Question Scythe Mugen 5 Rev.B temps on a 3800x

[Jason H.]

So I was using a Cooler Master Hyper T2 on my 2600, then I upgraded to the 3800x and the T2 was useless, as the Prism cooler was much better.

So I ended up buying the Scythe Mugen 5 Rev.B for my 3800x.

With the Prism using Furmark CPU Burn Test (Im just looking for instant results) I was maxing anywhere from 84-90c depending on if I was just testing after boot up, or after Ive been gaming all night.

On the Scythe Im maxing at about 78c using the same test. Which seems a bit high to me BUT I have an OC of 4.3ghz all core with 1.375v at idle and 1.33v under load. SO Im not sure if this higher temp using this better cooler is because of the OC or not but either way it doesnt seem like a major differance from the prism.

I also tried Prime95 using SMALL FFT with 16 threads. My temp went to 102c within 1 minute so I stopped the test. Im not sure if thats the kind of test I should run or not to test this new air cooler so any advice on all this would be great.

My main questions are though

• Do these temps sound normal for this CPU with this cooler with the 4.3ghz OC?
• Am I running the right Prime95 test?

SPECS

MOBO: MSI Tomahawk B450
CPU: Ryzen 7 3800x (4.3ghz 1.375v)
Cooler: Scythe Mugen 5 Rev.B
GPU: Gigabyte RTX 2060 Super OC 3x Windforce White (8pin only version)
MEMORY: HyperX RGB 2x8gb DDR4 3200mhz (OC'd)
Storage: Samsung 860 Evo 250gb / 500gb Seagate / 1tb Samsung 860 Qvo
PSU: Corsair cx550m
WIN: Windows 10 Pro
Case Cooling: Antec Prizm 3x 120mm rgb case fans with controller and 2 RGB strips / 1 default 120mm case fan
Case: Cougar MX330

 

[Phaaze88]

What happens when you remove the overclock?

 

[Jason H.]

What happens when you remove the overclock?

I went and removed the OC by turning Core back to Auto and Voltage back to Auto, also turned PBO back on from Disabled to Auto.

Using the quick Furmark CPU burner, HWMonitor shows its using 5w less (from 110w max to 105w max) at 4.15ghz under load with 1.3v (thats a 0.03 drop). The temp seemed to have MAYBE a 1c differance from having the manual OC enabled.

Also, when at idle, the voltage is crazy weird jumping from 1.1v -1.495v and single random cores boost up to 4.55ghz.

So all in all, no differance in thermals but weird voltages.

What does this mean for the cooler? I was under the assumption it was far better than the Prism?

EDIT: idle temps are about 3-4c hotter. minimum before was 37-38 now they are 40-41, even before the test.

 

[Phaaze88]

How can I make this simple, hmm...

1)Those 'spikes' you see are normal for Ryzen 3000. It's a different beast from your old 2600.
It is FREQUENTLY waking up and putting to sleep cores/threads as it needs them. Intel's chips do the same at stock operation, but not nearly as often.

2)Before the Agesa ABBA(?) bios changes, the voltage spikes weren't quite as high, but people were complaining about the cpu's not hitting their advertised boost clocks - AMD remedied that... the voltage is still within spec, according to them.

3)The cpus are temperature sensitive in a similar manner to Nvidia's Gpu Boost.

4)PBO is essentially a more aggressive version of the stock settings. It probably needs better cooling in order to take advantage of it though.

Here's what I would suggest trying:
-Disable PBO, and run Cinebench R20 single and multi-thread tests. Take note of the scores as well as your temps.
-Reapply your overclock, and do the same 2 runs.
Compare the results.

 

[Jason H.]

How can I make this simple, hmm...

1)Those 'spikes' you see are normal for Ryzen 3000. It's a different beast from your old 2600.
It is FREQUENTLY waking up and putting to sleep cores/threads as it needs them. Intel's chips do the same at stock operation, but not nearly as often.

2)Before the Agesa ABBA(?) bios changes, the voltage spikes weren't quite as high, but people were complaining about the cpu's not hitting their advertised boost clocks - AMD remedied that... the voltage is still within spec, according to them.

3)The cpus are temperature sensitive in a similar manner to Nvidia's Gpu Boost.

4)PBO is essentially a more aggressive version of the stock settings. It probably needs better cooling in order to take advantage of it though.

Here's what I would suggest trying:
-Disable PBO, and run Cinebench R20 single and multi-thread tests. Take note of the scores as well as your temps.
-Reapply your overclock, and do the same 2 runs.
Compare the results.

Well, I figured out what to do on Prime95 and I had to disable AVX and AVX2. I also did a Cinebench test on multi and single cores with the manual oc but Prime95 is the only one I did with and without the manual oc.

The results of the Cinebench with the manual oc was a Multi core score of 5003cb (83c) and single core of 492cb (56c). This was at 4.3ghz 1.37v idle, 1.308v under load.

The results of the Prime95 test were 92c with the manual OC, and 83c without the manual OC. Without the manual OC it was running PBO at 4.15ghz and 1.1-1.3v.


Any good for this cooler?

 

[CountMike]

It's possible that your MB, actually it's VRM, are holding you back and that may cause erratic voltages.
Another problem could be HWmonitor, not exactly best for Ryzen, HWinfo https://www.hwinfo.com/download/ can give you more accurate readings.

 

[Jason H.]

It's possible that your MB, actually it's VRM, are holding you back and that may cause erratic voltages.
Another problem could be HWmonitor, not exactly best for Ryzen, HWinfo https://www.hwinfo.com/download/ can give you more accurate readings.

The voltages even in the BIOS were at 1.44v in the hardware monitor section when using auto settings with PBO.

I also have hwinfo64 and it gives the same readings as hwmonitor.

I mean is 92c with prime95 with the manual oc normal for this cooler? Or 83-85c on Cinebench?

 

[CountMike]

The voltages even in the BIOS were at 1.44v in the hardware monitor section.

I also have hwinfo64 and it gives the same readings as hwmonitor.

I mean is 92c with prime95 with the manual oc normal for this cooler?

You have OC and torturing it with Prime is sure to lift temps sky high, It would take very beefy liquid cooler to lower it. This is not safe at all.
Another question, why are you OCing it like that ? Keeping it at 4.3 GHz is more of an underclock . CPU is capable to boost more than that with less heat. You may be actually loosing some performance, specially in single core/thread applications.

 

[Jason H.]

You have OC and torturing it with Prime is sure to lift temps sky high, It would take very beefy liquid cooler to lower it. This is not safe at all.
Another question, why are you OCing it like that ? Keeping it at 4.3 GHz is more of an underclock . CPU is capable to boost more than that with less heat. You may be actually loosing some performance, specially in single core/thread applications.

My Cinebench numbers go up with the manual OC by 200 on multi and like 80 on single. With stock settings and PBO, the clocks only reach 4.15 at 1.3v. under load during the tests. Only time I see 4.5ghz at 1.44-1.49v is when its idle.

so 92c aint safe? But the max temp is 95? And I would expect a higher temp while doing prime95 with the OC. Without the OC is 83-85c.

And its 4.3ghz on all cores. The 4.5ghz boost is only on single cores and it makes the voltages reach up to 1.49 in some cases. But I was told that is still within spec with PBO.

While gaming I seen a MAX temp of 75c with manual OC but mostly stayed around 60c.

 

[CountMike]

My Cinebench numbers go up with the manual OC by 200 on multi and like 80 on single. With stock settings and PBO, the clocks only reach 4.15 at 1.3v. Only time I see 4.5ghz at 1.44-1.49v is when its idle.

so 92c aint safe? But the max temp is 95? And I would expect a higher temp while doing prime95 with the OC. Without the OC is 83-85c.

And its 4.3ghz on all cores. The 4.5ghz boost is only on single cores and it makes the voltages reach up to 1.49 in some cases. But I was told that is still within spec with PBO.

Within specs but not safe in long run. Best boost for PBO is reached at 62-65c
95c is temperature when CPU should shut off, not some number for it to throttle down so 92c is far too close for safety and performance.
Another problem could be your MB, it's not really suitable for OCing an 8 core Ryzen, 3600(x) at most although that MB is one of best b450 boards as far as VRM is concerned. My MB has at least 50% better VRM and still has problems reigning in 3700x at over 4.3GHz.

 

[Jason H.]

Within specs but not safe in long run. Best boost for PBO is reached at 62-65c
95c is temperature when CPU should shut off, not some number for it to throttle down so 92c is far too close for safety and performance.
Another problem could be your MB, it's not really suitable for OCing an 8 core Ryzen, 3600(x) at most although that MB is one of best b450 boards as far as VRM is concerned. My MB has at least 50% better VRM and still has problems reigning in 3700x at over 4.3GHz.

Hmm good info.

Well I did some more testing (I am now using Ryzen Master to test OC's as its faster for testing). And I got my manual voltage at 4.3ghz down to 1.325v at idle, and 1.26v under load and the new temps are pretty decent.

86c with Prime95 small fft. 77c with Furmarks cpu burner test. And 79c with Cinebench multi core test.That is a 6-8c drop on all tests from the voltage change.

Does that seem more acceptable for this particular cooler?

 

[CountMike]

Hmm good info.

Well I did some more testing. And I got my manual voltage at 4.3ghz down to 1.325v at idle, and 1.2v under load and the new temps are pretty decent.

86c with Prime95 small fft. 77c with Furmarks cpu burner test. And 79c with Cinebench multi core test.That is a 6-8c drop on all tests from the voltage change.

Does that seem more acceptable for this particular cooler?

Yes, that's acceptable for OC.

 

[Jason H.]

Yes, that's acceptable for OC.

Idk why I just thought this cooler would perform a bit better. I mean even the Prism kept the cpu at 88-89c with the furmark cpu burner test, and that was with the same exact manual oc of 4.3ghz at 1.37v, and the new Scythe cooler only dropped the temp by like 5-7c because with that same oc it leveled off at like 82-83c. But now, with the lowered voltage, Im seeing the 10-11c drop that I originally expected. So I could expect if I used the stock Prism cooler with the lower voltage it would do pretty good.

SO that just leads me to think that the stock cooler isnt THAT much worse than this cooler.

 

[CountMike]

Idk why I just thought this cooler would perform a bit better. I mean even the Prism kept the cpu at 88-89c with the furmark cpu burner test, and that was with the same exact manual oc of 4.3ghz at 1.37v, and the new Scythe cooler only dropped the temp by like 5-7c because with that same oc it leveled off at like 82-83c. But now, with the lowered voltage, Im seeing the 10-11c drop that I originally expected. So I could expect if I used the stock Prism cooler with the lower voltage it would do pretty good.

SO that just leads me to think that the stock cooler isnt THAT much worse than this cooler.

Stock cooler is generally fine for normal use except OC. Having smaller mass it saturates with heat faster and long runs of high usage tend to overheat faster.

 

[Jason H.]

Stock cooler is generally fine for normal use except OC. Having smaller mass it saturates with heat faster and long runs of high usage tend to overheat faster.

So at full stock settings, PBO disabled, the Scythe Mugen 5 max temp on prime95 small fft is 82c.

Is that good or not.

 

[Phaaze88]

That's good, IMO. Remember, Prime95 loads aren't realistic, and are more indicative of a worst case scenario.

 

[Jason H.]

That's good, IMO. Remember, Prime95 loads aren't realistic, and are more indicative of a worst case scenario.

Yea my gaming temps on avg are about 65. My only issue is thats only a 5c drop from stock cooler, and when I was using the stock cooler, I had the 4.3ghz all core oc at 1.37v...

And in prime 95 on stock cooler with the all core OC the temps were 92-95. On the Scythe the all core 4.3ghz OC is 84-86c.

I see all over the internet the Stock cooler on prime95 at stock settings is like 80c, when I use stock settings on this scythe, its 82c......

See my problem here lol?

But then I see videos like this, where the 3800x on a 4.4ghz OC with a custom water loop, aida64 hits 76c. Which then mine hitting 84c on air cooling at 4.3ghz, in comparison, doesnt seem so odd. The odd thing is that this cooler just seems to be under performing compared to the stock cooler as other videos show the stock cooler hitting max 77c on prime with stock settings.. skip to 3 minute mark.

View: https://www.youtube.com/watch?v=-6VlrlS0f4Q&t=26s

 

[Phaaze88]

Yea my gaming temps on avg are about 65. My only issue is thats only a 5c drop from stock cooler, and when I was using the stock cooler, I had the 4.3ghz all core oc at 1.37v...

I might not have read your statement here correcty, but I believe the reason has to do with Ryzen's boost behavior - again, very similar to Nvidia's Gpu Boost - the Mugen 5 improved temps enough to allow the cpu to boost slightly higher.
Of course, that means power and temps go up also - just enough to keep from crossing another threshold.
5C may not seem like much, but it's a decent drop, plus the cooler is quieter, yes?

And in prime 95 on stock cooler with the all core OC the temps were 92-95. On the Scythe the all core 4.3ghz OC is 84-86c.

Well, it was the difference between dangerously hot, and borderline ok temps... that's pretty significant, IMO.

I see all over the internet the Stock cooler on prime95 at stock settings is like 80c, when I use stock settings on this scythe, its 82c......

You also have to consider that there are other variables affecting yours, and theirs' temps, like ambient temp(small impact) and case airflow.

 

[Jason H.]

I might not have read your statement here correcty, but I believe the reason has to do with Ryzen's boost behavior - again, very similar to Nvidia's Gpu Boost - the Mugen 5 improved temps enough to allow the cpu to boost slightly higher.
Of course, that means power and temps go up also - just enough to keep from crossing another threshold.
5C may not seem like much, but it's a decent drop, plus the cooler is quieter, yes?


Well, it was the difference between dangerously hot, and borderline ok temps... that's pretty significant, IMO.


You also have to consider that there are other variables affecting yours, and theirs' temps, like ambient temp(small impact) and case airflow.

My case airflow never exceeds 32c and even is sometimes 30c. Ambient temps, I mean its like 68-74 degrees F in my room most times so I mean that doesnt seem like a big deal. I have 2 intake, and 2 exhaust fans. I mean they aint no Noctua fans but they are Antec and they seem to move a decent amount of air for being rgb fans.

So, in your personal opinion, these temps, compared to the stock cooler temps, seem on par for what the cooler is?

Someone else is telling me its too high.

 

[Phaaze88]

So, in your personal opinion, these temps, compared to the stock cooler temps, seem on par for what the cooler is?

Yes.

 

[CountMike]

Caution !!
Prime 95 is a torture test meant to be used to force as much load as possible to find weak spot and should be used only for troubleshooting if you already have indications of something bad going on. Not many applications can force so much load. Every time you use it it's same as using computer normally for months if not years and it's overuse can lead to real HW problems, shortening of component life and even burnouts.
I think I mentioned it before, tests showed that Ryzen gets best PBO boost under 62-65c. If you OC all cores temp is allowed to go higher as PBO is not working at that time so 75 -80c is common and "normal" providing voltage is kept in check. Personally I wouldn't allow it any higher as there are also implications to other components like VRM and chipset temps.

 

[Phaaze88]

^Yeah... that's basically why I stopped using it to stress test with - using both Asus Realbench and Cinebench R20 'infinite loop' instead.

 

[Jason H.]

^Yeah... that's basically why I stopped using it to stress test with - using both Asus Realbench and Cinebench R20 'infinite loop' instead.

These are my HWInfo stats while playing apex legends with the 4.3ghz manual oc with 1.325v for 2 hours.

Replace the current 40's cpu temps with 60's as when I tabbed out temp dropped. But avg temp in game is 58c, max temp was 71.4c but I never visually seen it myself, it alway stayed at like 58-61c.