[SOLVED] Cooling Conundrum

[cogsman]

I have just completed my build with the following components:

• Ryzen 9 5900X
• ASUS ROG Strix B550i Gaming Mini ITX motherboard
• Be Quiet Dark Rock Pro 4 Cooler
• EVGA RTX XC3 3070
• 32GB Ballistix 3600Mhz ram
• 2 Be Quiet 140mm fans
• Seasonic Focus GX850 PSU
• Thermaltake Core V21 Case
• Sabrent Rocket 4 and Sabrent Rocket Q NVME drives mounted on top and bottom of board
• WD 1TB SSD on the SATA port

My case has a 200mm front fan (~150CFM) and my configuration is basically set up this way (if you are looking at the front of the case):

1. Air blows in from the 200mm fan
2. Blows towards the back of the case
3. The motherboard is in the back right of the case
4. The DRP4 cooler has the 2 standard fans on it, cooling the CPU and oriented so that they blow towards the back of the case
5. One 140mm fan is mounted on the back of the case, right behind the cooler, continuing the flow of air out the back of the case
6. One 140mm fan is mounted on the right side near the back of the case, right next to the DRP4 cooler, blowing air out of the side of the case (my thinking here is that this area of the case is where the most heat will be concentrated)
7. So I have the 1 intake at the front, running at full speed, and the 2 140mm fans for exhaust on PWM
8. The PSU fan is facing down so it is not adding or removing any air from the case
9. The GPU's 3 fans didn't even kick on during the load test

Last night I put it through its paces.

• With 100% load, the CPU Temp was reported at 73c and the CPU Package peaked at 89c but was typically at 85c for the duration.
• The CPU boosted to 4500Mhz during the load on a couple of cores.
• I noticed that the VRM heatsink fan speed was sitting at 0 rpm, which made me wonder if there was something wrong with the VRM fan. MB temp was <50c and the chipset temp was in the low 60s.
• This was a 20 minute load test (upconverting a batch of 80 videos using MOVAVI.

I am trying to understand if all this is typical. From what I am reading some say it is typical under load, others say it's too high, some are saying it's fine... I'm not sure what I was expecting but just want to find out if this hardware is good to go. Online, I can't find a consensus on what a typical temp range is (conflicting opinions). Regarding the VRM fan, there is nothing in the specs about when it would kick in. I just assumed it would always run. Is there something in the bios I am overlooking perhaps?

I used the pea method for thermal paste when I mounted the cooler. It made contact with the CPU; I know this because although it was hard to see, I could see and feel it slide a tiny bit on the CPU and I had not added the cross-bar mounting bracket for a final screwdown yet (by the way, the installation on this thing was WAY too frigging hard on such a small board. Partially my fault for choosing an ITX build, but also read that many prefer Noctua ease of mounting).

So I thought to myself, is the cooler's middle (inside) fan spinning, or is just sitting there? Did I mount that fan incorrectly? And is there an easy way for me to check that fan is working without taking the cooler apart? In the end, with this freaking behemoth of a cooler, is 85c the best I can get under load?

Any help or advice would be appreciated.

 

[cogsman]

My current build contains:

• Ryzen 9 5900X
• Core V21 Case
• 2 Be Quiet 140mm fans
• ASUS ROG Strix B550i Gaming Mini ITX motherboard
• Be Quiet Dark Rock Pro 4 Cooler
• EVGA RTX XC3 3070
• 32GB Ballistix 3600Mhz ram
• Seasonic Focus GX850 PSU
• Sabrent Rocket 4 and Sabrent Rocket Q NVME drives mounted on top and bottom of board
• WD 1TB SSD

My case has a 200mm front fan (~150CFM) and if you are looking at the front of the case:

1. Air blows in from the 200mm fan
2. Blows towards the back of the case
3. The motherboard is in the back right of the case
4. The DRP4 cooler has the 2 standard fans on it, cooling the CPU and oriented so that they blow towards the back of the case
5. One 140mm fan is mounted on the back of the case, right behind the cooler, continuing the flow of air out the back of the case
6. One 140mm fan is mounted on the right side near the back of the case, right next to the DRP4 cooler, blowing air out of the side of the case (my thinking here is that this area of the case is where the most heat will be concentrated)
7. So I have the 1 intake at the front, running at full speed, and the 2 140mm fans for exhaust on PWM
8. The PSU fan is facing down so it is not adding or removing any air from the case
9. The GPU's 3 fans didn't even kick on during the load test

I tested it and with 100% load, the CPU Temp was reported at 73c and the CPU Package peaked at 89c but was typically at 85c for the duration.

• The CPU boosted to 4500Mhz during the load on a couple of cores.
• I noticed that the VRM heatsink fan speed was sitting at 0 rpm, which made me wonder if there was something wrong with the VRM fan. MB temp was <50c and the chipset temp was in the low 60s.
• This was a 20 minute load test (upconverting a batch of 80 videos using MOVAVI.

I am trying to understand if all this is typical. From what I am reading some say it is typical under load, others say it's too high, some are saying it's fine... I'm not sure what I was expecting but just want to see if there will be a problem. Online, I can't find a consensus on what a typical temp range is (conflicting opinions). Regarding the VRM fan, there is nothing in the specs about when it would kick in. I just assumed it would always run. Is there something in the bios I am overlooking perhaps?

I used the pea method for thermal paste when I mounted the cooler. It made contact with the CPU; I know this because although it was hard to see, I could see and feel it slide a tiny bit on the CPU and I had not added the cross-bar mounting bracket for a final screwdown yet (by the way, the installation on this thing was WAY too frigging hard on such a small board. Partially my fault for choosing an ITX build, but also read that many prefer Noctua ease of mounting).

So I thought to myself, is the cooler's middle (inside) fan spinning, or is just sitting there? Did I mount that fan incorrectly? And is there an easy way for me to check that fan is working without taking the cooler apart? In the end, with this freaking behemoth of a cooler, is 85c the best I can get under load?

As I write this, I am using AIDA64Extreme to do another test which so far has lasted 20 minutes.
@100% load, CPU is at 79c according to AI Tweaker
90c for the CPU package
80 for the chipset
45 for MB
The VRM fan did kick in
It looks like it's boosting to 4.0Ghz
GPU is 61c

Any help or advice would be appreciated.

 

[mdd1963]

Those CPU temps seem quite reasonable for a budget air cooler that in assorted Carey Holzman Youtube videos seemingly had issues keeping a 9700K or 9900K under 85C in Prime 95/balanced, which , as we know, is less stressful than Prime95/small FFTs.

You might be simply overestimating the abilities of any cooler, air or fluid, to to instantly lower core/package temps, which are largely determined by clock speeds, core voltage and type/severity of load. As long as it it not throttling, or having temps gradually rising to near that point, I'd say you are good.

 

[drea.drechsler]

AMD - Robert Halleck specifically, technical marketing rep - has stated it's expected that 5000 series processors will run up to 90C or even 95C in extreme workloads. So that sounds about right.

AMD CPU Shipments were Down by up to 40% in Q3; Higher ASP as Desktop Ryzen Chips Got Pricier | Hardware Times

Like its rivals, AMD had a bittersweet third quarter, falling short of estimates but still recording year-on-year growth of 29%. Team Red’s report card was notably better than Team Blue’s. While the latter’s data center business had a disastrous Q3, AMD’s revenue gain was mainly the result of a...

www.hardwaretimes.com

 

[CountMike]

I have just completed my build with the following components:

• Ryzen 9 5900X
• ASUS ROG Strix B550i Gaming Mini ITX motherboard
• Be Quiet Dark Rock Pro 4 Cooler
• EVGA RTX XC3 3070
• 32GB Ballistix 3600Mhz ram
• 2 Be Quiet 140mm fans
• Seasonic Focus GX850 PSU
• Thermaltake Core V21 Case
• Sabrent Rocket 4 and Sabrent Rocket Q NVME drives mounted on top and bottom of board
• WD 1TB SSD on the SATA port

My case has a 200mm front fan (~150CFM) and my configuration is basically set up this way (if you are looking at the front of the case):

1. Air blows in from the 200mm fan
2. Blows towards the back of the case
3. The motherboard is in the back right of the case
4. The DRP4 cooler has the 2 standard fans on it, cooling the CPU and oriented so that they blow towards the back of the case
5. One 140mm fan is mounted on the back of the case, right behind the cooler, continuing the flow of air out the back of the case
6. One 140mm fan is mounted on the right side near the back of the case, right next to the DRP4 cooler, blowing air out of the side of the case (my thinking here is that this area of the case is where the most heat will be concentrated)
7. So I have the 1 intake at the front, running at full speed, and the 2 140mm fans for exhaust on PWM
8. The PSU fan is facing down so it is not adding or removing any air from the case
9. The GPU's 3 fans didn't even kick on during the load test

Last night I put it through its paces.

• With 100% load, the CPU Temp was reported at 73c and the CPU Package peaked at 89c but was typically at 85c for the duration.
• The CPU boosted to 4500Mhz during the load on a couple of cores.
• I noticed that the VRM heatsink fan speed was sitting at 0 rpm, which made me wonder if there was something wrong with the VRM fan. MB temp was <50c and the chipset temp was in the low 60s.
• This was a 20 minute load test (upconverting a batch of 80 videos using MOVAVI.

I am trying to understand if all this is typical. From what I am reading some say it is typical under load, others say it's too high, some are saying it's fine... I'm not sure what I was expecting but just want to find out if this hardware is good to go. Online, I can't find a consensus on what a typical temp range is (conflicting opinions). Regarding the VRM fan, there is nothing in the specs about when it would kick in. I just assumed it would always run. Is there something in the bios I am overlooking perhaps?

I used the pea method for thermal paste when I mounted the cooler. It made contact with the CPU; I know this because although it was hard to see, I could see and feel it slide a tiny bit on the CPU and I had not added the cross-bar mounting bracket for a final screwdown yet (by the way, the installation on this thing was WAY too frigging hard on such a small board. Partially my fault for choosing an ITX build, but also read that many prefer Noctua ease of mounting).

So I thought to myself, is the cooler's middle (inside) fan spinning, or is just sitting there? Did I mount that fan incorrectly? And is there an easy way for me to check that fan is working without taking the cooler apart? In the end, with this freaking behemoth of a cooler, is 85c the best I can get under load?

Any help or advice would be appreciated.

Those are pretty normal temps for that CPU and that cooler, Zen3 are allowed much higher temps than earlier Ryzen.
Only with a 360 liquid cooler you can expect somewhat lower temps.

 

[cogsman]

Thanks for this! I've been hearing the same. In fact, I've been doing some stress tests with various benchmarks, and everything equalizes between 87 and 90 under full load, with it mostly landing at 87. Maybe I was just a bit paranoid as my last AMD was an FX processor from 8 years ago, and my last processor was an i7-6700 so I'm not used to these temps lol.

 

[cogsman]

Thank you for these responses. I have been stress testing and everything equalizes between 87 and 90, leaning mostly towards the 87 side. I am not used to these high temps.

 

[CountMike]

Thanks for this! I've been hearing the same. In fact, I've been doing some stress tests with various benchmarks, and everything equalizes between 87 and 90 under full load, with it mostly landing at 87. Maybe I was just a bit paranoid as my last AMD was an FX processor from 8 years ago, and my last processor was an i7-6700 so I'm not used to these temps lol.

Lol, get used to those temps, most high core count CPUs are much hotter than older ones.

 

[drea.drechsler]

Thank you for these responses. I have been stress testing and everything equalizes between 87 and 90, leaning mostly towards the 87 side. I am not used to these high temps.

A few items that may be of interest if you're new to the AM4 platform...

Ryzen has dozens of temp sensors scattered over the processor's cores. It also uses an extremely aggressive boosting algorithm that will frequently boost cores to maximum clocks even when it's at 'idle'. You'll see a temp spike from one of the sensors in the boosting core and that spike will make fans pulse. Setting up a custom profile that ignores the spikes and follows the average temperature will avoid pulsing. The spikes are kind of like a lighted match in a room...the match is hot but the room isn't.

The best monitoring utility for AM4 is HWInfo64. It has temp readouts for both the instantaneous core temp and an average core temp; the average core temp is closest to the true thermal state. It also has a reliable VCore reading... the Core Voltage (STI2 TFN). That's the true core voltage as the processor reads it out in telemetry.

And a comment about 'stress testing'. Prime95 (small FFT's in particular) and as well Aida 64's, are simply unreal in the extreme with such tightly nested, perfectly optimized AVX rich code executing endlessly on all available cores simultaneously, never even the slightest pause to reconstruct data tables. So just being stable running it is quite good enough in my opinion; I run it for 20 to 30 min's max and not worry (much) about temperature. I'd worry about temperature when running a real-world stress test which would be something like Cinebench 23 for two hours to simulate a very long batch rendering. Or a 'sample' batch rendering of your own devising using your favorite rendering tool (I'd use Handbrake X.264).

There's usually setting in BIOS for something called 'platform thermal limit'. In my BIOS I have to enable PBO in manual mode to expose it. Set that to 89 or 90C (Tjmax for 5900X's)... the highest temp recorded during the CB23 stress test would be a good limit actually... and it will never go above that. It really works a trick on my system (although it's a 3700X it should work the same) as it doesn't 'hard throttle' the CPU, it just lowers the boost clocks and voltage to keep it under or really close to the set point.

 

[BogdanH]

I noticed that the VRM heatsink fan speed was sitting at 0 rpm, which made me wonder if there was something wrong with the VRM fan.

Your motherboard has fan for VRM?

 

[Paperdoc]

I see on your mobo's we page that it DOES have a VRM fan included. And that has its own dedicated VRM_HS_FAN header at the board top edge, to the rear from the three CPU_FAN etc headers. So to start with, just check that the connector IS plugged into that header securely. Details of all the fan headers on this mobo are not in the main User Manual. But there is a separate manual for the BIOS of this board. In that on p. 56 I see you can either Enable or Disable this header, so check there how that is set. There is no indication that this header can display the actual speed pf this fan.

 

[cogsman]

VRM fan was connected and I found that under load, it did spin up after the temp increased. I have heard now from this thread and others that 5000s just run hot, and these temps are not unexpected. As of a week later, this is running fine! The rig screams

Thanks to all for your help!

 

[cogsman]

This has all been really helpful! After a week, my load testing maxes at 90c but usually closer to 87c. VRM fan does spin up after a certain temp. All seems to be running fine. I appreciate all the help. The system screams!

 

[Paperdoc]

Glad you got it resolved, and have an impressive system.