The Noctua C14S is likely the best air cooler for this case. You will need to use the vented side panel though.
-
Save 15% on ALL SFF Network merch, until Dec 31st! Use code SFF2024 at checkout. Click here!
Chimera Industries Cerberus: The 18L, mATX, USA-made enclosure
- Thread starter PlayfulPhoenix
- Start date
Hey man! Amazing setup!! I would like as well to put my 2 GPUs under water, but I think is too much because of the cpu OC.Finally finished the full loop, so thought I should drop some impressions. After a severely mediocre experience with Watercool, I ordered an XSPC block for my RTX 2080FE. I wasn't sure what to expect, but damn, it sure feels nice. It was on the expensive side (likely partially due to the LEDs included that I didn't install) and I was hoping to save some dough by going all acetal, but the Razor Neo block fit the bill and I was ready to wrap up the project. Also, I really liked that I could use the stock backplate.
Temps are surprising but good. Before, with the CPU on water and the RTX stock cooler, I was getting ~70C and ~75C, respectively. Plus, the fans on the FE cooler are pretty dang loud, and the open air cooler really liked to warm up my HDDs (up to ~50C). It was extra noisy too on account of having had a defective fan.
With the block installed, I'm getting ~80C on CPU and ~40C on GPU. This is running Prime95 + Heaven for about twenty minutes or so. I feel pretty confident the temps had normalized. Honestly, I didn't expect the GPU to be so low, especially since I'm only running the single 240 rad.
So like, not stellar, but I'm very satisfied, especially since it's dead quiet. Definitely no room for any OCs though. Even though the CPU is on the hot side (but still safe), with the lower temps between the GPU and HDDs (now 34C) - plus no noise - I think the trade-off is worth it.
Some pretty awesome builds around here. @fabio killing it with those tight clearances. Would love to see a shot with everything loaded up once you're finished.
I will post a NEW picture this afternoon. I am using the left space to add a fan hub with a passtrough sensor to monitor the water temp!
I want to use all the space!
Someone can tell me if it is a good idea to connect the radiator fans to the fan controller to control the speed based on the water temp instead of the cpu temp?
Last edited:
I don't know how much the glass sidepanel limits cooler height, but with the standard panel the Scythe Fuma 2 Rev B also fits: https://smallformfactor.net/forum/t...x-usa-made-enclosure.454/page-227#post-105316
Probably the largest TDP air cooler that can fit in this case.
Probably the largest TDP air cooler that can fit in this case.
It does appear that it fits from your photographs. However, the heat pipes are149mm height though... vs 145mm height limit per Cerberus spec.
I ordered one just because
Impressive setup!
What is your cpu? Its temperature seems high (but maybe it's a 9900K...or any cpu with severe heat bottleneck, like almost all recent intel cpu..
Just be careful your coolant does not goes above 60°C
Yes, it is a newer Intel - i7 8700k. It's only been that hot when putting both GPU and CPU under a full load. I fired up Tomb Raider for a bit and Max seemed to be mid 60s, and I've been running encodes over night and I'm stable at my typical max temp - 70C.
I'll probably run another stress test again to see if I get the same results. It's very likely it hadn't bled enough yet (I also sprung a leak shortly after).
Thanks dude. I'll keep my eyes peeled for your update. I think you'd be pushing it! With two rads I think you'd be okay, but with two GPUs (and Titans at that, right?)... I don't know that a single 240 would cut it.
I also ordered one after someone showing/saying it was compatibleIt does appear that it fits from your photographs. However, the heat pipes are149mm height though... vs 145mm height limit per Cerberus spec.
I ordered one just because
For the price, it's an awesome performer. I believe I got mine for 40€, about half the price of the Noctua NH-D15, which doesn't outperform the Fuma by much according to reviews.
Another small update!
@SweetVandal I think, I've found a sweet spot:
Room temp 22 Degrees
AVX Load at 3.6GHz (Maybe I can put 3.8GHz reducing the offset, because the temp doesn't go above 60 Degree under Prime)
16 Cores at 4.2GHz, 2 Cores at 4.4GHz
2xTitan V around 40/46 Degree in Idle thanks to the Noctua A12x25 in front of
Added an Aquacomputer QUADRO and the radiator Passthrough sensor to check the water temp and the relative deltas (Are amazing!)
Next week the build will be totally finished! (P4000 swapped with a P5000 - Of course waiting for Noctua to release BLACK fans!)
Enjoy!
Here the temps after a quick Photoshop job!
@SweetVandal I think, I've found a sweet spot:
Room temp 22 Degrees
AVX Load at 3.6GHz (Maybe I can put 3.8GHz reducing the offset, because the temp doesn't go above 60 Degree under Prime)
16 Cores at 4.2GHz, 2 Cores at 4.4GHz
2xTitan V around 40/46 Degree in Idle thanks to the Noctua A12x25 in front of
Added an Aquacomputer QUADRO and the radiator Passthrough sensor to check the water temp and the relative deltas (Are amazing!)
Next week the build will be totally finished! (P4000 swapped with a P5000 - Of course waiting for Noctua to release BLACK fans!)
Enjoy!
Here the temps after a quick Photoshop job!
Last edited:
It feels that your 8700K is somehow, not delidded and overclocked. I hate all those hidden overclock on Z370/390 motherboard....awful.
Worse thing is that only few motherboards allow to use cpu like intended by Intel (no multi enhancement, etc...).
Another small update!
@SweetVandal I think, I've found a sweet spot:
Room temp 22 Degrees
AVX Load at 3.6GHz (Maybe I can put 3.8GHz reducing the offset, because the temp doesn't go above 60 Degree under Prime)
16 Cores at 4.2GHz, 2 Cores at 4.4GHz
2xTitan V around 40/46 Degree in Idle thanks to the Noctua A12x25 in front of
Added an Aquacomputer QUADRO and the radiator Passthrough sensor to check the water temp and the relative deltas (Are amazing!)
Next week the build will be totally finished! (P4000 swapped with a P5000 - Of course waiting for Noctua to release BLACK fans!)
Enjoy!
Here the temps after a quick Photoshop job!
Dang, dude. Super slick!
Not sure what you mean here. It's not delidded, and I don't believe my board is applying an overclock - my max clock has been 4.7, which I'm pretty sure is stock turbo. Also not sure if I'm running multi-core enhancement, I'll have to check. Thanks for the heads up.
That’s the crasy thing with those z370/390 motherboard...even by default or loading just xmp profile on ram, your motherboard is applying hidden overcloking.
For z370 on core i7 8700k, mb is applyying 4.7ghz on all cores...whereas it should be only 4.3 ghz on all cores...crazy behaviour on mb manufacturers.
For info on z390 mb with 9900k, mb are applying max turbo of 4.7ghz on all cores indefinitely where intel supposed to have only a few seconds...
Globally you think you have a 95w cpu...however you get a 150w chip....with defaults settings...and sometimes not even fixable....pfff
Base clock on the 8700k is 3.70 GHz, using the default settings on my Gigabyte Z370 my 8700k sits at 4.40 - 4.70 GHz. I have never seen 3.70 GHz.
Here is my Cerberus. Very happy with the build quality and will be carrying this guy to a big LAN in January!
If your cooling is enough, your 8700k should oscillate between 4.3 and 4.7 depending of number of cores used with standard behaviour of 95w tdp.
Sure it’s turbo boost that aims to target tdp while delivering max power.
however mb manufactueres are bypassing this turbo boost by far, exploding official tdp by far...without any notice to customer.
I was under the impression tdp was “officially” measured with it off, so you could get the claimed 95w with the 4.3. While with it on you could get the upper 4.x speeds depending on the cores being used that are way above tdp provided the conditions were good.
I think all they motherboard manufacturers do this with Intel’s approval, so they can claim it’s a 95w chip, but can also reach those impressive numbers they can advertise about.
I think all they motherboard manufacturers do this with Intel’s approval, so they can claim it’s a 95w chip, but can also reach those impressive numbers they can advertise about.
In fact Intel is defining some guidelines to reach 95w tdp...as K version are overclockable, it's open bar when you are exiting 95W tdp..
My main concern (like @SweetVandal ) is that, by default, MB manufacturers are not putting standard Intel TDP. By purpose they overclocked in a hidden way (playing on turbo boost setup) K cpu. MB manufacturers are doing that in order to tweak their benchmark if early reviewers are not aware of those parameters.
It was existing before, but it mainly came back on Z370 with Asus with Multi core enhancement (putting all 8700K cores @4.7Ghz instead of 4.3Ghz).
On Z390 mb it was even worse as all MB manufacturers tweak turbo boost on 9th gen Intel cpu...by default, no motherboard was respecting intel tdp of 95W....and it worked as vast majority of reviewers (youtube), with some noticeable exception like Linus Tech Tips (who deliberately forced intel 95W tdp). What is interesting that, except Asus boards, on Z390 MB, you cannot recover a 95w profile...except by manual tweaks...
Thus, at the end, early 9900K reviews are completely misleading : 9900K appeared to be very toasty (despite soledered IHS), power hungry but powerful. However if you are respecting intel 95W tdp, 9900K is a very good cpu (not speaking of price & avaibility), cooler than 8700K (not delidded one) and more powerful than 2700X and 8700K.
I'm pretty sure Intel is pushing such hidden overclocking in order to compete (not on raw power) with R7 2700X that is still mainstream cpu king (with R5 2600..
).
Last edited:
@MarcParis - too many posts to quote now, but...
Yes, that seems to be what I've been seeing. I did see that all of my cores hit a max of 4.7. After doing some light reading, it looks like Turbo Boost is only supposed to push a single core to 4.7. I wasn't actively monitoring them, so I wasn't sure if perhaps it was playing round-robin with each core or if it had pushed each on to 4.7 at once. For me, it's not something I'm especially concerned about. I have enough power overhead and the cooling (evidently) to keep it stable (I'll run encodes for 30+ hours, so 100% load for that whole stretch).
I had glanced at my TDP/power draw stats before catching up here and did see that my max(and I think average) was well above the 95W, so my motherboard is probably doing what you've indicated. I don't remember well enough to take a guess as to what it actually was, though. Thanks for the heads up. Learn something new all the time, I guess.
Edit: maybe it's worth pointing out that encoding is hungry for threads - so that 100% util is ~100% on all 12 threads.
Yes, that seems to be what I've been seeing. I did see that all of my cores hit a max of 4.7. After doing some light reading, it looks like Turbo Boost is only supposed to push a single core to 4.7. I wasn't actively monitoring them, so I wasn't sure if perhaps it was playing round-robin with each core or if it had pushed each on to 4.7 at once. For me, it's not something I'm especially concerned about. I have enough power overhead and the cooling (evidently) to keep it stable (I'll run encodes for 30+ hours, so 100% load for that whole stretch).
I had glanced at my TDP/power draw stats before catching up here and did see that my max(and I think average) was well above the 95W, so my motherboard is probably doing what you've indicated. I don't remember well enough to take a guess as to what it actually was, though. Thanks for the heads up. Learn something new all the time, I guess.
Edit: maybe it's worth pointing out that encoding is hungry for threads - so that 100% util is ~100% on all 12 threads.
Similar threads
