Toasty 2080ti FTW3 in Cooler Master H500P Mesh

[yoadknux]

Hello everyone,

I am glad that I have something that I want to post in this forum, always like reading the opinions of experienced hardware dudes.
I managed to snag a 2080ti FTW3 Ultra recently and put it into my H500P Mesh. I tested the card for stability (stock, I will later OC), and my favorite test is the 3dmark FireStrike Ultra stress loop. It pushes the power consumption of the card to ~375W according to multiple tools.

The setup is:

So I ran this test, and I got the following temperatures:


While 78 isn't really hot for an air cooled card on a stress test, the 90c memory might be too high, and also note that the FTW3 stock fan profile is quite loud, and the fans were at high RPM during this stress test - so the temperature to noise is bad.

Next, I ran the same test with the glass panel OPEN:


and temps were a bit more reasonable. This indicates that perhaps there is something wrong with the airflow in my case. Currently, it has 2 front 200mm fans blowing air inside, a rear 140mm exhaust, and top 2x140 exhaust through a cpu radiator. On paper should be plenty of airflow, but obviously there is something wrong.

I'd like to know, any tips from you guys on how to improve the tempeartures here? I get that they aren't high, but the fans are running way too loud, I want a softer profile.

Thanks!

 

[JrClocker]

The biggest problem with the air coolers on graphics cards is that they dump the heat back into the case. You have to get the heat out of the case...or the GPU pulls in warmer air to try and cool itself.

"Plenty of airflow" is subjective in my experience (just like fan noise).

- The front intake on that case looks restrictive...are you sure you have good air intake?
- Do you have the rear exhaust fan running 100%?
- Do you have the radiator AIO fans running at 100%? (the exhaust for this on the top looks restrictive

The case picture looks pretty constricted around the PSU housing (I guess that's what it is).

Taking the side cover off and seeing large temperature drops points towards poor air flow. You can try the following:

1. Turn top radiator fans to 100% and repeat test (will probably be loud). Did you see a temperature difference? If so, you need to look at higher static pressure fans for the top exhaust through the radiator.
- The best ones I have found are the Corsair ML Pro fans. If you want LEDs, get the ML Pro LED fans, not the RGB fans as the RGB ones have their max RPM limited.
- Noctua has their new Chromax line that are pretty good and not the ugly Noctua colors
- I found the Corsair ML fans to be better in my experience

2. Remove restricting front cover from your intake fans and repeat the test.
- If you see a difference, then you need intake fans that have a better static pressure rating.


The static pressure is how much "force" the fan can apply against a restriction. It's typically measured at 0 air flow.

CFM is how much air the fan can move. It's typically tested with 0 restriction.

What separates the "men from the boys" in regards to fans is what the curve looks like that connects the above two points, and if this curve moves linearly with respect to RPM.

... Oh and Fan Noise ...

The Noctua Chromax fans are pretty quiet...so are the Corsair ML fans...all you hear is the WOOSH of air moving.

 

[yoadknux]

The biggest problem with the air coolers on graphics cards is that they dump the heat back into the case. You have to get the heat out of the case...or the GPU pulls in warmer air to try and cool itself.

"Plenty of airflow" is subjective in my experience (just like fan noise).

- The front intake on that case looks restrictive...are you sure you have good air intake?
- Do you have the rear exhaust fan running 100%?
- Do you have the radiator AIO fans running at 100%? (the exhaust for this on the top looks restrictive

The case picture looks pretty constricted around the PSU housing (I guess that's what it is).

Taking the side cover off and seeing large temperature drops points towards poor air flow. You can try the following:

1. Turn top radiator fans to 100% and repeat test (will probably be loud). Did you see a temperature difference? If so, you need to look at higher static pressure fans for the top exhaust through the radiator.
- The best ones I have found are the Corsair ML Pro fans. If you want LEDs, get the ML Pro LED fans, not the RGB fans as the RGB ones have their max RPM limited.
- Noctua has their new Chromax line that are pretty good and not the ugly Noctua colors
- I found the Corsair ML fans to be better in my experience

2. Remove restricting front cover from your intake fans and repeat the test.
- If you see a difference, then you need intake fans that have a better static pressure rating.


The static pressure is how much "force" the fan can apply against a restriction. It's typically measured at 0 air flow.

CFM is how much air the fan can move. It's typically tested with 0 restriction.

What separates the "men from the boys" in regards to fans is what the curve looks like that connects the above two points, and if this curve moves linearly with respect to RPM.

... Oh and Fan Noise ...

The Noctua Chromax fans are pretty quiet...so are the Corsair ML fans...all you hear is the WOOSH of air moving.

Thanks for the detailed comment!

I ran a few more tests this morning. All the tests were performed with the side glass panel ON.

First, I took off the front cover, so it looks like this:


And the temps for this configuration were:


Next, with the same "free intake", I also took rear & top exhaust fans to 100%. There wasn't a big difference:


And finally, for reference, I took another run with normal fan speed and front "closed":


So my conclusions are:
a) The front mesh probably limits the airflow a bit - so my intake is weak.
b) Increasing the exhaust didn't reduce the temperatures by a significant amount, at least not in a situation where the intake was good.

My options are:
a) remove the dust filter from the front and stay only with the mesh.
b) replace the 2x 200mm fans with 3x 140mm chromax fans. I have two unused ones, just need a third one. Gonna lose on looks though haha

There is in fact a third option. A unique feature of this case is that it has an additional tray with holes about 4cm after the front panel, so it could fit another 140mm (or maybe 2x 140mm) in series with the front fans. I know that in general, fans don't work well in series, but I think the distance is large enough for them to work. I'll throw in a Chromax in there and see how it does. It can actually be connected to the front of the GPU, how cool! Anyway, I'll edit this post after I perform the test.

EDIT: Here are the results with the single Chromax placed in series:


Might just be what I need to keep the case looking sharp with the noise levels low.

 

[JrClocker]

I've never seen anybody do a push/pull setup for intake fans not through a radiator...but is appears to have worked for you.

When fans are in series, the CFM does not add but the static pressure does.

So, you basically verified:

1. Your intake is restrictive
2. Your intake fans have a crappy static pressure

 

[yoadknux]

I've never seen anybody do a push/pull setup for intake fans not through a radiator...but is appears to have worked for you.

When fans are in series, the CFM does not add but the static pressure does.

So, you basically verified:

1. Your intake is restrictive
2. Your intake fans have a crappy static pressure

Thank you for guiding me through this!
I see that you have the 2080ti FTW3 as well (on water, mine is on air), what are some typical temperatures, as well as core and memory overclocks you get with those? I tried using EVGA's tool, but unless I'm missing something, curve overclock is a pain with this tool (with MSI AB it's easy, but MSI AB doesn't control the fans on this card properly).

 

[JrClocker]

The EVGA tool has a button you click that will run a routine (about 10 minutes) and determine stable overclocking for each frequency bin. I don't remember where it is, but it's in the tool and has worked well for me.

I originally had the power limit turned up to maximum on the card...to get "the most". About 2 weeks ago, I dropped the power limit to 100%...and haven't noticed any performance difference (maybe 2% or so). It's not really worth it for me to burn an extra 75 W for 2% performance.

The card is a "hybrid"...AIO for the GPU and memory, and a low noise (large) fan for the VRMs.

Temperature wise, I don't recall the memory, but the GPU hits:

- 100% power limit: 52 C max (27 C ambient temperature)
- Max power limit: 57 C max (27 C ambient temperature)

The cases I am using for both Darth Threadius and The Sith are both "high airflow" cases. Check them out and compare against what you are using to see the difference.

 

[yoadknux]

The EVGA tool has a button you click that will run a routine (about 10 minutes) and determine stable overclocking for each frequency bin. I don't remember where it is, but it's in the tool and has worked well for me.

I originally had the power limit turned up to maximum on the card...to get "the most". About 2 weeks ago, I dropped the power limit to 100%...and haven't noticed any performance difference (maybe 2% or so). It's not really worth it for me to burn an extra 75 W for 2% performance.

The card is a "hybrid"...AIO for the GPU and memory, and a low noise (large) fan for the VRMs.

Temperature wise, I don't recall the memory, but the GPU hits:

- 100% power limit: 52 C max (27 C ambient temperature)
- Max power limit: 57 C max (27 C ambient temperature)

The cases I am using for both Darth Threadius and The Sith are both "high airflow" cases. Check them out and compare against what you are using to see the difference.

I actually ended up doing something similar - instead of lowering power directly, I undervolted. The difference between the two is that with lower power limit you will never crash but always experience clock fluctuations, while with undervolt you risk crashing but clock is stable. I solved the crashing part by just running loop after loop after loop of TimeSpy, FireStrike and Port Royal.
I have been so long on water with my 1080ti, that I forgot what its like to overclock on air - pointless. Undervolting on the other hand is great.

 

[freeagent]

Looking at those front fans hurts. I would recommend a Meshify C

But you already paid for that one so..

 

[yoadknux]

Looking at those front fans hurts. I would recommend a Meshify C

But you already paid for that one so..

Yeah, I can't help but feel disappointed. It's not that 78c is dangerous or affects performance... But this is meant to be the most premium air cooled 2080ti in a 180$ (before taxes) airflow oriented case.
Of course I solved this by undervolting to 0.95V, so now the gaming temps are 60-65c... still feelsbad though.
Might just sell this case and get something else, how's the Lian-Li O11 Dynamic Mini?

 

[yoadknux]

I think it's time I update this post. Just for future research, or maybe other people with this case, or with this card, or a similar setup...



I am now at a point where I have stress test temps of 69c, an on air cooled, overclocked 375W card with a closed case, at stock fan curve and no undervolt. How? It was a combination of a few things, each lowered the temps by 1-2c:
- putting 2x Noctua NF-A14 in series with the intake fans)
- Changing Thermal paste to NT-H2 and all the pads
- Vertically mounting the GPU - 3c difference



So one thing I realized that has nothing to do with the case is that this GPU has very strong fans that pull air from below the card and exhaust it to the side. Back in the day when I had Enermax Chakra and Thermaltake Element S, this wouldn't be a problem, because those cases had side fans that could be set to exhaust. But with this case and honestly any other case on the market the side panel blocks the hot air that is exhaused from the GPU - and this is what increases the GPU temps. Vertical mounting means that the GPU fans intake air from side and exhaust to top.

Now, something that I still don't like is those goddamn MEM2 and MEM3 temps. How can the VRAM be 16c hotter than GPU core and VRMs? Turns out this is a common issue with this card (At least with the air cooled version). What I think happens - This card has a front plate that touches the memory chips. But this front plate isn't coupled to a heatsink, so it simply acts as a heat BARRIER. The right approach in my humble opinion is to just go to the "standard route": Skip this front plate and just put pads directly between VRAM and main heatsink.

Maybe I will try it someday. But not now. Mounting a TRIPLE SLOT GPU on a 45$ eBay vertical mount is so hard, and it will probably be difficult to remove too. I was happy I did it though.

 

[WhitehawkEQ]

Thanks for the detailed comment!

I ran a few more tests this morning. All the tests were performed with the side glass panel ON.

First, I took off the front cover, so it looks like this:

Is the rear of that case against the wall? Pull it out about 2-3 inches (5cm-7.6cm) for good air flow.

 

[yoadknux]

Is the rear of that case against the wall? Pull it out about 2-3 inches (5cm-7.6cm) for good air flow.

Yep, I kept a few cms from the wall. Now with the GPU dumping heat into the case (instead of dumping it into the side panel), I feel hor air actually emerging from the rear. The top is a bit restrictive - It has 2x140 but it exhausts through a radiator, and into a net/filter/glass mesh thing. I might modify something there to improve top exhaust.

 

[EarthDog]

and into a net/filter/glass mesh thing. I might modify something there to improve top exhaust.

There is no need for a filter for exhaust. All it does is slow down airflow. Remove the exhaust filter.

 

[yoadknux]

There is no need for a filter for exhaust. All it does is slow down airflow. Remove the exhaust filter.

Yes, as long as the computer is running... But when it's not running, dust can get in from the top.

 

[EarthDog]

Yes, as long as the computer is running... But when it's not running, dust can get in from the top.

Not really... Negligible at best. Remove the exhaust filter.

 

[tRidiot]

I don't think dust 'falling' in or circulating into the case through Brownian motion is a major factor. You get 1000x more dust from intake when running. In fact, any dust that randomly falls or floats in through the top when off would likely 80%+ be blown back out as soon as you turn the case on.

I would also remove the exhaust filter.