[SOLVED] RTX 3070's low usage and determining cause of bottleneck ?

[Sepi Florez]

Good day to you all,

For some time I've been struggling with the fact that my RTX 3070 does not seem to be giving me the performance I was expecting. I play a lot of different games like Elden Ring, Spiderman Remastered, Apex Legends, Valheim. And in most games I have experienced the GPU usage being nowhere near its maximum capicity and also frametimes being really unstable. The weird thing is I have not been able to prove a bottleneck and I would like some help to make sense of this all.

My specs:
GPU: Gigabyte GeForce RTX 3070 Gaming OC 8G
CPU: AMD Ryzen 5 2600
Motherboard: ASRock B450M Pro4
Bios version/date: American Megatrends Inc. P1.60, 05/11/2018
RAM: G.Skill Aegis F4-3000C16D-16GISB (2x8gb)
OS: Windows 10 Pro 19044.1889
Storage: Samsung 850 EVO 250gb (windows installation drive) and Samsung 980 1tb
PSU: Corsair RM750 (age unclear as it was a 2nd hand purchase)
Main screen resolution: 2560 x 1440

From my observations it seems that my GPU almost never reaches anywhere near 90-100 percent usage but sits around 50-60% (depending on the game ofcourse). I would then think that my CPU might be the bottleneck, but I have not been able to see that in afterburner. Most of the times my CPU/CPU cores do not reach anywhere near max usage.

Now recently I started playing the new spiderman remastered and booted up afterburner to see if my GPU would be bottlenecked again and for the first time since getting my 3070 I saw that my CPU was actually being used to it maximum capacity. Now I tried overclocking my CPU to see if that would improve my GPU usage but it just lowered the CPU usage and the GPU usage stayed about the same (around 50%).

Now this endeavor with spiderman remastered really confused me and makes me doubt anything I know about troubleshooting this problem so would like to ask you guys for help and finally answer my question as to why my GPU not being used to its maximum capacity? I would also appreciate any advice you can give about resolving this issue for instance any upgrades I can make. Thanks in advance!

EDIT: After the overclock of the CPU and setting it back to stock settings, the CPU usage does not reach 100% anymore (in Spiderman Remastered). Still around the same GPU usage though, which makes this problem even weirder.

 

[Lutfij]

Welcome to the forums, newcomer!

You forgot to include the make and model of your PSU as well as it's age. BIOS version for your motherboard at this moment of time? What OS are you working with? SSD?

 

[jeremy0118]

Your processor is too weak for a 3070. According to bottleneck calculator its about 30% on 1080p which is quite significant

When playing games your graphic card won't perform on maximum possible performances (won't be utilized to the maximum), because you processor won't process and deliver data fast enough. This theoretically means that utilization of you processor will be on maximum and the potencial of graphic card will not be completely used.

 

[Sepi Florez]

Welcome to the forums, newcomer!

You forgot to include the make and model of your PSU as well as it's age. BIOS version for your motherboard at this moment of time? What OS are you working with? SSD?

I've updated my post to include the missing information.

 

[USAFRet]

Your processor is too weak for a 3070. According to bottleneck calculator its about 30% on 1080p which is quite significant

While that processor is a bit weak for that GPU, the "bottleneck calculator" is pure snake oil.
Junk science.
Should be ignored.

 

[USAFRet]

Numbers seem pretty accurate to me, where is your proof? anyways i dont need bottleneck calculator to tell him that cpu is far too weak for a 3070. "abit" is underselling it lol.

'proof' that the bottleneck calculator is junk?

It gives no context.
In Game A, that combination might be fine.
In Game B, not so fine.

Some games are more CPU, others more GPU.

Then, this:

Same system, 9 months apart.

 

[jeremy0118]

'proof' that the bottleneck calculator is junk?

It gives no context.
In Game A, that combination might be fine.
In Game B, not so fine.

Some games are more CPU, others more GPU.

Then, this:

Same system, 9 months apart.

Maybe the numbers changed because of modern game requirements. Ill just say with the calculator having a 12600kf combined with a 3070 at 1080p concidered a match in heaven, we are definetly on the same page. is it something you can trust with your life? nah. does it give a general idea, yes. is it junk science, maybe. of course you have cpu and gpu heavy titles , but on an average reading it does decent. i should have left it out tho.

 

[drivinfast247]

Is that a single 16gb stick or 2 8gb? Are they installed in slots a2-b2?

 

[Sepi Florez]

Is that a single 16gb stick or 2 8gb? Are they installed in slots a2-b2?

2 sticks of 8gb each. They are installed in slots a2 and b2, yes.

 

[Vic 40]

Motherboard: ASRock B450M Pro4
Bios version/date: American Megatrends Inc. P1.60, 05/11/2018

Vey old bios, update to the latest,
First install the latest AMD chipset drivers,
https://www.amd.com/en/support/chipsets/amd-socket-am4/b450
or use the autodetect,
https://drivers.amd.com/drivers/ins...in-edition-22.8.2-minimalsetup-220825_web.exe

then one by one install all later bios up to 5.30, lets leave the beta bios alone for now..
https://www.asrock.com/mb/AMD/B450m Pro4/#BIOS
Lutfij is very adamant on doing it that way.

 

[Sepi Florez]

Vey old bios, update to the latest,
First install the latest AMD chipset drivers,
https://www.amd.com/en/support/chipsets/amd-socket-am4/b450
or use the autodetect,
https://drivers.amd.com/drivers/ins...in-edition-22.8.2-minimalsetup-220825_web.exe

then one by one install all later bios up to 5.30, lets leave the beta bios alone for now..
https://www.asrock.com/mb/AMD/B450m Pro4/#BIOS
Lutfij is very adamant on doing it that way.

Thank you for your suggestion. I updated my BIOS version to the latest one and also installed the latest AMD chipset drivers. Both actions did not seem to fix the issue.

 

[Vic 40]

Your processor is too weak for a 3070.

Starting to think this might well have some truth in it.

From my observations it seems that my GPU almost never reaches anywhere near 90-100 percent usage but sits around 50-60% (depending on the game ofcourse). I would then think that my CPU might be the bottleneck, but I have not been able to see that in afterburner. Most of the times my CPU/CPU cores do not reach anywhere near max usage.

That the cpu doesn't reach 100% doesn't mean it's not a bottleneck, seen with Ryzen that it hardly reaches 100% in games, maybe the the old 4 core ones like the 1200/1300 and even maybe the 3300x, think either a driver issue or some other kind of core management. Seems that only bench tools like Cinebench seem to be able to make it run like that.

Ram is running at rated spec? XMP enabled?

You play the games at 1440p?

Some of the same games in the next two video's,

with 2600/3070

View: https://www.youtube.com/watch?v=XKoHVFzKxfA


with 5600x/3070

View: https://www.youtube.com/watch?v=Mqw1qpGPne0


you see differences in those and yes in not all of those games alot of difference, but gpu depency can be different and map/scene as well.

 

[Sepi Florez]

That the cpu doesn't reach 100% doesn't mean it's not a bottleneck, seen with Ryzen that it hardly reaches 100% in games, maybe the the old 4 core ones like the 1200/1300 and even maybe the 3300x, think either a driver issue or some other kind of core management. Seems that only bench tools like Cinebench seem to be able to make it run like that.

So you are saying that even though my CPU is not at 100% usage, it could still be the bottleneck in this scenario? Is there any way I can determine this?

Ram is running at rated spec? XMP enabled?

XMP enabled yes, running at 3000mhz.

You play the games at 1440p?

Some of the same games in the next two video's,

with 2600/3070
View: https://www.youtube.com/watch?v=XKoHVFzKxfA


with 5600x/3070
View: https://www.youtube.com/watch?v=Mqw1qpGPne0

you see differences in those and yes in not all of those games alot of difference, but gpu depency can be different and map/scene as well.

The videos you posted seem to have some big differences between settings and scenes. Not sure what I can derrive from comparing the two videos.

 

[Vic 40]

So you are saying that even though my CPU is not at 100% usage, it could still be the bottleneck in this scenario? Is there any way I can determine this?







The videos you posted seem to have some big differences between settings and scenes. Not sure what I can derrive from comparing the two videos.

Showed the video's as an aswer to the first question. Even with different scenes and afaik pretty much same settings you see quite a big difference between the two cpu's where the 5600x is capable of letting the 3070 go to ~100%. The 2600 always trails and both cpu don't reach 100% usage, something people have to accept with Ryzen.

 

[Sepi Florez]

Showed the video's as an aswer to the first question. Even with different scenes and afaik pretty much same settings you see quite a big difference between the two cpu's where the 5600x is capable of letting the 3070 go to ~100%. The 2600 always trails and both cpu don't reach 100% usage, something people have to accept with Ryzen.

I see now, so this means that the 2600 was just holding my 3070 back all this time and that not reaching 100% usage is just a ryzen thing. From looking at the videos, the 5600x seems like a worthwhile upgrade to get more out of the 3070. Would you recommend upgrading?

 

[Vic 40]

I would upgrade with such a gpu yes. Think it's worth it. The 5600 (non x) is also an option and likely somewhat cheaper.

difference,

https://pcpartpicker.com/product/PgcG3C/amd-ryzen-5-5600-36-ghz-6-core-processor-100-100000927box

https://pcpartpicker.com/product/g94BD3/amd-ryzen-5-5600x-37-ghz-6-core-processor-100-100000065box

that is US, migth be that you live somewhere else and that the difference in price might be smaller.

Just look for Ryzen 5600 reviews, see what the difference in performance is.

 

[Sepi Florez]

I would upgrade with such a gpu yes. Think it's worth it. The 5600 (non x) is also an option and likely somewhat cheaper.

difference,

https://pcpartpicker.com/product/PgcG3C/amd-ryzen-5-5600-36-ghz-6-core-processor-100-100000927box

https://pcpartpicker.com/product/g94BD3/amd-ryzen-5-5600x-37-ghz-6-core-processor-100-100000065box

that is US, migth be that you live somewhere else and that the difference in price might be smaller.

Just look for Ryzen 5600 reviews, see what the difference in performance is.

I think I'll consider upgrading to fix the issue. Thanks for your help and information!

 

[Vic 40]

You're welcome.