SKYREACH 4 MINI (S4M)

[1461748123]

I'm going to be using an i7-8800 + Noctua NH-L9i + ASRock Z370M-ITX/ac in the S4M. While waiting for the case to be shipped, I put the system together and set everything up in an open-air test setup. I went with the i7-8800 because of the lower TDP assuming a lower max temperature, but I'm seeing some disturbing temps with this processor. At stock speeds, I ran Prime 95 29.4 b7 FFT and quickly shot up to the 90's and reached 98C within about 2 minutes before I stopped it. While this was better than I saw using the stock cooler (100C in < 30 seconds), the temps are a lot higher than I expected for the 8800 @ default speed in open-air (ambient is somewhere in the mid-high 60's F). Idle temps and temps doing normal workloads are good though. I know the S4M has good ventilation, but I'm quite concerned that once I put this in the case that my temps are going to be worse. This system will be used for some heavy processing and will be used in warm ambient temperature.

I saw @CubanLegend's tests comparing the NH-L9i to the LP53 which showed the NH-L9i reaching 99C in 5 minutes, but that was on an i7-7700K and done inside the S4M-C. I figured the i7-8800 would have more headroom being a lower TDP part. I guess I was wrong because I'm seeing it hit that temp much faster, in open-air to boot.

If anyone else is using an i7-8800 + Noctua NH-L9i (or any other heatsink) with the S4M, could you post your P95 FFT results?

I know the LP53 is considered the best performing HS in this form factor and that delidding will yield a great reduction in heat, but the LP53 takes too long to get (and costs a fortune with new fan + mod required) and I don't want to delid and lose the warranty. Undervolting and/or underclocking may be the best solution.

Yeah, so I have a similar part - the i7 8400. Its a 6 core (no HT and slower clocks) and I'm seeing similar, though not quite as high temps. Last time I ran P95 for an extended period of time one of my cores got to 94 (the rest were mid 80s)

For me, its fine, because I'll only be using it for gaming, which will never come close to forcing all cores to 100%, 100% of the time, but yeah, people should not be surprised to see 65w chips pushing high temps with the NH-L9i. As good as it is, there is only so much you can do with the space available.

I really do suggest you to delid, although that might cause you to lose your warranty. The CPU is really hard to experience failure anyways, and delidding will give you up to ~25 degrees C difference. In combine with undervolting, you are looking at ~30 degrees C difference.
I'm currently using an 8700k with the noctua L9i. I replaced the slim fan with a normal 25mm one, but even at 100% utilization, stressing both CPU and FPU, the highest I get is ~90. Keep in mind that this is a 95w cpu compare to a 8700 which is a 65w cpu!

 

[Wasim]

I'm going to be using an i7-8800 + Noctua NH-L9i + ASRock Z370M-ITX/ac in the S4M. While waiting for the case to be shipped, I put the system together and set everything up in an open-air test setup. I went with the i7-8800 because of the lower TDP assuming a lower max temperature, but I'm seeing some disturbing temps with this processor. At stock speeds, I ran Prime 95 29.4 b7 FFT and quickly shot up to the 90's and reached 98C within about 2 minutes before I stopped it. While this was better than I saw using the stock cooler (100C in < 30 seconds), the temps are a lot higher than I expected for the 8800 @ default speed in open-air (ambient is somewhere in the mid-high 60's F). Idle temps and temps doing normal workloads are good though. I know the S4M has good ventilation, but I'm quite concerned that once I put this in the case that my temps are going to be worse. This system will be used for some heavy processing and will be used in warm ambient temperature.

I saw @CubanLegend's tests comparing the NH-L9i to the LP53 which showed the NH-L9i reaching 99C in 5 minutes, but that was on an i7-7700K and done inside the S4M-C. I figured the i7-8800 would have more headroom being a lower TDP part. I guess I was wrong because I'm seeing it hit that temp much faster, in open-air to boot.

If anyone else is using an i7-8800 + Noctua NH-L9i (or any other heatsink) with the S4M, could you post your P95 FFT results?

I know the LP53 is considered the best performing HS in this form factor and that delidding will yield a great reduction in heat, but the LP53 takes too long to get (and costs a fortune with new fan + mod required) and I don't want to delid and lose the warranty. Undervolting and/or underclocking may be the best solution.

There are couple of things to check in your Z370 BIOS. First check at which frequency your cores are running on under heavy load. By default both 8700 and 8700K all core boost is 4.3. If your cores are running higher than that this could mean that multi core enhancement ( MCE ) is enabled by default which could lock down all core @4.6 for non-K and @4.7 for K CPUs.

Secondly, please check whats your VCore voltage. Higher voltage mean higher temps. If its higher, you might consider bringing it down a little. My 8700k is stable at 1.16V. Your CPU may be stable at lower or higher voltage depending upon the quality of silicone.

You might also want to double check your NH L9i installation. Josh has a video on that.

Here are my temps of 8700k ( stock ) with Noctua L9i.

 

[Parge]

Josh has a video for all of life's problems.

 

[Untero]

I really do suggest you to delid, although that might cause you to lose your warranty. The CPU is really hard to experience failure anyways, and delidding will give you up to ~25 degrees C difference. In combine with undervolting, you are looking at ~30 degrees C difference.
I'm currently using an 8700k with the noctua L9i. I replaced the slim fan with a normal 25mm one, but even at 100% utilization, stressing both CPU and FPU, the highest I get is ~90. Keep in mind that this is a 95w cpu compare to a 8700 which is a 65w cpu!

8700k and 8700 are identical chips, the only difference is K is unlocked and operating at much higher base frequencies and slightly higher boost frequencies.
What that means is 8700 is decently cooler than 8700k at base frequencies and almost as hot at boost.

Edit: I'd love to be proven wrong since a cool 8700 chip is something I'd love to have in my S4M

Edit: v2.0 looks like youtube didn't want to give me any hope though.

 

[rfarmer]

I saw this posted on another forum.

Other Thoughts: +
I have noticed a bunch of reviews on just about all intel processors pointing to TDP as the measurement of power consumption.
This is inherently wrong, TDP (thermal design power) refers to the heat dissipation from the processor, not power use.
To measure power consumption, you can use "Intel Power Gadget", software written by a development team at Intel for windows, mac and surprisingly linux. The software was written specifically to measure real time power consumption on supported intel CPUs.
Using Intel Power Gadget, I measured the idle power consumption at just 14 Watts, and under full load during a Handbrake transcoding it reached a peak of 111 Watts of power consumption.

 

[Josh | NFC]

I saw this posted on another forum.

Other Thoughts: +
I have noticed a bunch of reviews on just about all intel processors pointing to TDP as the measurement of power consumption.
This is inherently wrong, TDP (thermal design power) refers to the heat dissipation from the processor, not power use.
To measure power consumption, you can use "Intel Power Gadget", software written by a development team at Intel for windows, mac and surprisingly linux. The software was written specifically to measure real time power consumption on supported intel CPUs.
Using Intel Power Gadget, I measured the idle power consumption at just 14 Watts, and under full load during a Handbrake transcoding it reached a peak of 111 Watts of power consumption.

I'll build on this.

This is true, but it is kinda like saying that in photography changing lenses doesn't change your perspective. No, changing lenses doesn't change your perspective--where you are in relationship to your subject gives you your perspective. But when you change a lens, you usually have to move in relation to your subject, thus your perspective changes. In other words it is helpful to understand a concept technically (and good) but it can be faster to explain someone a rule of thumb.

TDP envelope is not a measurement of a CPU's max power consumption, and I have never made that claim. For INTEL chips specifically for the generations I have covered TDP does a pretty good job matching what an Intel chip from Sandybridge to Skylake will draw at load. Not extreme duress or anything, just a general load.

The reason I used this measurement so liberally is because I had the very hard task ahead of me of evangelizing for ultra small form factor and DC-DC power supplies. This gave people a very easy guesstimate for gauging their system draw and if their power brick was going to be enough. I put alot of work into testing power supplies and then would match the beefy ones that had higher max load limits to the TDP recommendations I made. I have a literal crate of power supplies that I don't talk about, because I have put alot of work into "propaganda" for the few bricks I think are good.

I measure parts draw using software like Intel Power Gadget, my usb connected pro watt meter, and my handy multimeter which I can use to probe expansion cards and subtract from the total draw.

I hope this makes some sense.

As for using Prime 95

Don't use Prime 95 for anything other than COMPARING heatsinks. Prime95's purpose for me is to put wildly unrealistic loads on the CPU (which can be dangerous by the way) and heat everything up as much as you can. It is not a good indicator of how cool your system will run under real life loads. Use your real life load to determine what that is.

Check out Wasim's post above, I think that is helpful

Peace!

 

[Obelisk]

I'll build on this.

This is true, but it is kinda like saying that in photography changing lenses doesn't change your perspective. No, changing lenses doesn't change your perspective--where you are in relationship to your subject gives you your perspective. But when you change a lens, you usually have to move in relation to your subject, thus your perspective changes. In other words it is helpful to understand a concept technically (and good) but it can be faster to explain someone a rule of thumb.

TDP envelope is not a measurement of a CPU's max power consumption, and I have never made that claim. For INTEL chips specifically for the generations I have covered TDP does a pretty good job matching what an Intel chip from Sandybridge to Skylake will draw at load. Not extreme duress or anything, just a general load.

The reason I used this measurement so liberally is because I had the very hard task ahead of me of evangelizing for ultra small form factor and DC-DC power supplies. This gave people a very easy guesstimate for gauging their system draw and if their power brick was going to be enough. I put alot of work into testing power supplies and then would match the beefy ones that had higher max load limits to the TDP recommendations I made. I have a literal crate of power supplies that I don't talk about, because I have put alot of work into "propaganda" for the few bricks I think are good.

I measure parts draw using software like Intel Power Gadget, my usb connected pro watt meter, and my handy multimeter which I can use to probe expansion cards and subtract from the total draw.

I hope this makes some sense.

As for using Prime 95

Don't use Prime 95 for anything other than COMPARING heatsinks. Prime95's purpose for me is to put wildly unrealistic loads on the CPU (which can be dangerous by the way) and heat everything up as much as you can. It is not a good indicator of how cool your system will run under real life loads. Use your real life load to determine what that is.

Check out Wasim's post above, I think that is helpful

Peace!

Any recommendations on software that can run 24/7 for stress testing?

 

[Eigua]

Thanks for all the responses!

i7-8800? Am I getting something wrong..

Did you mean i7-8700 instead of i7-8800?

Yes. I don't know why I keep writing 8800 instead of 8700. I've been doing it for weeks. >_<

Yeah, so I have a similar part - the i7 8400. Its a 6 core (no HT and slower clocks) and I'm seeing similar, though not quite as high temps. Last time I ran P95 for an extended period of time one of my cores got to 94 (the rest were mid 80s)

For me, its fine, because I'll only be using it for gaming, which will never come close to forcing all cores to 100%, 100% of the time, but yeah, people should not be surprised to see 65w chips pushing high temps with the NH-L9i. As good as it is, there is only so much you can do with the space available.

Thanks! Well at least I know it's not just mine.

I really do suggest you to delid, although that might cause you to lose your warranty. The CPU is really hard to experience failure anyways, and delidding will give you up to ~25 degrees C difference. In combine with undervolting, you are looking at ~30 degrees C difference.
I'm currently using an 8700k with the noctua L9i. I replaced the slim fan with a normal 25mm one, but even at 100% utilization, stressing both CPU and FPU, the highest I get is ~90. Keep in mind that this is a 95w cpu compare to a 8700 which is a 65w cpu!

There are couple of things to check in your Z370 BIOS. First check at which frequency your cores are running on under heavy load. By default both 8700 and 8700K all core boost is 4.3. If your cores are running higher than that this could mean that multi core enhancement ( MCE ) is enabled by default which could lock down all core @4.6 for non-K and @4.7 for K CPUs.

Secondly, please check whats your VCore voltage. Higher voltage mean higher temps. If its higher, you might consider bringing it down a little. My 8700k is stable at 1.16V. Your CPU may be stable at lower or higher voltage depending upon the quality of silicone.

You might also want to double check your NH L9i installation. Josh has a video on that.

Here are my temps of 8700k ( stock ) with Noctua L9i.

Thanks for the info about the BIOS! I was watching the core speeds and it was ramping up and down the speed as needed. I've never seen the core reach 4.6. I'll double check the settings though.

Your temps look a lot better than mine. I'm definitely going to play around with undervolting then. This does sound like the easiest way to reduce temps without voiding a warranty. Unfortunately, I can't delid as I'm building this system for a friend.

The cooler is installed properly. I tried two different orientations on my board because both fit perfectly fine-- fin openings pointing toward the ram and parallel to the RAM. Without a doubt, temps were cooler with the fin openings running parallel to the RAM.

8700k and 8700 are identical chips, the only difference is K is unlocked and operating at much higher base frequencies and slightly higher boost frequencies.
What that means is 8700 is decently cooler than 8700k at base frequencies and almost as hot at boost.

Edit: I'd love to be proven wrong since a cool 8700 chip is something I'd love to have in my S4M

Edit: v2.0 looks like youtube didn't want to give me any hope though.

<facepalm> I was confused about the whole TDP thing. If I had known that, I'd have just popped for the 8700K, if for no other reason than possible future OCability should it ever be moved to a bigger case. I should have done more research. C'est la vie.

As for using Prime 95

Don't use Prime 95 for anything other than COMPARING heatsinks. Prime95's purpose for me is to put wildly unrealistic loads on the CPU (which can be dangerous by the way) and heat everything up as much as you can. It is not a good indicator of how cool your system will run under real life loads. Use your real life load to determine what that is.

Indeed. I was mainly using it as a upper threshold because I have two unknown factors here: 1) ambient temp the system is used at 2) increased temps in case as opposed to open-air. I would have been happy with a 90C high-end running FFT in my current open-air, low ambient test setup, knowing that it would have plenty of headroom to make up for the other 2 factors when doing long intensive rendering tasks.

I'll use all the great information posted here by everyone to help me figure out the next steps. Thanks!

 

[damekko]

Mail just arrived!!!

 

[Eigua]

If your cores are running higher than that this could mean that multi core enhancement ( MCE ) is enabled by default which could lock down all core @4.6 for non-K and @4.7 for K CPUs.

I've checked and the cores do sometimes go up to 4.6, but when under load, they max out at 4.3. Looking in the BIOS, there are no options for MCE or anything equivalent as those options do not appear for the non-K processors according to the manual. Makes sense.

Secondly, please check whats your VCore voltage. Higher voltage mean higher temps. If its higher, you might consider bringing it down a little. My 8700k is stable at 1.16V. Your CPU may be stable at lower or higher voltage depending upon the quality of silicone.

I've gone through all voltage options, but nothing has been able to lower the VID reported by the cores from 1.399v during FFT testing. If I limit the CPU to 85W max (uses 150W during this test stock), it's able to keep the temps much lower, but at a cost of speed, capping at 3.6GHz. I think I can find an acceptable balance here. Anyway, I'll take this over to the ASRock forum to dig into all the specifics of this board to see if I can get a proper undervolt. Thanks again!

 

[Craxas]

Mail just arrived!!!

WHERE THEM PICS AT BOI

 

[1461748123]

Mine is still on the way Considering it is estimated to deliver in

Mail just arrived!!!

Can’t wait for mine....

 

[damekko]

For your viewing pleasure guys.

@Craxas

Thank You Very Much @Josh | NFC !!!

 

[damekko]

Mine is still on the way Considering it is estimated to deliver in

Can’t wait for mine....

Hope you get it soon!

 

[20legend]

We can only look for now.. JFK customs

 

[Eigua]

I'm jealous. Waiting for the shipment notification...

 

[kittysox]

My black panels and silver bezel showed up today. Quality is outstanding!

 

[1461748123]

My black panels and silver bezel showed up today. Quality is outstanding!

Plz send pics

 

[Wasim]

I've checked and the cores do sometimes go up to 4.6, but when under load, they max out at 4.3. Looking in the BIOS, there are no options for MCE or anything equivalent as those options do not appear for the non-K processors according to the manual. Makes sense.



I've gone through all voltage options, but nothing has been able to lower the VID reported by the cores from 1.399v during FFT testing. If I limit the CPU to 85W max (uses 150W during this test stock), it's able to keep the temps much lower, but at a cost of speed, capping at 3.6GHz. I think I can find an acceptable balance here. Anyway, I'll take this over to the ASRock forum to dig into all the specifics of this board to see if I can get a proper undervolt. Thanks again!

Also check if your fan ramp up to its max speed (2600/2700 RPM) during stress test. If it does not, check if your fan is plugged into correct ( cpu header ) or maybe your bios fan tuning utility is holding it back. Also as Josh mentioned, Prime95 stress test is totally unrealistic for ur cpu/cooler. I always use Aida64, even this stress your cpu a lot, much more than a real world load would but its necessary for testing stability. Make sure you have the latest bios. Also its worth checking which thermal paste was used and how it was applied ( one comes with Noctua L9i is good far better than cooler master stock thermal paste).

 

[damekko]

Hi guys, i need a little bit of advice... i changed my stock cpu cooler to a noctua l9i and now i’m seeing that my cpu temps are jumping from 36 degrees to 46 degrees in an instant, it does it up and down on idle... is this normal??? I never notice this pattern with the stock one :-/