[SOLVED] Low timespy score 5900x

[Ijustdontknowanything]

So i just got a 5900x + RX 6900 XT combo and the timespy results are extremely low and i dont know why.
First test i actually got 7 850 on the cpu and i checked if XMP was enabled and it wasnt so i turned it on and ran another test and then i got 8 416. Which is also far to low for the 5900x.
I then went ahead and updated BIOS (and made sure XMP was enabled once again) and ran another test and got 8 358 on the CPU.

The GPU got 20 699 which is fine i think at stock.

What is going on with the CPU, i should be getting 12-15 000 with the 5900x...

One thing i will say is that when i look at my timespy result it says 1/12 on the "Physical / logical processors" part whereas on other peoples results with the 5900x it says 1/24. Whats that about?

Thanks!

 

[rgd1101]

fresh windows 10 install?

 

[Ijustdontknowanything]

fresh windows 10 install?

No i literally just replaced my CPU.

 

[Ijustdontknowanything]

Another thing i will say is that Ryzen Master is only showing 6 cores...

 

[rgd1101]

check msconfig to see if it limiting cpu

 

[Ijustdontknowanything]

I think i fixed it. I went into Ryzen Master and reset the whole thing, its possible that it kept some settings from my old CPU. Looks like "Legacy Compatibility Mode" was what was limiting it, basically made it into a 5600x. Just ran another test and got 13 606.

 

[drea.drechsler]

So i just got a 5900x + RX 6900 XT combo and the timespy results are extremely low and i dont know why.
First test i actually got 7 850 on the cpu and i checked if XMP was enabled and it wasnt so i turned it on and ran another test and then i got 8 416. Which is also far to low for the 5900x.
I then went ahead and updated BIOS (and made sure XMP was enabled once again) and ran another test and got 8 358 on the CPU.

The GPU got 20 699 which is fine i think at stock.

What is going on with the CPU, i should be getting 12-15 000 with the 5900x...

One thing i will say is that when i look at my timespy result it says 1/12 on the "Physical / logical processors" part whereas on other peoples results with the 5900x it says 1/24. Whats that about?

Thanks!

Timespy...or any of 3dmark's BM's...are bad for assessing CPU performance. Use Cinebench23 to test CPU performance.

Since it's synthetic and not really representative of real-world gaming it's also not all that great for assessing GPU performance. But at least it's a stick in the sand to more easily compare against others as it forces fairly common settings on the part of the renderer to get valid results.

 

[Ijustdontknowanything]

Timespy...or any of 3dmark's BM's...are bad for assessing CPU performance. Use Cinebench23 to test CPU performance.

Since it's synthetic and not really representative of real-world gaming it's also not all that great for assessing GPU performance. But at least it's a stick in the sand to more easily compare against others as it forces fairly common settings on the part of the renderer to get valid results.

I have heard that before but i think its still a good test to gauge if your system is performing as it should and help you find easy to fix issues, which is exactly what it did for me today. But yeah its probably not the most accurate, i just ran another test and got 13 739 which is 133 more than the last test and i changed absolutely nothing.

 

[drea.drechsler]

I have heard that before but i think its still a good test to gauge if your system is performing as it should and help you find easy to fix issues, which is exactly what it did for me today. But yeah its probably not the most accurate, i just ran another test and got 13 739 which is 133 more than the last test and i changed absolutely nothing.

Variances of 10% or less are considered irrelevant, so it could be up to 1333 points different and not matter.

 

[Ijustdontknowanything]

Variances of 10% or less are considered irrelevant, so it could be up to 1333 points different and not matter.

Oh dang thats a lot, i'll look into getting Cinebench for sure. Thanks!

 

[drea.drechsler]

Oh dang thats a lot, i'll look into getting Cinebench for sure. Thanks!

What are you looking for?

Remember that CPU's and GPU's are thermally constrained, so back to back runs are going to naturally have different outcomes.

Also know that the OS is still doing it's 'thing' in the background which can affect outcomes from run to run.

The point is, don't expect tight repeatability, especially if you've not established a rigorous testing protocol similar to how the (better) review web sites do.

 

[Ijustdontknowanything]

What are you looking for?

Remember that CPU's and GPU's are thermally constrained, so back to back runs are going to naturally have different outcomes.

Also know that the OS is still doing it's 'thing' in the background which can affect outcomes from run to run.

The point is, don't expect tight repeatability, especially if you've not established a rigorous testing protocol similar to how the (better) review web sites do.

I'm not really one of those testing, overclocking, undervolting people haha i just wanted to know that my system is running properly and is stable, thats about it.
The PC is more than good enough for now at stock everything so i probably wont overclock anything at any point even if i could end up with crazy numbers, top 10 on the leader boards or whatever. Doesn't interest me.

 

[drea.drechsler]

I'm not really one of those testing, overclocking, undervolting people haha i just wanted to know that my system is running properly and is stable, thats about it.
...

Yeah...that's what I was doing too when I first stepped into Ryzen. And I found out just how badly many benchmarks can be to compare performance against "similar" systems.

3DMark in particular since even though it does force you to use common settings for the renderer it doesn't for settings you can make to the hardware. So when you look at the online comparisons you end up comparing to system with massively overclocked hardware...sometimes on LN2...and have made many, many, many runs (trying to register a top score) that get logged in the database, All those scores skew the averages for making UL's comparison charts.

What I do is go looking at review sites and compare to the results they got in their reviews to see if my card is working as it should. If it's close, I'm content. But it's usually better since reviewers use very strict protocols, some set by the mfr's as a condition for getting pre-launch hardware to have reviews to publish on launch day.