NOVA - Optimizing perf, noise and temperature

[FerrariVie]

This is a little dated by now as haven't used either for a few years, but I always used to prefer EVGA Precision X over Afterburner, just one of those personal preference things I guess.

It may be worth checking it out though to give comparative results, it may be more or less accurate (although it's now called Precision X1, that's how out of date I am ):

EVGA - Software - EVGA Precision X1ᐪᔿ

Introducing EVGA Precision X1ᐪᔿ. With a brand new layout, completely new codebase, new features and more, the new EVGA Precision X1ᐪᔿ software is faster, easier and better than ever. When paired with an NVIDIA Turing graphics card, the new EVGA Precision X1ᐪᔿ will unleash its full potential with...

www.evga.com

Always thought that it was only meant to be used on EVGA's GPUs? It indeed looks a lot nicer than Afterburner

If it has an in-game overlay, then I'm sold

 

[SpyderTracks]

Always thought that it was only meant to be used on EVGA's GPUs? It indeed looks a lot nicer than Afterburner

If it has an in-game overlay, than I'm sold

It used to be excellent, can be used on any manufacturer of GTX1600 series and up.

I have no idea what it’s like now, hopefully they haven’t messed it up!

 

[FerrariVie]

@maxff9 , posting this to help you with the repasting: (thanks @enduser for posting it in another thread)

 

[maxff9]

Nice video, thank you. It was a bit too fast to show the repasting process tough.
I am not so confident to repaste because 1) I never did it before 2) I don't even know if my thermals are good/bad.
I should first assess the quality of the paste but if I open, I'll have to replace it anyway. The observation is destructive, it's annoying.

For just browsing, the CPU temp stays at 52°C, not bad right ? I guess I'll have to run Cinebench a couple of times xD

One thing I don't really get... This NOVA has a desktop CPU but the built of the chassis doesn't deliver the adequate amount of power to the CPU. At the end, we have a component heating a lot and never reaching its full potential ? Should I feel scammed ?!

I am also trying to find a good and comfortable way to play. On the desk, the screen is too low and I have to look down to see the screen. I have a external monitor but I'd need to buy a keyboard... And my chair in my studio is the worst piece of wood I had the chance the sit on haha. Trying to take care of my back and neck those days, lot of optimization to do. But that's just me explaining my life.

 

[FerrariVie]

Nice video, thank you. It was a bit too fast to show the repasting process tough.
I am not so confident to repaste because 1) I never did it before 2) I don't even know if my thermals are good/bad.
I should first assess the quality of the paste but if I open, I'll have to replace it anyway. The observation is destructive, it's annoying.

For just browsing, the CPU temp stays at 52°C, not bad right ? I guess I'll have to run Cinebench a couple of times xD

One thing I don't really get... This NOVA has a desktop CPU but the built of the chassis doesn't deliver the adequate amount of power to the CPU. At the end, we have a component heating a lot and never reaching its full potential ? Should I feel scammed ?!

If you're having issues with thermals on this chassis, then you definitely have problems with the paste. It is designed to support up to a 12 cores Ryzen9 3900, so there's no way why it shouldn't be able to handle an R5 with half cores. But I agree that different CPUs will behave differently, as I've seen some benchmarks showing that the R7 3700x runs cooler than the R5 3600, probably because the former is now being fully utilised. The only way you can compare that accurately is by checking with another person that has exactly the same chassis and CPU/GPU as you.

I am also trying to find a good and comfortable way to play. On the desk, the screen is too low and I have to look down to see the screen. I have a external monitor but I'd need to buy a keyboard... And my chair in my studio is the worst piece of wood I had the chance the sit on haha. Trying to take care of my back and neck those days, lot of optimization to do. But that's just me explaining my life.

Don't you have a laptop stand (with or without cooling fans)? I do have a standard Amazon Basics stand that doesn't have any fans, but does have 5 different height adjustment levels. On the second-highest level, I find it on the perfect height for my desk, where I don't need to look down to see the top of the screen. And even though it doesn't have fans, it helps with thermals since it's a metal mesh that supports the laptop, so it gives it extra room to pull air.

 

[maxff9]

I should check for that stand but isn't it weird to have the keyboard tilted? I am used to have my hands flat on the keyboard.
My friend seems to have the same thermals as I have. Or we both have normal thermals or bad ones.
We'll run some tests.
Hahaha at end I only played 30min since I have my laptop... Kinda sad in a way. Can't wait to have the setup exactly as I want it.

 

[FerrariVie]

I should check for that stand but isn't it weird to have the keyboard tilted? I am used to have my hands flat on the keyboard.
My friend seems to have the same thermals as I have. Or we both have normal thermals or bad ones.
We'll run some tests.
Hahaha at end I only played 30min since I have my laptop... Kinda sad in a way. Can't wait to have the setup exactly as I want it.

I got used to having the keyboard tilted, at least for usual browsing and emails. Just for gaming that I still prefer the height positions that are a bit lower, kind of a middle term between perfect ergonomics and the lowest available position on the stand. Racing games are the only type that I actually used the lowest height available.

It's the best that I can do with the space that I have on my desk... Not enough room to use an external keyboard with the laptop screen and the only option would be to have an external monitor+keyboard, but that would also require me to do something smart with the laptop itself, like a stand that could keep it closed on a vertical position (in order to fit on the table). Not sure if that is going to be a problem for thermals though, as the laptop would have the lid closed all the time.

 

[FerrariVie]

Thanks for your dedication!
I find it quite tedious to hunt for Windows background processes. I took a look at some softwares to do it automatically but everything seems shady.
I just kill my apps such as Dropbox and other front end apps but I don't know what else could be done.

I hope afterburner doesn't have too many bugs because this would make the whole study very complicated. Or maybe something overriding it ? It's already very complicated with variable not possible to log.
I'll order the thermal paste this weekend and keep this thread updated.

I finally got the time to make note of Firestrike measurements when limiting GPU clocks:

I know this is not scientifically ideal since I've run each max clock just once, but it's kinda time-consuming to do them all 5 times or so an then take averages, and there's also the probability that running 35 tests non-stop will make components hotter and bring results down with time. But at least you can compare your results with mine

 

[maxff9]

Thank you very much!
I received the thermal past but didn't repaste yet. Still not sure if it's worth.

What would be your conclusion of your tests? What clock is best for you?

 

[FerrariVie]

Thank you very much!
I received the thermal past but didn't repaste yet. Still not sure if it's worth.

What would be your conclusion of your tests? What clock is best for you?

Looking at the graph, it seems that 1755 would be ideal when looking at temperatures x performance ratio. I'm currently using 1695Mhz most of the time, so I should give 1755 a try on real gaming to see how it looks like. Another thing that I could see from that analysis is that limiting to 1830Mhz is giving me more average clocks (and final score) than having no limits (stock) and almost the same performance as 1920 with. So I'll probably need to test those 2 again, but it case the results are similar, I'll never use the stock unlimited profile again and stick to a max of 1830.

Hey, I've repasted my Nova yesterday after running those benchmarks for you, as finally the NT-H2 was available on amazon after a few months of being out of stock.. I'll give you some initial (not very conclusive) findings later

 

[maxff9]

Nice graph by the way !

From stock to 1620 MHz: for a performance loss of 3%, you gain 3°C, it's worth it.

I could play a bit more with my laptop - but no demanding games (Borderlands 3, Phasmaphobia, Payday 2). Always playing on stock settings. The area left of the touchpad can get quite hot during gameplay. If I play and I am alone (99.99% of the time), the fans noise doesn't bother me. It's another story as soon as the game is silent... But also, I want to reduce the temps to extend the durability of the laptop. I also bought a super cheap stand (like 7€) and it does the job. The screen is higher and I don't wreck my neck looking down.
I also bought this wireless mechanical keyboard and a bluetooth mouse. I plan to play sitting in this ikea chair.
I played only once with the keyboard and I am very not sure I'll like it. It's clicky and the buttons are very sensitive. I may go to a small membrane cheap keyboard if I don't enjoy this one.

Going back to our business:
Indeed your results do not show the average clock speed, which is directly correlated with the score.
I don't get the 1755 choice. The ratio score/temp is the 2nd one. The 1st one being 1620 MHz.
I still wonder if going from stock to any settings would affect any underlying algorithm impacting the overall performance. I say that because setting the max clock speed at 1920 gives lower results than stock, while it should be the same. (at the same time and as you said, we don't have any standard deviations so it could be totally normal).
I am reconsidering to use customs settings. Probably I'll set a max clock speed to reduce a bit the temp. I am sure I won't see any difference ingame.

Overall your perfs are better than mine (for the reason right ?) and your temp are much better. I'll try to find time to open and repaste everything *stress*.
Thanks again for the discussion.

 

[FerrariVie]

Going back to our business:
Indeed your results do not show the average clock speed, which is directly correlated with the score.
I don't get the 1755 choice. The ratio score/temp is the 2nd one. The 1st one being 1620 MHz.
I still wonder if going from stock to any settings would affect any underlying algorithm impacting the overall performance. I say that because setting the max clock speed at 1920 gives lower results than stock, while it should be the same. (at the same time and as you said, we don't have any standard deviations so it could be totally normal).
I am reconsidering to use customs settings. Probably I'll set a max clock speed to reduce a bit the temp. I am sure I won't see any difference ingame.

Overall your perfs are better than mine (for the reason right ?) and your temp are much better. I'll try to find time to open and repaste everything *stress*.
Thanks again for the discussion.

The choice behind 1755Mhz is because it has better performance than 1695, while keeping around the same temperatures. Firestrike temperatures are rounded, so not sure if that was 57.4° for example, but still is a great increase in performance for around the same 57° temperatures. But only a test on a real game will tell the true difference. And yes, 3% difference in performance is going to be roughly 5 FPS (10 maximum) on a game that runs at 144 FPS, so it's not noticeable and it does improve temperatures (which is noticeable).

Overall your perfs are better than mine (for the reason right ?) and your temp are much better. I'll try to find time to open and repaste everything *stress*.
Thanks again for the discussion.

Well, those are related. I get better GPU performance because my temps are better, but my temps are better because my CPU is worse You get my point here

Again, all those tests were done before repasting and I planned to redo them after repasting, but by the time that I finished the whole process, my ambient temperature went up by 5 degrees (very unusual hot days that started exactly while I was repasting ) and that would not be representative of the truth. So I'm waiting for another day when I will have time and 20° in the room and can run the tests properly

I'll share my thoughts and some pictures about the repasting process later today

 

[maxff9]

Yes please share pictures for the repasting I'll follow because you'll bring me confidence. I'll run more tests after that.

Do you any good way to monitor the performance while gaming ? Every gaming session is different (even if it's the same game), so it's pretty hard to have consistency. And I am not sure the Firestrike mimics well real experience (actually I think it's quite good).

What games do you like to play ?

 

[FerrariVie]

Yes please share pictures for the repasting I'll follow because you'll bring me confidence. I'll run more tests after that.

Do you any good way to monitor the performance while gaming ? Every gaming session is different (even if it's the same game), so it's pretty hard to have consistency. And I am not sure the Firestrike mimics well real experience (actually I think it's quite good).

I use afterburner's overlay, you can setup a lot of different data points... I use the ones on the photo below, with the addition of FPS and CPU/GPU usage % (that I had turned off on this screenshot for some reason?), and already removed the power and voltage limit indicator.
Cyberpunk Screenshot

Just not sure if you mean saving logs? If yes, I've never used it, but I believe afterburner also saves logs:


I like Firestrike quite a lot, but it's short and real gaming won't last for just a few minutes. So because you're getting 60° in firestrike for sure won't mean that you'll get the same on gaming (I would say real gaming is like 10° higher for me?), but it's a good reference point to quickly compare different computers, as well as different setups within the same one.

What games do you like to play ?

Mainly offline games, since I don't have time to compete with ppl that spend hours and hours to get top equipment and always be stronger than everyone else. So I usually play cyberpunk (not anymore, since it's quite short and I've finished it already), Red Dead Redemption 2 and Forza Horizon 4. Forza would be the only game that I like to sometimes compete online, as competitiveness is based on having/not having a wheel and since I don't have one and play on the keyboard, I'm ok with being in the middle of the pack or losing . At least not spending many hours playing doesn't give one that much of an disadvantage and the game is quite fun

I used to play Football Manager quite a lot on my old PC due to the lack of performance to play newer games, but it got to a point where even FM was becoming too slow. But since I've moved to the Nova, I assume I've not played FM that much anymore.

 

[FerrariVie]

Regarding the repasting process, it was quite easy. Just removing some screws, clean old paste (harder part of the process), add new paste and put screws back in. The only thing to make sure you're doing is to follow the order of the numbers, inverse for removing the heatsink (8 to 1) and ascending (1 to 8) to screw it back. Oh, and don't make my mistake of forgetting to plug the GPU fan cable back

3 major takeaways from the whole repasting of the Nova was:

1. PCS is indeed using a good quality thermal paste on the CPU and the person that did mine did a perfect job . Your mileage may vary though, depending on how skilled the person that did yours was and a bit of luck

https://imgur.com/SNPhVB0

1. When talking about the GPU, however, things get complicated. It's clear that the quality of the paste is not the same (even the color is different) and it was all dried-up. My laptop is only 6 months old and I did stress test CPU and GPU for about an hour before (run Firestrike Graphics and Physics like 6 times to make that chart) in order to soften the thermal paste, which did work for the CPU. So there's no excuse, they're very likely using whatever paste came from Clevo, which should be of really bad quality to dry up after just 6 months

https://imgur.com/fywNYYO

1. Be careful with screw number 5. Just to the right side of it, there's like a plastic part that protects the CPU fan screw. In my opinion, that plastic part should not be directly below the heatsink, as it can impact the mounting pressure and prevent screw 5 from going all the way down. The heatsink and the plastic part are more or less on the same height level, so when you put the heatsink back in place and before adding screw 1 back in, make sure that the heatsink is not touching that plastic near screw 5 and keep making sure that the heatsink doesn't move until you've put screw 5 back. That's the only tricky part of the process and, for me, it's a design problem of the Nova. The plastic cover could have been smaller or lower to not impact the heatsink.

The photo doesn't look good because of the perspective (it looks like the heatsink is on top of the plastic part, but it isn't by just a millimetre), but at least can show you what I'm talking about.


So as I said, I could not assess the benefits yet as my room temperatures have gone up by quite a lot recently. But I've run firestrike anyway and the GPU temp was 2 degrees lower, which is a good sign considering the short test and the higher ambient temp. But regarding the CPU, I couldn't see any difference yet.

 

[SpyderTracks]

When talking about the GPU, however, things get complicated. It's clear that the quality of the paste is not the same (even the color is different) and it was all dried-up. My laptop is only 6 months old and I did stress test CPU and GPU for about an hour before (run Firestrike Graphics and Physics like 6 times to make that chart) in order to soften the thermal paste, which did work for the CPU. So there's no excuse, they're very likely using whatever paste came from Clevo, which should be of really bad quality to dry up after just 6 months

GPU's don't use thermal paste, they use thermal pads, entirely different thing, it's not dried up, it's an entirely different thermal product.

 

[maxff9]

Good job, I appreciate that you posted those advice.
I'll wait maybe 2-3 months before doing it as it seems PCS is doing a good job. I took some isopropanol to clean the old paste.

For sure it is useful to monitor the ambient temperature before running temperature benchmarks. Please keep us updated if you think you have better thermals.

What about the GPU ? If they used a thermal pad and we put the paste, are we messing with the thermals or it's ok ?

 

[FerrariVie]

Good job, I appreciate that you posted those advice.
I'll wait maybe 2-3 months before doing it as it seems PCS is doing a good job. I took some isopropanol to clean the old paste.

For sure it is useful to monitor the ambient temperature before running temperature benchmarks. Please keep us updated if you think you have better thermals.

What about the GPU ? If they used a thermal pad and we put the paste, are we messing with the thermals or it's ok ?

I would say by my initial testing that the paste is giving me better results than whatever was there in the first place, hence why I thought the Nova was originally using a bad thermal paste.

@SpyderTracks , you probably know more than me. But thermal pads are not the same ones used for the VRMs and those other smaller components? Like the blue ones on the picture below (GPU part of the heatsink):


Those didn't get dry and were still spongy while the GPU chip one had a circular shape (like a paste that spread out after the heatsink pressure), unless the pads used on the GPU are of a different type that kind of melted and dried out after use.

Now that I think of it, could it be the same thing that Intel used (or still use, not sure) on their stock coolers? I've had lots of Intel desktops in the past and remember seeing some circular or rectangular-shaped "thing" (don't know how to call them anymore ) that were stuck the heatsink. Always thought of those as being thermal paste, but one that's easy to store, move around and easy to apply (since they're not creamy), but focused a lot more on durability than on thermal performance:

https://imgur.com/ip2oo

https://www.legitreviews.com/wp-content/uploads/2018/02/amd-headsink-ryzen.jpg

Anyway, it's not that the GPU paste/pad/compound wasn't working, it was doing its job but it was very hard to remove and the new thermal paste seems to be working a lot better for heat dissipation.

 

[SpyderTracks]

I would say by my initial testing that the paste is giving me better results than whatever was there in the first place, hence why I thought the Nova was originally using a bad thermal paste.

@SpyderTracks , you probably know more than me. But thermal pads are not the same ones used for the VRMs and those other smaller components? Like the blue ones on the picture below (GPU part of the heatsink):
View attachment 23906

Those didn't get dry and were still spongy while the GPU chip one had a circular shape (like a paste that spread out after the heatsink pressure), unless the pads used on the GPU are of a different type that kind of melted and dried out after use.

Now that I think of it, could it be the same thing that Intel used (or still use, not sure) on their stock coolers? I've had lots of Intel desktops in the past and remember seeing some circular or rectangular-shaped "thing" (don't know how to call them anymore ) that were stuck the heatsink. Always thought of those as being thermal paste, but one that's easy to store, move around and easy to apply (since they're not creamy), but focused a lot more on durability than on thermal performance:

https://imgur.com/ip2oo

https://www.legitreviews.com/wp-content/uploads/2018/02/amd-headsink-ryzen.jpg

Anyway, it's not that the GPU paste/pad/compound wasn't working, it was doing its job but it was very hard to remove and the new thermal paste seems to be working a lot better for heat dissipation.

You can usually tell a thermal pad (even if it’s deteriorated as it sounds like this one had) as they do act as a sort of “glue” and make it quite hard to get the heatsink off. If you’ve ever taken apart a desktop GPU, you’ll know exactly what I mean, they take a surprising amount of force to separate the heatsink from the thermal pad.

Performance wise, a paste would likely always be better, I think the reason pads tend to be used on GPUs is because they don’t go to such high temps in the first place, have a larger surface area, and are assembled at source and so it speeds up application time.

Thermal Grizzly do some excellent thermal pads if you’re ever looking for some, come in all shapes and sizes and you can cut them to size if needed.

 

[FerrariVie]

You can usually tell a thermal pad (even if it’s deteriorated as it sounds like this one had) as they do act as a sort of “glue” and make it quite hard to get the heatsink off. If you’ve ever taken apart a desktop GPU, you’ll know exactly what I mean, they take a surprising amount of force to separate the heatsink from the thermal pad.

Performance wise, a paste would likely always be better, I think the reason pads tend to be used on GPUs is because they don’t go to such high temps in the first place, have a larger surface area, and are assembled at source and so it speeds up application time.

Thermal Grizzly do some excellent thermal pads if you’re ever looking for some, come in all shapes and sizes and you can cut them to size if needed.

I get it now, and that's exactly why I remembered my past desktop days... It was a bit hard to remove this heatsink as well, I could notice that the CPU side was off with no pressure at all, but the GPU was kinda glued in, but not as hard to remove when compared to a desktop CPU back in those days, but required a bit more pressure than just simply lifting it.

The confusion was just on how it was called, as I thought it was just a more solid kind of paste, but now you're telling me they're called pads. And like you said, it probably is already glued to the heatsink from Clevo, while the CPU is installed by PCS and they're probably applying the paste themselves. But since PCS is already doing one paste job, not sure why they couldn't do it for the GPU as well... but that's Clevo's fault for sending the pad glued on the heatsink.