[SOLVED] Lower results than World's average in Firestrike/TimeSpy

[insolito]

Hello, I'm trying to benchmark my new hardware:

CPU: AMD Ryzen 9 5900x
MB: ROG Strix x470-F Gaming
GPU: ROG Strix 6800XT LC OC
RAM: 2x8GB Corsair 3000Mhz @3200Mhz

But I'm getting below average results, like in Fire Strike 32671 points. World's Average is 37484 points.

There is my results:
TimeSpy: https://www.3dmark.com/3dm/57101353
FireStrike: https://www.3dmark.com/3dm/57104355

What could be the cause for these low score? Thanks in advance!

 

[helpstar]

most systems are much more overclocked on such benchmark sites

other than that your motherboard might hold back the full potential of the 5900X

 

[rgd1101]

make and model of the psu?
cpu/gpu temp?
fresh os install?

 

[insolito]

most systems are much more overclocked on such benchmark sites

other than that your motherboard might hold back the full potential of the 5900X

But these results are lower than average for much points

 

[insolito]

make and model of the psu?
cpu/gpu temp?
fresh os install?

Gigabyte 750W 80 Plus Gold Modular - G750H

CPU Temp 68~70 C / GPU Temp 50~55 C

Didnt make a fresh OS install, but Windows 10 get it to work immediately, also i had reinstall chipset and gpu drivers

Also i using the lastest bios available updated with AMD AM4 AGESA V2 PI 1.2.0.0

Can the resolution or the refresh rate influence these results? My display is set to 2560x1080 @160hz

 

[rgd1101]

do a fresh os install and test again

 

[DSzymborski]

"Average" is average for the hardware, not average for the hardware at a specific overclock. In a population that is going to have a lot of overclockers, the average stock configuration is going to be below-average.

And yes, install Windows properly. I guarantee you most of the people serious about overclocking and getting better scores aren't simply slapping their old OS install onto their new hardware to cut corners. Anything worth doing is worth doing right.

 

[USAFRet]

Didnt make a fresh OS install

Problem #1

getting below average results

Problem #2, comparing your system to those that have been overclocked, tweaked and optimized for that particular test

 

[insolito]

Ok, thanks guys, i will reinstall W10 on it, and how about the resolution and refresh rates? can it impact the results?

 

[alceryes]

Run a userbenchmark and post a link to the results so we can have a look.

Also, your memory appears to be running a little slow for a premium build.

 

[Deleted member 2838871]

But I'm getting below average results, like in Fire Strike 32671 points. World's Average is 37484 points.

There is my results:
TimeSpy: https://www.3dmark.com/3dm/57101662
FireStrike: https://www.3dmark.com/3dm/57104355

What could be the cause for these low score? Thanks in advance!

Where are you getting this "World's Average" from? Your 32k Firestrike shows 99th percentile... and so does my 35k. https://www.3dmark.com/fs/24586563

BTW, both your links are for Firestrike.

"Average" is average for the hardware, not average for the hardware at a specific overclock. In a population that is going to have a lot of overclockers, the average stock configuration is going to be below-average.

And yes, install Windows properly. I guarantee you most of the people serious about overclocking and getting better scores aren't simply slapping their old OS install onto their new hardware to cut corners. Anything worth doing is worth doing right.

This too. ^

Problem #1


Problem #2, comparing your system to those that have been overclocked, tweaked and optimized for that particular test

... and this. ^

 

[insolito]

BTW, both your links are for Firestrike.

My bad, updated, TimeSpy link is https://www.3dmark.com/3dm/57101353

Where are you getting this "World's Average" from?

i am getting from the 3DMark program after the test finish, it doesnt show in the result link:

 

[Deleted member 2838871]

My bad, updated, TimeSpy link is https://www.3dmark.com/3dm/57101353

Ok... Yeah, that 16k is 20% less than my 20k https://www.3dmark.com/spy/17188357 but again you're still in the 99th percentile and it could be the GPU? I don't know jack about AMD and how that card compares to the 3090 but that would be my first guess. But then again both your CPU and GPU scores are lower than mine... so maybe it's the overclock/clock speed differences.

Also, your memory appears to be running a little slow for a premium build.

Ahhh yeah... this too.

i am getting from the 3DMark program after the test finish, it doesnt show in the result link:

Ok, I know what you're talking about now. But as others have said... a lot of it is coming from those that have serious OCs on their rigs with all kinds of tweaks and optimizations... and maybe even some liquid nitrogen.

Anyway, for me I see "99th percentile" and I'm good with it... it's really nothing to stress over IMO.

 

[insolito]

Run a userbenchmark and post a link to the results so we can have a look.

That one? https://www.userbenchmark.com/UserRun/38781316

Also, your memory appears to be running a little slow for a premium build.

I could not stable it above 3200Mhz with XMP :/

 

[Deleted member 2838871]

That one? https://www.userbenchmark.com/UserRun/38781316

I could not stable it above 3200Mhz with XMP :/

Need to run that test again... and your memory speed is a little slower than mine which would also affect those results.

 

[alceryes]

We need to get a completed test. Something's going on with your GPU.
Try this - reset all BIOS tweaks/overclocking and GPU tweaks/overclocking to their defaults and re-run userbenchmark (and the other benchmarks).

In Userbenchmark, a relative performance above 50% is good and above 70% is a finely tuned system. 'Relative' performance is NOT the benchmark score. It is the percentage listed in the component description.

See here -

https://imgur.com/0O5Ojgk

View: https://imgur.com/0O5Ojgk

 

[Zerk2012]

Need to run that test again... and your memory speed is a little slower than mine which would also affect those results.

You have completely different hardware their no way to compare the 2 PC's and say 1 is not running right.

 

[Deleted member 2838871]

You have completely different hardware their no way to compare the 2 PC's and say 1 is not running right.

I never said his PC wasn't running right... I'm just trying to point out differences why his scores are lower... which seems to be his question... and also pointed out he's STILL in the 99th percentile.

1. GPU
2. Memory speed

... and as others have said, his system probably isn't running the same tweaks and optimizations that the premium scores are... and my system isn't either.

What a world we live in... people in the 99th percentile on 3DMark and complaining about it saying there's something wrong with their PC. Personally I don't think there's anything wrong with OP's PC... I just think his hardware isn't as good as mine and the rest of the PCs that scored higher.

 

[Krotow]

OP, don't bother about those numbers. Benchmarking at today is huge pissing contest. Up to tens of grands worth equipment - machines built only to beat particular benchmark with specially prepared components, liquid nitrogen cooling etc. etc. It is not worth to involve yourself in this if you don't have spare 15-25 grands and ambitions of Kardashian (yes, pun intended). If your machine is stable with default XMP/DOCP parameters for your RAM, consider yourself lucky. I would only recommend to consider MOBO upgrade to X570 or good B550 board. And RAM upgrade to 3600/CL16 or 4000/CL18 modules in near future. Current MOBO and RAM does not work with full potential of your CPU.

 

[Deleted member 2838871]

OP, don't bother about those numbers. Benchmarking at today is huge pissing contest. Up to tens of grands worth equipment - machines built only to beat particular benchmark with specially prepared components, liquid nitrogen cooling etc. etc. It is not worth to involve yourself in this if you don't have spare 15-25 grands and ambitions of Kardashian (yes, pun intended). If your machine is stable with default XMP/DOCP parameters for your RAM, consider yourself lucky.

+1 Totally agree.

I would only recommend to consider MOBO upgrade to X570 or good B550 board. And RAM upgrade to 3600/CL16 or 4000/CL18 modules in near future. Current MOBO and RAM does not work with full potential of your CPU.

... and therein lies the problem.

 

[insolito]

Just did a fresh install and got better numbers,

TimeSpy: https://www.3dmark.com/3dm/57116392
FireStrike: https://www.3dmark.com/3dm/57117259
Userbenchmark: https://www.userbenchmark.com/UserRun/38789701

And now i am close to the Average tests (using same cpu and gpu):

Need to run that test again...

Now with a fresh install the userbenchmark test is complete: https://www.userbenchmark.com/UserRun/38789701
But i dont know why it reports "Performing below potential (54th percentile) "

 

[Deleted member 2838871]

Just did a fresh install and got better numbers,

Now with a fresh install the userbenchmark test is complete: https://www.userbenchmark.com/UserRun/38789701
But i dont know why it reports "Performing below potential (54th percentile) "

Yeah the OS fresh install is always best. Grats on the score bump. As for userbenchmark... you'll get that "performing below potential" until you OC your GPU. I got the same... and once I OCed it started showing the "performing way above expectations." https://www.userbenchmark.com/UserRun/38137610

Happy gaming!

 

[rgd1101]

Now with a fresh install the userbenchmark test is complete: https://www.userbenchmark.com/UserRun/38789701
But i dont know why it reports "Performing below potential (54th percentile) "

as said as above don't look at the percentile as it is useless. look at the 2,439 User benchmarks, average bench 166%

 

[alceryes]

Now with a fresh install the userbenchmark test is complete: https://www.userbenchmark.com/UserRun/38789701
But i dont know why it reports "Performing below potential (54th percentile) "

Don't worry about the 'below potential' tag. Userbenchmark wants you to test and test again, and hit their links for sponsors, and buy stuff from their links.

Like I previously said, anything above 50% is good. The higher scores are probably tweaked/overclocked settings and not applicable as a good comparison.