[SOLVED] Low score on time spy with ryzen 7 5800x

hoyes127

Commendable
May 3, 2020
5
1
1,515
0
I always get around 9500 score on my CPU and when comparing to the average it should should be way up in the 11000's. I saw in similar threads it might be temperature but i monitored it during a benchmark and it was steady at low 60's. Any ideas`?

a previous result: http://www.3dmark.com/spy/28721290

I later ran cinebench r23 multi core tests and they were around 12500 when a 5800x should be around 15000...
 
Last edited:

hoyes127

Commendable
May 3, 2020
5
1
1,515
0
full system spec? include brand and model of the psu
new build? latest motherboard bios? fresh os install?
My specs:
MSI MPG B550 GAMING PLUS
AMD Ryzen 7 5800x 8-core processor 3.80 GHz
DUAL-RTX2060-O6G
Kingston FURY Beast RGB DDR4 3200MHz 2x16GB
Corsair RM750x - 750 watt- 80 Plus Gold

My BIOS version is American Megatrends International, LLC. 1.50, 01.14.2021
I last had a fresh os install on 05.01.2022
 
I don't think an outdated BIOS is a cause for concern, especially when you consider reviews of the 5800X are getting what OP expects to get. The 3DMark report is reporting the RAM is running at basically the advertised speed.

The 3DMark reporting the average CPU speed was "3,864 MHz" is kind of concerning. Can you post a screenshot (either to imgur or imgbb) of the results screen in 3DMark with the graph that shows various stats?
 
I don't think BIOS being outdated would affect performance all that much either. And if it did, it might even be for the better since I'm aware several people like a couple of the older AGESA's for their 5000 overclocking (1.2.0.6 seems to be a favorite). But it might be worthwhile anyway.

But one thing to definitely look into as a high priority is cooling issues. If I'm understanding what's posted it looks like there are two GTX 2060's in the case. One alone can be a furnace for heat output...two I can only imagine. Pay very careful attention to case ventilation if using an air cooler...or better yet seriously consider getting a 240mm AIO if you can front mount it. Ryzen CPU's are very thermal sensitive: cool them well enough and that alone is pretty much like overclocking.

Then check out some youtubes on tweaking the curve optimizer settings for your 5800X and let it breathe with some undervolted cores. You don't have to go all in to see immediate benefits.
 
Last edited:

Karadjgne

Titan
Ambassador
Is the cpu undervolted? Vcore undervolt will limit boosts, temps will limit boosts, auto voltage will limit boosts. Static OC will trash single thread ability, especially coupled with minimal voltage to try lowering temps.

Basically, if you treat a Ryzen like its an intel, scores will drop, especially in TimeSpy.
 

Karadjgne

Titan
Ambassador
I get higher scores in CBr20 in both single and multi with a set 1.325v vid than I do with voltage on Auto, which puts my Ryzen upto 1.475v single and 1.425 multi.

Temps with a static 4.4GHz locked core run 82°C, temps with dynamic cores which hit 4.29GHz and vid at 1.325 hit 62°C. Scores run 5000 multi at 4.4GHz and 5100 at 4.29GHz.

It's a balancing act, applies to any Ryzen of any gen, finding the right combination of voltages and temps to maximize performance, regardless of cpu frequency, which doesn't always mean faster is better. They are Ryzens, not Intels, cpu frequency is only one part of the equation, not the overriding drive force like in Intels.

With Ryzens IPC literally drops as voltages and temps rise, a Ryzen at 60-70°C literally 'thinks' faster than the same Ryzen in the same load at the same speeds and 70-80°C

Look at those 2 pictures. The cpu on the left with lower VID has all cores current above 4.59GHz. The cpu on the right with higher VID has 3 cores barely over 3.7GHz, killing multi threads, even if the single thread speed is slightly higher. That's a performance loss, not gain.
 
Last edited:
To put it simply, boost is not working, suggest resetting CMOS, enable XMP and use Core optimizer to see what BIOS comes up with.
Just in case, check your Power options in windows > Maximum processor state if it's less than 100%.
 
I get higher scores in CBr20 in both single and multi with a set 1.325v vid than I do with voltage on Auto, which puts my Ryzen upto 1.475v single and 1.425 multi.

Temps with a static 4.4GHz locked core run 82°C, temps with dynamic cores which hit 4.29GHz and vid at 1.325 hit 62°C. Scores run 5000 multi at 4.4GHz and 5100 at 4.29GHz.

It's a balancing act, applies to any Ryzen of any gen, finding the right combination of voltages and temps to maximize performance, regardless of cpu frequency, which doesn't always mean faster is better. They are Ryzens, not Intels, cpu frequency is only one part of the equation, not the overriding drive force like in Intels.

With Ryzens IPC literally drops as voltages and temps rise, a Ryzen at 60-70°C literally 'thinks' faster than the same Ryzen in the same load at the same speeds and 70-80°C
If you're modifying Zen 3 setting the VID and whatnot directly to undervolt, then I'd argue that's not what you're supposed to do. This is what PBO Curve Optimizer is for.

https://hothardware.com/reviews/amd-power-curve-optimizer-guide-zen-3

Also I don't see how IPC drops with higher voltages and temperature. IPC is literally the result of how the instruction pipeline was designed. If it appears to drop because of voltage or temperature (or just temperature, since power dissipation is a function of voltage), it's because the frequency is fluctuating. Because keep in mind: the CPU frequency can change however often the CPU wants. Sometimes it's in the millisecond range, sometimes it's in the minutes range. That's why HWiNFO has an "effective clock speed" field now.

Look at those 2 pictures. The cpu on the left with lower VID has all cores current above 4.59GHz. The cpu on the right with higher VID has 3 cores barely over 3.7GHz, killing multi threads, even if the single thread speed is slightly higher. That's a performance loss, not gain.
This is because the one on the left is running an all-core stress test. The one on the right is running a single core stress test. I mean I guess that would've helped put things in context.
 
Last edited:

hoyes127

Commendable
May 3, 2020
5
1
1,515
0
I always get around 9500 score on my CPU and when comparing to the average it should should be way up in the 11000's. I saw in similar threads it might be temperature but i monitored it during a benchmark and it was steady at low 60's. Any ideas`?

a previous result: http://www.3dmark.com/spy/28721290

I later ran cinebench r23 multi core tests and they were around 12500 when a 5800x should be around 15000...
After a few hours of looking through bios and settings i found the very simple answer of increasing the maximum processor state to 100% in the power options....
View: https://imgur.com/a/9EVTSe1
i set it at 99 like a year back and forgot about it
 
Reactions: hotaru.hino
After a few hours of looking through bios and settings i found the very simple answer of increasing the maximum processor state to 100% in the power options....
View: https://imgur.com/a/9EVTSe1
i set it at 99 like a year back and forgot about it
If this was to disable turbo boost, I've found this setting is kind of wonky. There's a hidden option you can enable that lets you adjust turbo boost behavior if you're interested.
 

ASK THE COMMUNITY