New Build

[mama]

I'm open to suggestions for improvement. The below results are after a a couple hours of solid gaming.

 

[Zach_01]

Please fill your profile with detailed system specs for easy access.

We cant really say much of those screenshots. HWmonitor is kind of a toy and you need a tool.

Announcements

Important announcements, new releases

www.hwinfo.com

Also this for RAM settings and memory subsystem in general

ZenTimings

ZenTimings is a simple and lightweight app for monitoring memory timings on Ryzen platform.

zentimings.protonrom.com

 

[mama]

Please fill your profile with detailed system specs for easy access.

We cant really say much of those screenshots. HWmonitor is kind of a toy and you need a tool.

View attachment 183590

Announcements

Important announcements, new releases

www.hwinfo.com

Also this for RAM settings and memory subsystem in general

ZenTimings

ZenTimings is a simple and lightweight app for monitoring memory timings on Ryzen platform.

zentimings.protonrom.com

Thanks for that. I've looked at those tools you recommended and attach the following screenshots for information. I am a little concerned with GPU temps. The results are after gaming in Anno 1800 at 1440p highest settings for a couple of hours. Any advice for tweaking or further set up

appreciated.

 

[Zach_01]

There is nothing to worry about the GPU temps. Its noramal to get to those max temps when the GPU is utilized at 99~100%.
The max operating temperatures are:

GPU: 100C
GPU hotspot: 110C
Memory: 100C

GPU VRM: 115C
Mem VRM: 115C

What is shown in your sreenshots is the max reached temps. Average values (last column) are important too. As I said these are normal max temps.
Here is one of mine.
Further down from temps you can see the power draw of GPU and memory, and also the thermal limits of components in %.

 

[mama]

Thanks. I accept that the temps on the card are within acceptable parameters and I am quibbling but I saw this: AMD Radeon RX 6800 XT Review - NVIDIA is in Trouble | TechPowerUp and wondered why a custom card would arguably have less efficacious results. Sure the clocks are higher but usually thermals are better than the reference too.

This is my first build. I researched the parts and how I wanted it to go together but I am very much learning the craft and want to improve. I appreciate any advice on my setup generally.

 

[Zach_01]

I’m sure this looks odd to you, but first as you said your card has higher clocks and second you can’t compare your card with another one that has been benchmarked elsewhere and probably with different conditions, as different ambient temp with different case or no case at all.
Your Corsair case is with a glass front or mesh?

 

[mama]

Glass front and sides (and top). Large gaps on the corners (and filters) for airflow.

 

[Zach_01]

Well no matter how large are the corner gaps, it can’t compare as if it had a mesh front. Airflow would increase vastly. Anyway your temps are fine so it’s not a must, at least GPU temp wise. How is you CPU temp under heavy load? Did any test? No need to run some torture test, just normal stuff. Like Cinebench R20 in a loop run for a few minutes. If CPU passes 80C maybe you should consider improving case airflow. Max operating temp is 95C for CPU, but it shouldn’t go near that.
Run some games too for a couple of hours each, with HWiNFO open(minimized) so you check later.

Some suggestions
You have 6 main columns on HWiNFO which is dense, so you can’t see the forth value of the Avg. It has its meaning. So maybe it’s better to make them 5. The 5th will be longer and will need scrolling but it’s ok. You can always have it shown what you want. SSD/HDD info is not all that important.

Min and Max values may “exist” only for very short periods of time. Average is important too.
In order to make sense though HWiNFO has to “record” values only during the task you want. So when you want to see game or any kind of benchmark results, you hit the clock down right and start the app/game as soon as possible. After finishing, as fast as you can you maximize HWiNFO and take a screenshot for later study. You can paste it (Ctrl+V) on windows painting app to save the image.

You can do this procedure for:
1. Games (at least 2 hours)
2. Benchmarks (single or multi thread for as long as they last)
3. Idle or mix of light loads like browsing, watching videos (at least 2 hours)

Post them here if you like

Have updated your
Windows
Board BIOS
Chipset drivers (always straight from AMD drivers download page)

You may find a lot of the info and values on HWiNFO sensors window unknown or irrelevant but believe me it will help you understand you system better, and faster. And in order to improve it, if it has any headroom, you should first get to know and learn it.
I and others I’m sure, can help you achieve that.

 

[mama]

Well no matter how large are the corner gaps, it can’t compare as if it had a mesh front. Airflow would increase vastly. Anyway your temps are fine so it’s not a must, at least GPU temp wise. How is you CPU temp under heavy load? Did any test? No need to run some torture test, just normal stuff. Like Cinebench R20 in a loop run for a few minutes. If CPU passes 80C maybe you should consider improving case airflow. Max operating temp is 95C for CPU, but it shouldn’t go near that.
Run some games too for a couple of hours each, with HWiNFO open(minimized) so you check later.

Some suggestions
You have 6 main columns on HWiNFO which is dense, so you can’t see the forth value of the Avg. It has its meaning. So maybe it’s better to make them 5. The 5th will be longer and will need scrolling but it’s ok. You can always have it shown what you want. SSD/HDD info is not all that important.

Min and Max values may “exist” only for very short periods of time. Average is important too.
In order to make sense though HWiNFO has to “record” values only during the task you want. So when you want to see game or any kind of benchmark results, you hit the clock down right and start the app/game as soon as possible. After finishing, as fast as you can you maximize HWiNFO and take a screenshot for later study. You can paste it (Ctrl+V) on windows painting app to save the image.

You can do this procedure for:
1. Games (at least 2 hours)
2. Benchmarks (single or multi thread for as long as they last)
3. Idle or mix of light loads like browsing, watching videos (at least 2 hours)

Post them here if you like

Have updated your
Windows
Board BIOS
Chipset drivers (always straight from AMD drivers download page)

You may find a lot of the info and values on HWiNFO sensors window unknown or irrelevant but believe me it will help you understand you system better, and faster. And in order to improve it, if it has any headroom, you should first get to know and learn it.
I and others I’m sure, can help you achieve that.

Thanks.

I think airflow in the case is pretty good. I have 6 case fans in addition to the CPU fan. 4x180mm and 2x240mm. One 240mm fan and one 180mm fan are located at the top/rear as exhausts.

I did a 10 minutes R23 multi test. The CPU temps looks fine to me but maybe you have more insight. I did the HWiNFO test as you suggested (see attached). R23 score was 20797 which I think is okay. I also did a CPU-Z bench for single thread (see attached).

Windows is up to date. My Chipset driver is up to date but my BIOS is the most current except for a beta BIOS which has just been released which includes facility for Curve Optimisation in Precision Boost 2. It's in beta so I'm not sure I want to go there yet and I understand Ryzenmaster will be updated soon to incorporate the feature rather than it being a strictly BIOS implementation. I might wait for that.

Anyway, whatever you can suggest I'm all ears.

 

[mama]

Thanks.

I think airflow in the case is pretty good. I have 6 case fans in addition to the CPU fan. 4x180mm and 2x240mm. One 240mm fan and one 180mm fan are located at the top/rear as exhausts.

I did a 10 minutes R23 multi test. The CPU temps looks fine to me but maybe you have more insight. I did the HWiNFO test as you suggested (see attached). R23 score was 20797 which I think is okay. I also did a CPU-Z bench for single thread (see attached).

Windows is up to date. My Chipset driver is up to date but my BIOS is the most current except for a beta BIOS which has just been released which includes facility for Curve Optimisation in Precision Boost 2. It's in beta so I'm not sure I want to go there yet and I understand Ryzenmaster will be updated soon to incorporate the feature rather than it being a strictly BIOS implementation. I might wait for that.

Anyway, whatever you can suggest I'm all ears.

Correction: 4x120mm fans and 2x140mm fans.

I wouldn't mind some help in interpreting this information. This is taken following a couple of hours of gaming at 1440p.

 

[Zach_01]

There are some discrepancies between different sensors for same things. Like CPU power draw, Current(A).
“CPU PPT” reports 109W max power but “CPU Core Power” report 152W.
Same for “CPU EDC” and “CPU Core Current”.
And the thing is that CPU PPT/EDC are for the entire CPU package and the other I’m referring to, are cores alone.
I’m not saying that CPU is dis function in some way. Not at all. It’s just board misreadings.

If you see VRM output (cores +SoC) for current and power draw it’s closer to CPU Core Power/Current than PPT/EDC. So it could be just VRM false readings that cause these miscalculations.

I would believe more PPT/EDC in conjunction with PowerReportingDeviation, as these are straight from CPU internally.

You can do this test
Run CinebenchR20 milticore and take a screenshot of HWiNFO in the middle of the run, before it ends. All test should last around 30, tops 35sec for the 5900X.
No need for the R20 to show on screenshot. Just as your previous it’s fine, as long as the test still runs when you take it.

Apart from this, everything else looks fine.

 

[mama]

There are some discrepancies between different sensors for same things. Like CPU power draw, Current(A).
“CPU PPT” reports 109W max power but “CPU Core Power” report 152W.
Same for “CPU EDC” and “CPU Core Current”.
And the thing is that CPU PPT/EDC are for the entire CPU package and the other I’m referring to, are cores alone.
I’m not saying that CPU is dis function in some way. Not at all. It’s just board misreadings.

If you see VRM output (cores +SoC) for current and power draw it’s closer to CPU Core Power/Current than PPT/EDC. So it could be just VRM false readings that cause these miscalculations.

I would believe more PPT/EDC in conjunction with PowerReportingDeviation, as these are straight from CPU internally.

You can do this test
Run CinebenchR20 milticore and take a screenshot of HWiNFO in the middle of the run, before it ends. All test should last around 30, tops 35sec for the 5900X.
No need for the R20 to show on screenshot. Just as your previous it’s fine, as long as the test still runs when you take it.

Apart from this, everything else looks fine.

Ok, Cinebench test result attached.

I understand the PPT for the chip is limited to 142W so this result is within that range. It was also in the range in the previous reading. I'm not really sure about the rest.

 

[mama]

Looks okay to me.

 

[Zach_01]

Ok, Cinebench test result attached.

I understand the PPT for the chip is limited to 142W so this result is within that range. It was also in the range in the previous reading. I'm not really sure about the rest.

On this screenshot all power readings are making sense (compared to one another) and are normal for your CPU.

Looks okay to me.

This is normal too

What is odd (a little) to me is the CPU temp. When CB benched the temp was lower even though CPU PPT (watt) is higher, compared to gaming load.
It could mean that GPU heat (~300W) is affecting CPU temp when you game by increasing your case's ambient. I cant think of anything else that does makes sense.

 

[mama]

That might be so but either way I don't think the temps are high. Might try the new BIOS and undervolt in a month or to but other than that I think the thing is okay for a while. Thanks for you observations and help.

Yes, looking at the HWiNFO screenshots again, the CPU is hotter when gaming than on Cinebench when GPU use is low. Maybe I need more airflow over the top of the card... I have SAM enabled: maybe it is something to do with that. Or maybe it's the particular game which is Anno 1800. I'll try something else and see.

Same game played for a couple of hours but bumped up the case and CPU fans. Seems to have made a difference.

 

[Zach_01]

As I thought, its the GPU heat increasing case temp. Its normal. I dont think SAM has anything to do with it directly. Yes, it may increase CPU load in theory, but I would say it is increasing CPU efficiency by using GPU VRAM for the CPU.
Anyway I think the system is behaving very normal on all aspects, and all values are nominal.
There is always room for tuning the DRAM, CPU etc but since the system is new and nothing wrong with it, I suggest keep it as is and enjoy it for now.

 

[mama]

Thanks again.