[SOLVED] System Performing Below Expectations

[maisoncf91]

Hello,

For my first post ever I want to ask you fine folks about some optimization.

I have been tweaking settings in bios as well as within Windows itself and HAVE seen a performance increase.

History: About 1 month ago I disassembled my PC. I reapplied thermal paste to the CPU and GPU. I was having heating issues and found some deeply embedded dust in various places. The whole thing is dust free now. At this point I did a fresh install of Windows 10. COMPLETELY fresh. Uninstalled the system and all programs entirely before reinstall. About 2 weeks later I got 2 SSD's and migrated my HDD and the OS to am SSD. Boot times definitely improved. That brings me to now.

I have the following system...

MOBO - MSI 970a-G46
CPU - AMD FX 8350 Black Edition
GPU - ASUS RX 480 4GB
RAM - 12GB DDR3 XMP 1333mhz
Storage - 240GB PNY SSD x2, 500GB HDD x1
PSU - 750 W Corsair
Cooling - Stock CPU cooler, 4 additional fans

Peripherals

Corsair Void Pro Headset
Razer Mamba Chroma RGB Wireless Mouse
Razer Blackwidow Chroma Keyboard
Razer Chroma Firefly
24in Sceptre 1080p 60hz monitor


CPU - OC 4300mhz 1.4v 200x21.5
Socket temp under load 50c
Chip temp under load 45-52c

GPU - Auto OC in Radeon
Temps under load 55-60c
RAM - XMP 1333mhz

I get my CPU up to 4.4 stable easily with no overheating but dropped it to 4.3 to see if it was impacting performance being higher.

I have been running userbenchmark over and over and continuously get "Your PC is running below expectations" by a significant margin. Everything is green except GPU is yellow which is to be expected and RAM is yellow.

It says my SSD's are performing WAY below expectations. I verified trim, reformatted them, turned indexing off, turned off sysmain. Still the same. I don't get crashes or anything when stability testing. I just don't know how to get my system to perform where it should be.

I have high background CPU usage from time to time (25%) and don't know where it's coming from, but it averages around 9%.

I can attach the UBM when I get home later, but yea. Any ideas in the time being to discuss that I may or may not have tried would be great.

Thank you!

 

[USAFRet]

First off, UBM is a flawed tool. It gives no context.

Frequently, you might see "way below expectations" and "168% Outstanding", for the same drive at the same time.

Use an actual benchmark tool for the drives.
CrystalDiskMark or AS SSD.

 

[maisoncf91]

That's kind of frustrating to hear.

Kinda silly that so many "professional" diagnostics places/services request you to run it of it's broke.

I'll download CrystalDiskInfo when I get home tonight. When I run it, will it "rate" the performance for me? Or do I need to know what I'm looking at?

Thanks!

 

[USAFRet]

That's kind of frustrating to hear.

Kinda silly that so many "professional" diagnostics places/services request you to run it of it's broke.

I'll download CrystalDiskInfo when I get home tonight. When I run it, will it "rate" the performance for me? Or do I need to know what I'm looking at?

Thanks!

CrystalDiskMark, for performance.
CrystalDiskInfo, for information on the drive.

Like this:
CrystalDiskMark

AS SSD

UBM is sort of good to compare your same system to "Last month" and "Next Month"
Other than that, it gives no context. "way below"....what, exactly?

 

[maisoncf91]

CrystalDiskMark, for performance.
CrystalDiskInfo, for information on the drive.

Like this:
CrystalDiskMark

AS SSD

UBM is sort of good to compare your same system to "Last month" and "Next Month"
Other than that, it gives no context. "way below"....what, exactly?

Thanks

CrystalDiskMark, for performance.
CrystalDiskInfo, for information on the drive.

Like this:
CrystalDiskMark

AS SSD

UBM is sort of good to compare your same system to "Last month" and "Next Month"
Other than that, it gives no context. "way below"....what, exactly?

Thank you so much! I'll post the results here later for review.

I appreciate it!

 

[geofelt]

Are you happy now with your pc?
FX-8350 was a fine processor when it was launched some 8 years ago at $195.
Today, a i3-10100 at $110 will perform twice as well.

FX is a hot processor, and the motherboards need very good vrm cooling or else they will throttle.
If you have any doubts about cpu performance, look into heat issues.

The only real good benchmark is YOUR workload.

 

[cryoburner]

First off, UBM is a flawed tool. It gives no context.

Frequently, you might see "way below expectations" and "168% Outstanding", for the same drive at the same time.

Other than that, it gives no context. "way below"....what, exactly?

UserBenchmark might not be perfect, and does some questionable things with how some of their comparisons are performed and presented, but it makes perfect sense for a component to be described as both "way below expectations" and "outstanding".

"Way below expectations" is describing how that particular component compares to its "expected" performance based on tests of the exact same component running in other people's systems, whereas "outstanding" is describing how it compares to that category of components in general. So in that example, a fast SSD could be performing well below where it should be for some reason, but still well above what is typical for storage devices in general. So if UserBenchmark says that, it's a clear indication of something negatively affecting the component's performance.

They provide more details too, that could perhaps better indicate what exactly is going on with the drives, but we would need a link to the actual results page to gather much from that.

 

[USAFRet]

UserBenchmark might not be perfect, and does some questionable things with how some of their comparisons are performed and presented, but it makes perfect sense for a component to be described as both "way below expectations" and "outstanding".

"Way below expectations" is describing how that particular component compares to its "expected" performance based on tests of the exact same component running in other people's systems, whereas "outstanding" is describing how it compares to that category of components in general. So in that example, a fast SSD could be performing well below where it should be for some reason, but still well above what is typical for storage devices in general. So if UserBenchmark says that, it's a clear indication of something negatively affecting the component's performance.

They provide more details too, that could perhaps better indicate what exactly is going on with the drives, but we would need a link to the actual results page to gather much from that.

And then when you test with a drive benchmarking tool, it reads as giving the exact performance it should be.

The problem is, there is no context of what "way below" actually means.

This is one of my SATA III SSD's, 1TB 860 EVO

"way below"?
'Outstanding"?
Or, the numbers from CrystalDiskMark, which are pretty much exactly what a SATA III SSD should be?

People glom on to that red text of "way below expectations", and assume a major problem with their system. Something that needs to be "fixed".

 

[maisoncf91]

Are you happy now with your pc?
FX-8350 was a fine processor when it was launched some 8 years ago at $195.
Today, a i3-10100 at $110 will perform twice as well.

FX is a hot processor, and the motherboards need very good vrm cooling or else they will throttle.
If you have any doubts about cpu performance, look into heat issues.

The only real good benchmark is YOUR workload.

 

[maisoncf91]

Okay, so here are the UBM scores. I fixed the SSD issue. I didn't have ACHI on like I thought.... -_- now they are working more like I expected. Anything else that can be seen?

https://www.userbenchmark.com/UserRun/34110894

Added a memtest86 log as well.

memtest86

Thanks!

 

[cryoburner]

Okay, so here are the UBM scores. I fixed the SSD issue. I didn't have ACHI on like I thought.... -_- now they are working more like I expected. Anything else that can be seen?

The hardware appears to be performing more or less as should be normal for that hardware. I don't notice anything glaringly wrong, at least. They did note high background CPU utilization while the test was running, though that could have potentially been from your web browser or something.

The problem is, there is no context of what "way below" actually means.

The context would be the average of everyone else with an 860 Evo 1TB running the same test on their systems. Looking at the limited portion of results you provided, those sustained write numbers are a couple-hundred megabytes lower than what the drive appears to commonly achieve on other systems running that particular test. Maybe its just down to the drive not being fully trimmed, or background tasks affecting performance or something, but for whatever reason, 90% of those who ran the benchmark on that model of drive saw better performance, hence why they flagged it as performing "below expectations" in your case. It's possible there could be some issue there, though it's also possible that the issue might not actually have much of an impact on real-world performance.

Meanwhile, the "outstanding" refers to the performance relative to what they consider a typical SSD, which is currently a Samsung 850 Pro. That's a lot more arbitrary, but perhaps can give people a general perception of how their components perform against "typical" hardware in use. It would probably be better if they downplayed that part of the results though. I do think they could definitely improve the way the results are presented.

As for CrystalDiskMark, it's not always going to be entirely clear how the results compare to others, as there's no averages available to compare the numbers against. You kind of have to track down results in reviews and posted by others online, and assume they might be representative of typical performance.

That's one thing UserBenchmark does well, is to provide a good reference against how a given piece of hardware performs in other systems, including a graph clearly indicating where a particular result lies on the performance curve for a given model, along with a good overview of the system as a whole. While UserBenchmark might have some notable issues, like weighting the results in some questionable ways when ranking certain components against others, I don't know of any other benchmark suite that offers anything comparable.

 

[maisoncf91]

The hardware appears to be performing more or less as should be normal for that hardware. I don't notice anything glaringly wrong, at least. They did note high background CPU utilization while the test was running, though that could have potentially been from your web browser or something.


The context would be the average of everyone else with an 860 Evo 1TB running the same test on their systems. Looking at the limited portion of results you provided, those sustained write numbers are a couple-hundred megabytes lower than what the drive appears to commonly achieve on other systems running that particular test. Maybe its just down to the drive not being fully trimmed, or background tasks affecting performance or something, but for whatever reason, 90% of those who ran the benchmark on that model of drive saw better performance, hence why they flagged it as performing "below expectations" in your case. It's possible there could be some issue there, though it's also possible that the issue might not actually have much of an impact on real-world performance.

Meanwhile, the "outstanding" refers to the performance relative to what they consider a typical SSD, which is currently a Samsung 850 Pro. That's a lot more arbitrary, but perhaps can give people a general perception of how their components perform against "typical" hardware in use. It would probably be better if they downplayed that part of the results though. I do think they could definitely improve the way the results are presented.

As for CrystalDiskMark, it's not always going to be entirely clear how the results compare to others, as there's no averages available to compare the numbers against. You kind of have to track down results in reviews and posted by others online, and assume they might be representative of typical performance.

That's one thing UserBenchmark does well, is to provide a good reference against how a given piece of hardware performs in other systems, including a graph clearly indicating where a particular result lies on the performance curve for a given model, along with a good overview of the system as a whole. While UserBenchmark might have some notable issues, like weighting the results in some questionable ways when ranking certain components against others, I don't know of any other benchmark suite that offers anything comparable.

Thank you for that breakdown I appreciate it! I was actually able to get the SSD score way up to above expectations by turning on ACHI in Bios. I thought I already had but probably didn't save.

Would you happen to have any input on my memtest86 results? I have no idea how to read that...

Thanks!

 

[USAFRet]

Right.
I guess my main issue is that big red "way below expectation". They don't state what it is way below.

Far too much 'in your face', and people think there is a major issue that needs to be fixed.