Tests and NDA-testing with preview samples *lol*

Sorry Bauer, der musste sein ;) … Trotzdem interessant und danke für den Test :D
Posted on Reply
Add your own comment