WEBVTT

00:00:00.080 --> 00:00:08.240
Intel's core i9 12900k in performance mode gets hot too hot to cool with the

00:00:05.680 --> 00:00:13.679
traditional heatsink too hot even for an all-in-one liquid cooler ugh smells

00:00:10.559 --> 00:00:15.519
nasty surely our mighty custom loop can

00:00:13.679 --> 00:00:20.880
save the day all right so we're getting there but

00:00:17.680 --> 00:00:22.000
what if we had three tubes instead of

00:00:20.880 --> 00:00:26.800
two luex is a newcomer in the custom water

00:00:24.800 --> 00:00:31.840
cooling space and claims that their tri-swift water block is the best on the

00:00:29.679 --> 00:00:36.559
market surpassing the latest and greatest from well-known brands like ek

00:00:33.840 --> 00:00:41.120
and swiftek their approach divide the coolant to conquer the heat according to

00:00:39.360 --> 00:00:46.399
tynan from our engineering department their theory does hold some water but

00:00:44.079 --> 00:00:51.280
there's only one way to see if adding extra speed holes to your water block

00:00:48.480 --> 00:00:56.000
will make your CPU as fast as this segue to our sponsor ridge ridge wallet has

00:00:54.000 --> 00:01:00.239
redefined their traditional wallet with its compact frame and rfid blocking

00:00:58.239 --> 00:01:03.840
plates the bulge in your pants shouldn't be from your wallet use offer code Linus

00:01:02.640 --> 00:01:08.240
to save 10 and get free worldwide shipping today

00:01:06.240 --> 00:01:12.720
this is the quantum velocity squared water block from ek it's one of the top

00:01:10.400 --> 00:01:19.280
performing blocks on the market and like our challenger the fluix tri-swift it

00:01:15.520 --> 00:01:20.799
can cost a whopping 200 plus us dollars

00:01:19.280 --> 00:01:26.080
depending on options like material choice that's just for the block which

00:01:24.000 --> 00:01:29.840
means that to test these monsters we first had to dig up all the other

00:01:28.080 --> 00:01:34.159
components of a quick custom water cooling loop then we grabbed our CPU

00:01:32.320 --> 00:01:38.640
frame from thermal grizzly cranked our Noctua fans and d5 pump to their max

00:01:36.400 --> 00:01:43.759
speeds and configured the classroom blender benchmark to a whopping 3 000

00:01:41.439 --> 00:01:49.040
path samples with our CPU in the default power profile our ek block here barely

00:01:46.399 --> 00:01:53.600
cracked 72 degrees once our loop reached thermal equilibrium that's a tough mark

00:01:51.360 --> 00:01:57.360
to beat it's significantly better than some of the air and all-in-one water

00:01:55.520 --> 00:02:01.920
coolers that we've looked at recently but fluix is going to have to do it because

00:02:00.000 --> 00:02:06.880
when you're spending this kind of money on cooling every fraction of a degree

00:02:04.640 --> 00:02:11.520
matters and as i learned in my product management days the demand for the

00:02:09.280 --> 00:02:15.840
second best custom water block is essentially zero i don't think that's

00:02:13.840 --> 00:02:20.959
going to be a problem though since fluix promises to win by a lot more than a

00:02:18.319 --> 00:02:26.640
fraction of a degree i mean look at this two inlets instead of one that's right

00:02:23.599 --> 00:02:27.440
double the water flow to your

00:02:26.640 --> 00:02:32.560
wait it's just a y splitter

00:02:30.800 --> 00:02:37.280
we're back sorry they explain this better than i can here's the product

00:02:34.560 --> 00:02:41.599
description emphasis mine the fluix tri-swift is the highest performance

00:02:39.440 --> 00:02:44.720
block on the planet 1.5 millimeter microchannels proprietary sinusoidal

00:02:43.280 --> 00:02:49.599
pattern jet plate great turbulence mixing chamber something something this

00:02:46.879 --> 00:02:55.680
allows our tri-swift water block to lower CPU temperatures by 8 to 10

00:02:53.280 --> 00:03:00.959
degrees celsius during intensive workloads compared to substitute water

00:02:58.560 --> 00:03:05.519
blocks here's the thing guys we don't really

00:03:02.879 --> 00:03:10.560
review water blocks anymore but these are some bold claims almost as bold as

00:03:08.640 --> 00:03:14.959
my claim that you'll find all the best behind the scenes extras and exclusives

00:03:12.159 --> 00:03:19.519
over on Floatplane.com like alex making joystick condoms anywho

00:03:17.680 --> 00:03:23.760
buried in that long-winded marketing spiel that you just endured are some

00:03:21.840 --> 00:03:27.599
actual grains of truth tynan is gonna fill you in

00:03:25.599 --> 00:03:31.519
they're basically saying that their twin inlets assisted by these wavy jet plates

00:03:29.760 --> 00:03:35.280
will increase their fluid turbulence improving the thermal transfer between

00:03:33.200 --> 00:03:39.280
the copper fins and the water in your loop it's a fundamentally sound theory

00:03:37.280 --> 00:03:43.040
because turbulent flow does help to prevent the formation of boundary layers

00:03:41.280 --> 00:03:47.200
a boundary layer is a thin zone where the friction between the fluid and the

00:03:44.720 --> 00:03:51.280
solid slows down the flow causing the already warm fluid to kind of stick to

00:03:49.200 --> 00:03:54.319
the copper slowing down thermal transfer you can see this in action for yourself

00:03:52.640 --> 00:03:57.760
with a simple experiment the room you're sitting in right now almost certainly

00:03:55.920 --> 00:04:01.200
contains weak air currents but you probably can't feel them this is thanks

00:03:59.760 --> 00:04:06.720
to the natural friction between your body and the surrounding air that creates a zone of pre-warmed air next to

00:04:04.480 --> 00:04:09.840
your skin a boundary layer simply wave your hand in the air like this to

00:04:08.080 --> 00:04:13.439
disrupt this boundary layer and you'll feel an immediate though temporary

00:04:12.159 --> 00:04:19.680
cooling effect neat right certainly but what it isn't is new

00:04:17.680 --> 00:04:24.240
fluix's first mistake was citing the deep water cooling magic to me which i

00:04:22.560 --> 00:04:28.080
was there when it was written first generation commercial water blocks from

00:04:25.840 --> 00:04:33.040
the early 2000s did suffer from lackluster performance for a number of

00:04:30.000 --> 00:04:34.960
reasons first up their simple maze style

00:04:33.040 --> 00:04:40.000
designs did almost nothing to increase surface area directly over the CPU die

00:04:37.440 --> 00:04:44.800
and second because most diy water coolers were using low pressure cheap

00:04:42.320 --> 00:04:49.520
aquarium pumps they intentionally used very wide water channels to avoid

00:04:46.960 --> 00:04:54.479
restricting flow that means that much of the water flow inside them was laminar

00:04:52.080 --> 00:04:57.840
rather than turbulent that's bad because you'll find substantial boundary layers

00:04:56.720 --> 00:05:03.120
basically well everywhere in a danger den maze 3

00:05:00.240 --> 00:05:07.680
from 2002 but then something happened our water cooling lord and savior

00:05:05.120 --> 00:05:12.720
certified extreme systems legend stu cathar forster created the little river

00:05:10.320 --> 00:05:16.320
white water it was the first commercial block that intentionally restricted the

00:05:14.720 --> 00:05:20.560
flow of water at the block inlet creating a jet of water that blasted the

00:05:19.039 --> 00:05:24.639
copper cold plate why did he do this turbulent flow to say

00:05:23.199 --> 00:05:28.880
that it was a game changer would be an understatement with a high pressure pump

00:05:26.800 --> 00:05:33.440
the white water and its successors the cascade and the storm would absolutely

00:05:31.600 --> 00:05:39.039
dominate the custom water cooling scene for the next few years the storm was

00:05:36.240 --> 00:05:45.360
super cool but devilishly complicated to manufacture the high volume version

00:05:41.199 --> 00:05:48.240
actually used a mid plate with 35 micro

00:05:45.360 --> 00:05:52.639
jets that fired into machined cups to create a hyper turbulent zone directly

00:05:50.720 --> 00:05:57.840
above the CPU die it left everything else in the dust and when truly stronger

00:05:55.280 --> 00:06:02.000
competitors finally emerged nearly all of them leaned on the same fundamental

00:05:59.919 --> 00:06:06.800
design principles out competing the storm only by increasing surface area

00:06:04.240 --> 00:06:10.000
directly over the CPU die optimizing the shape of the cooling patch for the heat

00:06:08.639 --> 00:06:14.800
spreader designs of contemporary processors or by bowing out the base for

00:06:13.120 --> 00:06:20.080
extra mounting pressure these evolutionary changes yielded a grand

00:06:17.280 --> 00:06:25.919
total 6 degree improvement over the 10-year span that we tested back in

00:06:22.639 --> 00:06:28.160
2016. since then skived copper bases

00:06:25.919 --> 00:06:32.880
where they actually shaved the copper and bended up to make the fins have

00:06:30.319 --> 00:06:36.960
taken over for cast or machined micro pins so while i haven't personally

00:06:34.960 --> 00:06:41.840
tested them i suspect that the top designs today like this one from ek are

00:06:40.080 --> 00:06:47.520
probably at least a couple degrees better than what we saw that means then

00:06:44.400 --> 00:06:49.440
long story short by my rough math the

00:06:47.520 --> 00:06:54.560
total water block performance improvements from basically flat chunks

00:06:51.759 --> 00:06:58.960
of copper in 2002 to the skived copper bases we see today is in the

00:06:56.319 --> 00:07:03.759
neighborhood of 17 degrees with all of that in mind fluix's

00:07:01.280 --> 00:07:10.160
patent-pending fluid turbulence technology has a lot of work ahead of it

00:07:06.160 --> 00:07:13.520
to beat anything by 8 to 16 degrees as

00:07:10.160 --> 00:07:16.639
they claim let alone a modern block that

00:07:13.520 --> 00:07:18.479
by the way definitely makes use of jet

00:07:16.639 --> 00:07:23.280
induced turbulence like every other decent design on the market so how's the

00:07:20.639 --> 00:07:23.280
performance

00:07:25.360 --> 00:07:31.440
this is bad not just a little bit bad like comically

00:07:30.560 --> 00:07:36.000
bad hey future line is coming at you fluix

00:07:33.919 --> 00:07:41.360
responded to our results and said that they have a new cold plate coming for

00:07:37.840 --> 00:07:44.080
12th gen it's possible nay probable that

00:07:41.360 --> 00:07:47.680
this will improve performance but the cold hard truth is that at least some of

00:07:46.160 --> 00:07:51.680
the things they're doing like accounting for the concavity of the ihs they should

00:07:50.240 --> 00:07:54.639
have known if they were serious about designing a high performance water block

00:07:53.759 --> 00:08:01.199
and they were up against a competitor that doesn't even use a skived copper base

00:07:58.960 --> 00:08:06.960
not to mention that it looks amazing where fluix is at best hoping to get

00:08:04.080 --> 00:08:11.120
closer in performance and also looks like it came from aliexpress there's not

00:08:09.520 --> 00:08:15.680
a huge market for ugly custom water cooling hardware either the truth is i i

00:08:14.160 --> 00:08:22.000
just don't make the rules and that's what they had for sale so that's what we reviewed all right back to the video

00:08:19.199 --> 00:08:28.080
with our CPU in default mode maximum package temperatures hit 85 degrees with

00:08:25.360 --> 00:08:35.440
the fluix block with an average of 82. that is over 10 degrees higher than our

00:08:31.599 --> 00:08:37.039
ek block it pulled about 10 more watts

00:08:35.440 --> 00:08:40.959
okay so fair enough but our average core clocks were

00:08:38.959 --> 00:08:45.680
actually slightly worse which would indicate to us that that higher power

00:08:42.880 --> 00:08:50.800
draw was likely due to higher thermals not higher performance cranking things

00:08:48.240 --> 00:08:54.480
up with our 12 900k in performance mode letting it turbo to high heaven and pull

00:08:52.640 --> 00:09:00.959
nearly 300 watts our ek block did manage to pull it off i

00:08:58.320 --> 00:09:06.000
mean it's hot at 97 degrees but it didn't throttle meanwhile the tri-swift

00:09:04.240 --> 00:09:10.160
throttled so quickly we didn't even bother letting it complete the full 3000

00:09:07.920 --> 00:09:13.760
sample render having not looked closely at this thing

00:09:11.920 --> 00:09:18.320
before greenlighting this review these results are

00:09:15.279 --> 00:09:20.240
much worse than i expected so bad in

00:09:18.320 --> 00:09:26.000
fact that i actually damn it where'd i put it dug up my old storm v2 from the

00:09:23.279 --> 00:09:29.839
mothballs to test against it the unit i have was manufactured under licensed by

00:09:27.839 --> 00:09:35.360
swift tech back in the mid 2000s and cost about 80 us dollars

00:09:32.560 --> 00:09:40.640
it uses a pretty janky mounting adapter that Jake made by hand but

00:09:38.480 --> 00:09:43.600
i cut that with tin snips we had very few tools back then

00:09:42.240 --> 00:09:47.040
yes at least one a

00:09:45.920 --> 00:09:50.399
roasted you tried both oh just for lulls i tried

00:09:49.200 --> 00:09:55.279
swapping it in case i was doing something wrong you're leaking you're looking flaking

00:09:52.959 --> 00:09:58.640
oh that's not optimal we have not pre-tested this

00:09:56.800 --> 00:10:03.279
we have no idea what's going to happen so whatever happens happens here we go

00:10:00.640 --> 00:10:09.839
bring on the stone okay it's not as good as it seems to be

00:10:05.760 --> 00:10:11.360
so far it's set like 73 74 on the

00:10:09.839 --> 00:10:16.000
package but the fluid hasn't heated up yet so we're gonna have to give it a little bit we can't take these results

00:10:14.079 --> 00:10:20.240
this final just yet you can see here i held the fans to speed up the fluid

00:10:18.399 --> 00:10:25.560
heating up right here and we reached a maximum of 87 but since then it has

00:10:22.800 --> 00:10:31.920
actually settled back down to 83 to 85 degrees celsius on the CPU

00:10:28.959 --> 00:10:34.560
package we landed at 84 degrees now i want to show you something here's a

00:10:33.360 --> 00:10:38.399
better look at the storm we've got our micro jet mid plate there's those

00:10:36.560 --> 00:10:42.560
finicky little bastards and then we've got the cups on the bottom you can see

00:10:39.839 --> 00:10:47.360
as i alluded to before this is a really really small cooling patch where there's

00:10:45.200 --> 00:10:51.839
actual turbulent flow it was designed for direct dye cooling on older cpus

00:10:50.160 --> 00:10:57.360
that didn't have a heat spreader over top of the dye so the fact that this

00:10:54.240 --> 00:11:00.399
performs anywhere near the fluix

00:10:57.360 --> 00:11:03.680
tri-swift which has a gigantic cooling

00:11:00.399 --> 00:11:06.880
patch and gets to take advantage of much

00:11:03.680 --> 00:11:09.279
much more modern machining techniques

00:11:06.880 --> 00:11:13.920
it's kind of hilarious clearly there's more turbulence here

00:11:11.279 --> 00:11:17.519
than here for comparison i also took apart the quantum velocity squared

00:11:16.560 --> 00:11:25.040
and oh that's interesting this does not use skiving

00:11:22.240 --> 00:11:30.959
even though ek is not using skivved fins you can see they are much finer than the

00:11:27.440 --> 00:11:32.640
ones in the fluix tri-flow oh and this

00:11:30.959 --> 00:11:36.959
is notable there's this mid-plate here this slotted

00:11:35.200 --> 00:11:43.200
mid-plate that sits over top of the fence gee i wonder what that

00:11:40.839 --> 00:11:46.880
does so then by going fluix we not only get

00:11:45.440 --> 00:11:52.399
performance that's comparable to a design that was optimized for cpus from

00:11:48.720 --> 00:11:54.320
2005 we add complexity to our loop with

00:11:52.399 --> 00:11:58.959
more potential points for leakage thanks to the y-splitter yay it's

00:11:57.040 --> 00:12:04.399
ugly as hell and yes again with the price it's a whopping

00:12:02.040 --> 00:12:09.839
230 us dollars on sale even adjusted for inflation it

00:12:07.279 --> 00:12:14.720
costs twice as much as the storm with some budget left over for our awesome

00:12:11.519 --> 00:12:16.800
new swim trunks on lttstore.com the only

00:12:14.720 --> 00:12:21.680
way that i can imagine flux justifying their turbulent juice marketing nonsense

00:12:19.680 --> 00:12:25.600
is if they've only ever tested this thing using the anemic pump that they

00:12:23.600 --> 00:12:30.480
bundle with the aio version from talking to tynan it is possible

00:12:28.160 --> 00:12:34.399
that with an extremely weak pump their design may

00:12:32.079 --> 00:12:39.040
actually help to increase turbulence in the CPU block but

00:12:36.399 --> 00:12:43.600
while i could test this thing frankly i have spent enough of both my time and

00:12:41.120 --> 00:12:48.560
money evaluating this piece of junk and the reality is that anyone spending

00:12:46.240 --> 00:12:51.279
this kind of moolah on a water block is going to drop more than 40 bucks on

00:12:50.320 --> 00:12:55.279
their pump and they're going gonna get a proper d5 so probably the only good

00:12:53.360 --> 00:12:58.959
thing that came out of this whole video is that it was an excuse for me to talk

00:12:57.040 --> 00:13:03.200
about water cooling again after so long maybe someday we'll see a wild new block

00:13:01.680 --> 00:13:07.680
design that nobody ever thought of and it'll be a quantum leap forward in

00:13:04.720 --> 00:13:10.000
performance but it is not this day if you want to get into custom loops we're

00:13:08.959 --> 00:13:14.240
going to have everything we used linked down below but for now we would probably

00:13:12.560 --> 00:13:19.600
recommend sticking with the tried and true ek block it's much less expensive

00:13:17.920 --> 00:13:23.120
yeah that's the that's the real kicker the 200 one i showed you guys before

00:13:21.760 --> 00:13:28.320
that's with all the extra bells and whistles it's like actually a lot cheaper and while they're not

00:13:25.839 --> 00:13:33.200
manufactured in the usa at least they're made in a machine shop rather than in a

00:13:30.560 --> 00:13:38.240
shoe factory and this segue was made in our segway factory to our sponsor

00:13:36.079 --> 00:13:41.920
squarespace need a website but don't have the know-how squarespace makes it

00:13:40.399 --> 00:13:45.360
easy there's a wide selection of award-winning templates all optimized

00:13:43.839 --> 00:13:50.160
for mobile so it looks great on all devices create members only content for

00:13:47.839 --> 00:13:53.920
extra revenue using squarespace's member areas grow and engage your audience with

00:13:52.320 --> 00:13:58.079
the powerful and easy to use email campaign system and if you ever need

00:13:55.760 --> 00:14:02.959
additional help squarespace also offers webinars help guides and you can contact

00:14:00.399 --> 00:14:07.760
their 24 7 customer support via live chat and email so get started today and

00:14:05.440 --> 00:14:13.360
head to squarespace.com forward slash LTT to get 10 off your first purchase

00:14:11.519 --> 00:14:19.040
if you guys enjoyed this video make sure to check out our easiest water cooled pc

00:14:16.000 --> 00:14:23.800
video we are well aware now of how much

00:14:19.040 --> 00:14:23.800
time it takes to do a proper custom loop