[SOLVED] Why is my RX580 underpeforming?

Dec 17, 2019
14
2
15
So I recently build a new system. Everything was great except my gpu. I thought that my games were lacking some frames and after running user benchmark twice my hypotheses was confirmed with my gpu scoring far below average (3rd percentile). Trying to find out why this has been the case for me has so far been unsuccessful. Does anyone know what the problem is?

Parts:
Ryzen 5 2600
1x16GB DDR4 3000Mhz
Asus prime B450M-A
480GB M.2 SSD
450W 80+ Gold Psu
Asus Dual RX580 O4C

Trying to tweak some settings in the Radeon Adrenaline software has been unsuccessful. I also have the latest version of windows 10.
 
Solution
FPS:51.7
Score:1301
Min FPS:8.5
Max FPS:102.6
System
Platform:Windows NT 6.2 (build 9200) 64bit
CPU model:AMD Ryzen 5 2600 Six-Core Processor (3393MHz) x6
GPU model:Radeon RX 580 Series 26.20.15002.61 (4095MB) x1
Settings
Render:Direct3D11
Mode:1920x1080 8xAA fullscreen
PresetCustom
QualityUltra
Tessellation:Extreme
I am also...
Dec 17, 2019
14
2
15
exact model of the psu?
cpu not going into turbo. did you do a fresh os install. what the cpu temp?
Seasonic FOCUS 450 Gold
Dont have cpu monitoring software. I do know however that the cpu idle temperature is around 30 celcius thanks to the bios
I have not done a fresh windows reinstall either.
 
Dec 17, 2019
14
2
15
That single stick of ram is killing your performance. Ryzen has to have fast ram, and in dual channel. If that windows install is from a previous build, that is not helping you either.
True. I am considering buying another identical ram stick or just sending the stick back and buying some dual channel memory.
 

joeblowsmynose

Distinguished
So I just ran mine - I also have 580 but an 8 gb version.

1st, I'm not sure about userbenchmark 'conclusions" so that is why I ran mine to compare.

https://www.userbenchmark.com/UserRun/22776483

A couple things ... the 4gb variant has 1750mghz effective ram speed, while the 8 gb version has 2000mghz - so my card is going to be faster.

My clock speed is a fair bit higher as well, but yours appears at spec.

Your CPU is also performing "below expectations" on that benchmark ... hmmm

I think the sentiment that its your single channel RAM causing a bottleneck might be true. This would potentially drag down both the GPU and CPU scores - as we see in the results.

Don't buy another stick to "add" (if you have the choice) -- send this one back and and get the dual pack - RAM is binned in pairs and you might take a hit in best timings and clocks, if you pair ram that isn't binned together.

You also might want to try a different benchmark or two as well and see if you are seeing the same deficits there.
 
Last edited:

joeblowsmynose

Distinguished
One more point ... if you get the dual channel ram (you want to anyway, even if that's not causing this issue), and performance is still low look more closely into your CPU ...

Maybe check if your CPU clock speeds are running at spec. (you can do this now of course - report back anything that looks weird there as well)

Ryzen Master, CPUID or HWMonitor should each give you accurate CPU clock speed info - you can observe them while running the test.
 

joeblowsmynose

Distinguished
GPU: AMD RX 580 - 49.1%
UserBenchmarks: Game 49%, Desk 67%, Work 62%

... cpu was performing above average so I think thats just me doing the benchmark one to many times.

This benchmark is tool does not produce consistent results ... I just ran a few tests and the results for both my GPU and drives swing wildly and I didn't even get a CPU score for one run.

I did get my R7 1700 into the 99th percentile though with a bit of OCing :)

Try a different benchmark tool (3D Mark?) , but do replace that RAM with dual channel stuff, regardless.
 
Dec 17, 2019
14
2
15
This benchmark is tool does not produce consistent results ... I just ran a few tests and the results for both my GPU and drives swing wildly and I didn't even get a CPU score for one run.

I did get my R7 1700 into the 99th percentile though with a bit of OCing :)

Try a different benchmark tool (3D Mark?) , but do replace that RAM with dual channel stuff, regardless.
Thanks for the response.

I really don’t think that the cpu is a problem because I experienced the same inconsistent results with the benchmark and I just happend to send a benchmark where my cpu was performing bad.

The dual channel ram is however very much agree on so I am going to purchase some better fast dual channel ram.

Today I haven’t had time to install alternative benchmarking software so I am going to do that tomorrow. But I’ll keep you updated.
 
  • Like
Reactions: joeblowsmynose
Dec 17, 2019
14
2
15
This benchmark is tool does not produce consistent results ... I just ran a few tests and the results for both my GPU and drives swing wildly and I didn't even get a CPU score for one run.

I did get my R7 1700 into the 99th percentile though with a bit of OCing :)

Try a different benchmark tool (3D Mark?) , but do replace that RAM with dual channel stuff, regardless.
Unigine Heaven Benchmark 4.0
FPS:88.0
Score:2216
Min FPS:9.1
Max FPS:167.5
System
Platform:Windows NT 6.2 (build 9200) 64bit
CPU model:AMD Ryzen 5 2600 Six-Core Processor (3393MHz) x6
GPU model:Radeon RX 580 Series 26.20.15002.61 (4095MB) x1
Settings
Render:Direct3D11
Mode:1920x1080 fullscreen
PresetCustom
QualityHigh
Tessellation:Disabled

Here is Heaven Benchmark 4.0
 

joeblowsmynose

Distinguished
Can you run with
Unigine Heaven Benchmark 4.0
FPS:88.0
Score:2216
Min FPS:9.1
Max FPS:167.5
System
Platform:Windows NT 6.2 (build 9200) 64bit
CPU model:AMD Ryzen 5 2600 Six-Core Processor (3393MHz) x6
GPU model:Radeon RX 580 Series 26.20.15002.61 (4095MB) x1
Settings
Render:Direct3D11
Mode:1920x1080 fullscreen
PresetCustom
QualityHigh
Tessellation:Disabled

Here is Heaven Benchmark 4.0


The only thing I could find online to compare is this:


Unigine Heaven Benchmark 4.0
FPS:46.4
Score:1169
Min FPS:13.9
Max FPS:102.7
System
Platform:Windows NT 6.2 (build 9200) 64bit
CPU model:AMD FX(tm)-6350 Six-Core Processor (3900MHz) x3
GPU model:Radeon RX 580 Series 22.19.666.1 (4095MB) x1
Settings
Render:Direct3D11
Mode:1920x1080 8xAA fullscreen
PresetCustom
QualityUltra
Tessellation:Extreme


But this was set with "Ultra" preset and then the resolution brought down to 1080p for the run. You can change to these settings and run again and compare the scores. Note that your CPU is a fair bit better than his, but I suspect you should get roughly the same score (~1150), or maybe a little better.

I can do a run on my system when I get home.
 
Dec 17, 2019
14
2
15
Can you run with



The only thing I could find online to compare is this:


Unigine Heaven Benchmark 4.0
FPS:46.4
Score:1169
Min FPS:13.9
Max FPS:102.7
System
Platform:Windows NT 6.2 (build 9200) 64bit
CPU model:AMD FX(tm)-6350 Six-Core Processor (3900MHz) x3
GPU model:Radeon RX 580 Series 22.19.666.1 (4095MB) x1
Settings
Render:Direct3D11
Mode:1920x1080 8xAA fullscreen
PresetCustom
QualityUltra
Tessellation:Extreme


But this was set with "Ultra" preset and then the resolution brought down to 1080p for the run. You can change to these settings and run again and compare the scores. Note that your CPU is a fair bit better than his, but I suspect you should get roughly the same score (~1150), or maybe a little better.

I can do a run on my system when I get home.
FPS:70.5
Score:1776
Min FPS:14.4
Max FPS:144.9
System
Platform:Windows NT 6.2 (build 9200) 64bit
CPU model:AMD Ryzen 5 2600 Six-Core Processor (3393MHz) x6
GPU model:Radeon RX 580 Series 26.20.15002.61 (4095MB) x1
Settings
Render:Direct3D11
Mode:1920x1080 fullscreen
PresetCustom
QualityUltra
Tessellation:Extreme

Now I am very confused.
Maybe there is some setting which I have enabled/disabled which the benchmark that you send me did not. I do not know which setting that would potentially be.
 

joeblowsmynose

Distinguished
FPS:70.5
Score:1776
Min FPS:14.4
Max FPS:144.9
System
Platform:Windows NT 6.2 (build 9200) 64bit
CPU model:AMD Ryzen 5 2600 Six-Core Processor (3393MHz) x6
GPU model:Radeon RX 580 Series 26.20.15002.61 (4095MB) x1
Settings
Render:Direct3D11
Mode:1920x1080 fullscreen
PresetCustom
QualityUltra
Tessellation:Extreme

Now I am very confused.
Maybe there is some setting which I have enabled/disabled which the benchmark that you send me did not. I do not know which setting that would potentially be.

AA set to 8x in the sample I posted. Your's looks like none -- that would do it. Try adding that in the settings.