[SOLVED] Can a CPU bottleneck actually worsen a better GPU performance?

[Mr_Furball]

Hello,

I've been searching a bit about the RTX 3060 and 3060 Ti and looking to possibly pair them with an 3700x. From what I can see, the 3060 can be paired with this CPU with almost no games having any problems with bottleneck. With the 3060 Ti, it appears to happen a bit more often. Maybe a 5800x would be ideal to this GPU, but it when it happens, doesn't seem like an extreme bottleneck nor does it happen with every game, but I saw two comparison videos that were kind of weird.

Fortunately, the video is from the same channel and the RTX 2060 numbers on those test were almost exactly the same, so we get sort of a consistecy, but the thing is that the channel compared the RTX 2060 with the RTX 3060. In this first video, the RTX 3060 had a bit of a bottleneck but it's still being used above 90% and gives better performance than the 2060, which is also at 90%, and it makes sense. On the second video, they tested an 2060 with a 3060 Ti and the bottleneck of the Ti was so extreme that it got the same performance than the 2060. Also, in the game prior in the video, COD Warzone, it was actually worse. This brings me two questions. First, I thought that even you make your GPU bottlenecked, it will still give at least the same performance of a non bottlenecked card at it's best and also, it seems very weird that the 3060 vanilla bottlenecks 10-15% and the Ti goes to 40%. This jump seems weird. I think it would make more sense if it was a RTX 3080, I guess. But still, shouldn't the 3060 Ti performe at least at the same level that the 3060 vanilla in Tomb Raider and COD? Can you actually get worse performance with a more powerful but bottlenecked GPU? Also, I know we can't take these videos as 100% true, but we need to have some base of what to believe.

View: https://www.youtube.com/watch?v=hhho0ggXhK4
- 2060 vs 3060 Ti - 14:50 COD Warzone weird bottleneck / 15:50 Tomb Raider weird bottleneck

View: https://www.youtube.com/watch?v=zm7pLfeRnfE
- 2060 vs 3060 - 3:33 COD Warzone bottleneck but still makes sense / 5:11 Tomb Raider

Sorry if my thread was hard to read. Im just confused about this and didn't know in what order to present the questions. Thank you!

 

[USAFRet]

"bottleneck" is one of the most misunderstood, misused, and overused words in current computer jargon.

CPU A and GPU X.
Reasonably matched parts. Same time period.
We see 100FPS at medium graphics level.

Swap in a better, newer GPU Z.
"OMG! This is a bigger bottleneck!!!"

What happens to performance?
We still get 100FPS, but better graphics.

 

[Mr_Furball]

"bottleneck" is one of the most misunderstood, misused, and overused words in current computer jargon.

CPU A and GPU X.
Reasonably matched parts. Same time period.
We see 100FPS at medium graphics level.

Swap in a better, newer GPU Z.
"OMG! This is a bigger bottleneck!!!"

What happens to performance?
We still get 100FPS, but better graphics.

But in both video, the quality of the game is the same and the performance is worse. If the graphical quality of Tomb Raider was in a higher setting in the 3060 Ti comparison and if it was getting the same FPS as the 3060 in it's video, I would have understand that that's the limit of performance for the 3700x CPU, but it's the same settings and actually worse performance. That's why I'm asking.

 

[RodroX]

Hello,

I've been searching a bit about the RTX 3060 and 3060 Ti and looking to possibly pair them with an 3700x. From what I can see, the 3060 can be paired with this CPU with almost no games having any problems with bottleneck. With the 3060 Ti, it appears to happen a bit more often. Maybe a 5800x would be ideal to this GPU, but it when it happens, doesn't seem like an extreme bottleneck nor does it happen with every game, but I saw two comparison videos that were kind of weird.

Fortunately, the video is from the same channel and the RTX 2060 numbers on those test were almost exactly the same, so we get sort of a consistecy, but the thing is that the channel compared the RTX 2060 with the RTX 3060. In this first video, the RTX 3060 had a bit of a bottleneck but it's still being used above 90% and gives better performance than the 2060, which is also at 90%, and it makes sense. On the second video, they tested an 2060 with a 3060 Ti and the bottleneck of the Ti was so extreme that it got the same performance than the 2060. Also, in the game prior in the video, COD Warzone, it was actually worse. This brings me two questions. First, I thought that even you make your GPU bottlenecked, it will still give at least the same performance of a non bottlenecked card at it's best and also, it seems very weird that the 3060 vanilla bottlenecks 10-15% and the Ti goes to 40%. This jump seems weird. I think it would make more sense if it was a RTX 3080, I guess. But still, shouldn't the 3060 Ti performe at least at the same level that the 3060 vanilla in Tomb Raider and COD? Can you actually get worse performance with a more powerful but bottlenecked GPU? Also, I know we can't take these videos as 100% true, but we need to have some base of what to believe.

View: https://www.youtube.com/watch?v=hhho0ggXhK4
- 2060 vs 3060 Ti - 14:50 COD Warzone weird bottleneck / 15:50 Tomb Raider weird bottleneck

View: https://www.youtube.com/watch?v=zm7pLfeRnfE
- 2060 vs 3060 - 3:33 COD Warzone bottleneck but still makes sense / 5:11 Tomb Raider

Sorry if my thread was hard to read. Im just confused about this and didn't know in what order to present the questions. Thank you!

Hi, we forgive you don't worry

Let me recap what I got from all that.

You got a R7 3700X or perhaps wana buy one?, and you are looking to pair it with a RTX 3060 or 3060TI. So you wana know if the R7 3700X will bottleneck (stupid, horrible word, pelase don't use it again, please) limit the performance of any those cards.

I got a few things to ask/tell you,

1. If you already have a PC with this R7 3700, then you didn't listed your system, brand and model of every component please, specially the PSU.

2. We could assume the resolution at what you wana play your games, but that will be very wrong, Will be way better if you share with us the details of your current display (at least resolution and refresh rate).

3. We have no idea if Warzone and Tomb Raider games are important to you, or if in fact you are going to be playing other games. And this is key information because you can not translate the result of two games to the whole universe out there.

4. Yes those videos, those videos shows you a "picture" an instant of game play betwen two GPUs (we wana believe the rest of the system remained the same). Problem with this is that those videos are from january and february 2021. And thats important because Windows 10 had a lot of updates, and yeah nvidia launched countless (well you can count them, Im not goin to do it) of drivers. So the information/results probably wont hold today if we redo the test (could be better or worst, who knows).

5. Finally, Will the R7 3700X limit the RTX 3060/TI?, perhaps in some games, specially if you play at lower resolution (720p/1080p). Would a R7 5800X be a better CPU than the R7 3700X, yes of course, but also the R5 5600X is, and it cost a lot less than the R7 5800X. Heck even a Core i5 10400 will be a better gaming CPU than the R7 3700X (and yes this one cost even less than the R5 5600X).

6. Don't confuse "bottleneck" I mean a CPU limiting a GPU (or viceversa) with a sudden/sporadic frame drop, those may or not be related to the CPU at all.


So yeah, there are lots of stuff that can be said about this topic. I could keep writing, but is late and I need to go sleep.

Cheers!