RTX 3060 Ti not keeping up with Ryzen 5 3600?

[merlin95]

Hey all. I've been testing my 3060 ti 8GB and It seems to be not keeping up with my Ryzen 5 3600.

Tested in shadow of the tomb raider running on "highest" setting at 1080p no fps limit. (Photo included) it's saying I'm 36% gpu bound and the graphs show at some points the gpu is faster but at most it's slower than the cpu.

Is this right? The graphs are different each time usually gpu takes longer more than half the time

My overclock settings are

Gpu- power limit 110

Core clock +160

Memory clock +1000

Cpu- 4.1ghz ryzen 5 3600

16gb ddr4 3200 RAM

B450m mortor max motherboard

Game is running from the m.2 drive

My temps are not an issue they reach 80c at max no higher. Im running this all just air cooled.

I have photos included to help.

Thank you anyone :) have a great day all.

[SupaKomputa]

You cannot measure GPU is faster than CPU, it does different thing.

GPU bound meaning the GPU is the bottleneck here, meaning you can get better performance if you put a better gpu with that cpu.

There's nothing to worry about 100fps is just what a 3060ti should do.

[Aereldor]

Time to break out the old bottlenecking analogy. 

 

Think of PC performance as horses pulling a carriage. At 1080p, the GPU isn't too stressed, so the GPU 'horse' can run at full speed. Since the GPU horse is running at full speed, the CPU horse can also run at full speed.

 

At 1440p, that's almost twice as many pixels as 1080p. The carriage is twice as heavy, and the GPU horse can't go as fast anymore. As a result, no matter how strong the CPU horse is, it can only run at the speed set by the GPU horse. 

This can work vice versa... The 'GPU horse' is focused on leading the carriage through stuff like textures, models, complex lighting, render distance. The 'CPU horse' is focused on leading the carriage through logic, large numbers of NPCs, world generation in games like Minecraft. So you'd understandably be limited by the GPU at high resolutions, with high-res textures and high-poly models, and limited by the GPU when there's lots of logic to compute, lots of NPCs etc.

 

This will vary not just from game to game but from area to area - in The Witcher 3, you'd be GPU limited in Skellige with all the foliage and the more demanding lighting, but CPU limited in Novigrad with the hundreds of NPCs.

 

In Shadow of the Tomb Raider, with global illumination/ray tracing, TAA, advanced Ambient Occlusion... Max settings, 100fps is simply the best you'll get with a 3060 Ti. I would recommend using some kind of DLSS solution if you want more frames.

[TylerDurden!]

23 minutes ago, merlin95 said:

have a great day all.

that "bound" thing is sorta misleading, for your system, the gpu is the bottleneck 36% of the time during the benchmark, the other 64% could be a combo of cpu & ram.  If you are at 0% gpu bound, then the rest of your system is the bottleneck for the benchmark.  For your system, I would just call it normal.  But then again you are using GPU that is twice the cost of your CPU

[merlin95]

19 hours ago, TylerDurden! said:

that "bound" thing is sorta misleading, for your system, the gpu is the bottleneck 36% of the time during the benchmark, the other 64% could be a combo of cpu & ram.  If you are at 0% gpu bound, then the rest of your system is the bottleneck for the benchmark.  For your system, I would just call it normal.  But then again you are using GPU that is twice the cost of your CPU

thank you very much for commenting the cost of the gpu to cpu is whats stumping me, ive even got a card that boosts higher than average 3060 ti cards aswell and to see that was so odd... its strange to say but something feels off when using the card, and as far as its ranking in bench marking software its what everyone else is getting from the card too (mines abit higher) 

could there be more driver updates needed to make all games work better on it? it has been out for awhile i know but it never reached enough people. so minimal diagnostic info shared like with gpuz. and such, could that affect it? i could be so wrong.. ive never been lucky enough to get a new gpu not long after launch.

is it worth noting the pci-e slot on my board is gen 3 where as the gpu is gen 4. i know its backwards compatible but is there a performance lose there?

[merlin95]

19 hours ago, SupaKomputa said:

You cannot measure GPU is faster than CPU, it does different thing.

GPU bound meaning the GPU is the bottleneck here, meaning you can get better performance if you put a better gpu with that cpu.

There's nothing to worry about 100fps is just what a 3060ti should do.

do you think its strange that a ryzen 5 3600 is out matching the new 3060 ti? i know they do different things but i do understand somewhat the info shared between the cpu and gpu while gaming. 36% gpu bound means 66% is cpu or ram.

iam running the gpu on a gen 3 pci-e slot, i heard the difference between 4.0 and 3.0 isnt much but still a thought maybe?

 

[AngryBeaver]

40 minutes ago, merlin95 said:

do you think its strange that a ryzen 5 3600 is out matching the new 3060 ti? i know they do different things but i do understand somewhat the info shared between the cpu and gpu while gaming. 36% gpu bound means 66% is cpu or ram.

iam running the gpu on a gen 3 pci-e slot, i heard the difference between 4.0 and 3.0 isnt much but still a thought maybe?

 

People focus too much on the whole bottleneck thing.
 

The truth is it is an optimal situation to have the gpu be the bottleneck. Now if you were to run that test again at 720p then your CPU would be the bottleneck. This mostly comes down to draw calls. Your cpu has to do the calculation for the game, some potential physics, background tasks, and then on top of that do draw calls which are dependent on your fps in the game. So as frame rates go up the work is more cpu dependent, when the resolution goes up and fps decreases then the work is placed on the gpu more and it becomes the limiting factor.

 

The truth is no matter what you do something will always be a limiting factor. For the best case scenario in games it is best for it to be the GPU.

[merlin95]

On 1/21/2021 at 3:17 PM, AngryBeaver said:

People focus too much on the whole bottleneck thing.
 

The truth is it is an optimal situation to have the gpu be the bottleneck. Now if you were to run that test again at 720p then your CPU would be the bottleneck. This mostly comes down to draw calls. Your cpu has to do the calculation for the game, some potential physics, background tasks, and then on top of that do draw calls which are dependent on your fps in the game. So as frame rates go up the work is more cpu dependent, when the resolution goes up and fps decreases then the work is placed on the gpu more and it becomes the limiting factor.

 

The truth is no matter what you do something will always be a limiting factor. For the best case scenario in games it is best for it to be the GPU.

oh perfect, thank you for that, you explained it so well. there will always be a bottleneck from somewhere i forgot about that fact, and knowing the gpu is a better one to have bottlenecking is what i needed to know.,

perfect thank you again