Cryorig C7 vs. Noctua L9i (with and without fan duct)

[marcv81]

Hi team,

I considered the C7 and the L9i for a DAN A4 SFX v3 build. I didn't know which one I would prefer so I tried both. I'm sharing my findings in case it helps someone.

TL;DR: If you can tolerate the weird fan noise get the C7. Otherwise get the L9i but make a fan duct.

Temperature

I used a i7-8700 and a dual fan 1060. I ran Prime95 and Furmark simultaneously for 30 minutes and kept the highest moving average temperatures for the CPU, GPU and SSD over a 2 minutes period. I repeated this 3 times for each cooler on distinct days and kept the average. The results were very consistent, always within 1C of the average. The aircon was set to 27C.

Cryorig C7:
- CPU: 78C
- GPU: 71C
- SSD: 53C

Noctua L9i:
- CPU: 88C
- GPU: 70C
- SSD: 63C

Noctua L9i with fan duct:
- CPU: 79C
- GPU: 70C
- SSD: 55C

The CPU cooler did not change the temperature on the GPU side. The C7 performed the best. The CPU and SSD were 10C hotter with the L9i than with the C7. After adding a fan duct the CPU and SSD were 8-9C cooler, but still 1-2C hotter than with the C7.

Sound

This section is subjective; I did not measure the sound level.

The C7 sits too close to the DAN A4 SFX side panel. This creates audible turbulence. I find it disturbing, as if the fan was faulty. The L9i is a bit quieter. More importantly it sounds like one should expect from a fan.

The L9i fan adjusts its speed progressively. The C7 is prone to sudden changes in speed with CPU activity spikes. I find it slightly unpleasant.

Fan duct

After toying with cardboard prototypes I ordered a laser cut fan duct (USD15). Each layer was cut on a 5mm acrylic sheet, for a total thickness of 10mm. I used the longer screws provided with the L9i (M3 x 26mm). The screws should not be tightened too much to avoid cracking the acrylic.

 

[jtd871]

You need to specify what your CPU is and what your fan curves are. I barely hear my C7 on my i5-7500, so I respectfully disagree about the fan noise being turbulent.

 

[marcv81]

I used an i7-8700 with the long term power limit set to 65W. For a fair thermal comparison the fans were forced to their maximum speed.

I concede that this test is not representative of real use. With the default fan curve both the C7 and the L9i are virtually silent when browsing. When gaming both the C7 and the L9i start to become audible. At this load the C7 only emits a slight high-pitched noise. I only get a loud whistle when running CPU-intensive tasks for more than 15 minutes. The i7-8700 has 6 cores and will quickly reach its 65W TDP limit with most CPU-bound workloads.

I am not bothered that the C7 is loud, and to be fair the L9i is not that quiet either. What bothers me is that the C7 sounds as if defective. I plan to bring my computer to the office to speed up heavy workloads and I don't want to get the looks.

@jtd871, I think we can agree that whether one can tolerate the C7 is down to build specs, use cases, and personal preferences. I'd still recommend the L9i to anyone who can be bothered to attach a fan duct.

Finally to be fair to Cryorig there is nothing wrong with the C7. I can make the L9i whistle when I run it outside the case an place the side panel right next to it. It just happens that in the DAN A4 SFX the C7 fan sits close to the side panel while the L9i has more space.

 

[sam_no90]

What's you cpu clock when running P95? 3.6GHz?
The temps looks even lower than a delid 8700 @ 4.3GHz

 

[MarcParis]

Thanks for sharing your experience.

Airduct is a well-known and effective cooling method, providing nice airflow/cooling and avoiding turbulence.

Stupid questions :

• what will be noctua L9i with 25mm fan instead?
• On noctua l9i is it better with fan as intake or exhaust mode? (in exhaust mode, generally, overall temperature (except cpu) is better. Normally exhaust mode is the best for crammed case..

 

[Brock]

Thanks for sharing your experience.

Airduct is a well-known and effective cooling method, providing nice airflow/cooling and avoiding turbulence.

Stupid questions :

• what will be noctua L9i with 25mm fan instead?
• On noctua l9i is it better with fan as intake or exhaust mode? (in exhaust mode, generally, overall temperature (except cpu) is better. Normally exhaust mode is the best for crammed case..

While I was waiting on my Dan case I had my itx Mobo in my atx case (s340 elite) with the l9i on exhaust mode and when I got my Dan case I switched it to intake mode thinking because the fan was so close to the side of the case it could pull cool air in easier and my temps are cooler in my dancase than they were in the nzxt. Not too scientific but just my observation .

 

[MarcParis]

While I was waiting on my Dan case I had my itx Mobo in my atx case (s340 elite) with the l9i on exhaust mode and when I got my Dan case I switched it to intake mode thinking because the fan was so close to the side of the case it could pull cool air in easier and my temps are cooler in my dancase than they were in the nzxt. Not too scientific but just my observation .

In fact, you are right. Gamer's nexus experienced the same behaviour on ITX testing.
In dan case, as your cpu cooler fan is really close to panels vents, you are creating a fresh air duct, whereas in nzxt S340, your fan is just cycling used air. Definitively, noctua l9i is not suited for big case..

 

[Brock]

In fact, you are right. Gamer's nexus experienced the same behaviour on ITX testing.
In dan case, as your cpu cooler fan is really close to panels vents, you are creating a fresh air duct, whereas in nzxt S340, your fan is just cycling used air. Definitively, noctua l9i is not suited for big case..

Ya. I figured as much. The thought process was I'll have the Mobo put together so I can just toss it in the Dan case real quick when it gets here... Still spent 2+ hours building it trying to get the 545 aio in there until I gave up. Lol. You happen to know of anyone that has compared temps between the l9i and 545lc?

 

[deadline]

I apologize for necroing this thread, but is it important that this fan duct is more or less touching the side panel, or is - say - a 5-7mm clearance okay?

 

[Thehack]

You want 2mm or less.

 

[deadline]

You want 2mm or less.

My plan is to use a Cryorig C7 cu (ordered) with a 25mm Noctua fan. Apx 57mm height in total, in a Louqe Ghost that will fit up to 66mm height. That means I'll go for the same set as @marcv81, but with each piece 4mm instead of 5mm. Would this be correct you reckon? Thanks!

 

[Thehack]

My plan is to use a Cryorig C7 cu (ordered) with a 25mm Noctua fan. Apx 57mm height in total, in a Louqe Ghost that will fit up to 66mm height. That means I'll go for the same set as @marcv81, but with each piece 4mm instead of 5mm. Would this be correct you reckon? Thanks!

That sounds about right. Though more than likely you'll need one 5mm piece and on 3mm piece, since they normally come in 3mm and 5mm width, at least the ones i've seen.

 

[deadline]

That sounds about right. Though more than likely you'll need one 5mm piece and on 3mm piece, since they normally come in 3mm and 5mm width, at least the ones i've seen.

Gotcha, thanks!

 

[deadline]

The company that can print them asked for DXF - files. Is there anyone who knows where I can get the ones from threadstarter? He hasn't been online for a while... I can pay if someone would create them for me

 

[Narwhal]

I used an i7-8700 with the long term power limit set to 65W. For a fair thermal comparison the fans were forced to their maximum speed.

I concede that this test is not representative of real use. With the default fan curve both the C7 and the L9i are virtually silent when browsing. When gaming both the C7 and the L9i start to become audible. At this load the C7 only emits a slight high-pitched noise. I only get a loud whistle when running CPU-intensive tasks for more than 15 minutes. The i7-8700 has 6 cores and will quickly reach its 65W TDP limit with most CPU-bound workloads.

I am not bothered that the C7 is loud, and to be fair the L9i is not that quiet either. What bothers me is that the C7 sounds as if defective. I plan to bring my computer to the office to speed up heavy workloads and I don't want to get the looks.

@jtd871, I think we can agree that whether one can tolerate the C7 is down to build specs, use cases, and personal preferences. I'd still recommend the L9i to anyone who can be bothered to attach a fan duct.

Finally to be fair to Cryorig there is nothing wrong with the C7. I can make the L9i whistle when I run it outside the case an place the side panel right next to it. It just happens that in the DAN A4 SFX the C7 fan sits close to the side panel while the L9i has more space.

what are the cpu clocks?

 

[marcv81]

@sam_no90, @Narwhal, please correct me if I'm wrong but as far as I understand the CPU clock speed should not matter. I set the TDP limit to 65W; the chipset adjusts the CPU clock down to make the estimated power consumption fit within the desired limit. As far as I remember the CPU clock was in the 3.6-3.9GHz range, but if my understanding is correct I theoretically could have ran a benchmark which generates less heat per instruction and reaches 65W at 4.3GHz. If someone knows better please let us know!

@deadline, PM me for the files maybe?

 

[marcv81]

The files can now be found here: https://github.com/marcv81/l9i-fan-duct.

 

[deadline]

The files can now be found here: https://github.com/marcv81/l9i-fan-duct.

You are a star!

 

[Narwhal]

@sam_no90, @Narwhal, please correct me if I'm wrong but as far as I understand the CPU clock speed should not matter. I set the TDP limit to 65W; the chipset adjusts the CPU clock down to make the estimated power consumption fit within the desired limit. As far as I remember the CPU clock was in the 3.6-3.9GHz range, but if my understanding is correct I theoretically could have ran a benchmark which generates less heat per instruction and reaches 65W at 4.3GHz. If someone knows better please let us know!

@deadline, PM me for the files maybe?

we were wondering about your clocks because in my case I get much hotter temps with the ssame cpu and as I understand it the cpu is 65w base (3.2ghz) and peaks at 122w when turboing at 4.3

 

[marcv81]

Thanks for your reply @Narwhal. Now I see where you're coming from.

I'm not sure what the TDP means. My GTX 1060 has a 120W TDP and never ever exceeds it. The i7-8700 on the other hand happily exceeds its TDP under heavy loads.

I take the processor TDP as an indication for OEMs and system builders about what cooler size they should pick. I checked 3 different laptop models; they all have their long term power cap set to the processor TDP, and their short term power cap set to 25% more. I decided to try a similar approach.

I did some performance benchmarks on workloads that matter to me: compiling the Linux kernel, and training an XGBoost model I had at hand at the time.

80W short term cap / 65W long term cap
- Linux: 95s
- XGBoost: 61s

No power cap
- Linux: 86s, 10% faster
- XGBoost: 58s, 5% faster

As expected the system performs a bit better with no power cap. The difference is however not important enough to me and I stuck to the 80W/65W power cap. These are the settings I used to run the temperature tests presented in the opening post.

You say the processor is 65W @ 3.2GHz and 122W @ 4.3GHz. I have not tried to measure the power consumption when limiting the frequency but I believe you. I took an opposite approach: I limited the power consumption and measured the performance.