GPU/CPU Bottleneck

[Garrosh]

                     Hello there, this is my first post on Linus  and sry if my post will almost an article like and kinda long but i need help with this  

 

 

So the title is basicly regarding the topic that am asking about , i read alot about cpu and gpu bottleneck and asked some friends of mine even and yet i cant 100% understand if thats a real issue or something that a person should rly take care of or its just a hoax so i decided to raise a case example of myself and a friend of mine and from the answers i get i might understand.

So currently i have a PC thats specs : i5 6500 on an

                                                              H170 gigabyte gaming 3 mobo
                                                                GTX 970
                                                                  1x16gb ram

My main activities are ofc gaming and when i built this PC was main goal 60 fps but now i wanna play any triple AAA game and specially Cyberpunk when it comes out next year and my current games like WOW Startcraft 2 Diablo OW etc on highest FPS possible or lets say 144fps to 120fps averagely on 1080p reso.

My question now is with my i5 6500 can i lets say buy an rtx 2080 super or if i manage to actually wait for Ampere in 2020 and get whats equivlant to 2080 super by then which will ofc be much tougher , will there be a bottleneck coz this cpu would be so weak to accomodate such high end gpus that would limit the gpu potential and limit me in fps and if its a no would this basicly mean i can keep upgrading my gpu to the end of times so long as my cpu is alive and games only optimize 4 cores and doesnt need more.
When exactly as a gamer i should worry that my cpu is weak and needs to be upgraded to coexist with high end gpus to give me the highest FPS possible.


The answer to this concern will be rly important for me to let me spend money in the right direction as right now i have 2 routes i can go for , either grab rtx 2080 super (or just wait for ampere next year) and get a 1080p 144hz monitor with it /or  else i start building a new full out setup consisting of a new case even (i7 9700k/Gigabyte Aorus Ultra Z390/850W Gold PSU/ H500M/360 mm Cooler etc ) and then wait till 2020 and grab an ampere high end card since if i do this setup build now  i wont afford an rtx 2080 now

 

<PS:if u are wondering why would i buy a new setup now without a new GPU and get a new gpu next year and why wouldnt i just wait for next year and get all at once its simply because i dont like spending a hefty bunch of funds all at once as this pains me so i am basicly building it in steps to >

[2SidedPolygon]

Your friend's build almost certainly is experiencing major bottlenecking, and with an upgrade to a 2080 (Super) I imagine yours would as well. If you gamed at higher resolutions, it wouldn't matter as much, but your post suggests that you care more about framerate, in which case the CPU's limitations would begin to show. I wouldn't, however, get a new build without a GPU either. If I were in your position, I would wait until I had enough money to make a new build with a GPU, and stick with what I had until then.

[Nord]

Well, there is a literal ton of information and questions regarding "bottlenecking" & how one would or should define it.

But rather than writing a 5.000 word story on it, I suggest these videos to you:
(don't let the titles put you off, both sooner or later will make sense in regard to the question you are asking)

Test was run with an overclocked Titan X on 1080p. The card is obviously "weak" by today's standards but it's still about 2x as fast as a 970 would be and should sit around the 1660 ti / 2060 (stock) area in terms of performance. Also they got a lot of CPU test videos similar to this, starting at the 2500 up to the 8700 and everything in between, plus they are putting in and showing proper testing methodology on the topic, which near to no reviewer, including LTT, fails to do, averages mean nothing to actual gaming performance, its simply the fastest way to test)

 

So just watch these and make your own opinion on the subject.

 

 

And as a personal note, game engine & optimisation matters the most for FPS, which is why most reviewers will almost always benchmark the same games. Because, simply put, most games are garbage in these aspects. Including big hits such as fortnite.

For example starcraft 2 has an utterly crappy single core based engine, so you wont have high FPS in that game, ever. In fact, you can be happy if you stay at 60 at all times and nothing you can put in your PC will change that. Maybe in 50 years when a single CPU core runs on like 20GHZ but until then, no such luck.
While overwatch on the other hand is very well optimised (so I hear) which should just lead to "minor" difference in the CPU used, unless you try to shoot above 144fps.

But this would lead to the next subject on the matter, which would be that FPS are the least important, it's just the entry “phrase” that people learn. Way more important than FPS are frametimes and even more important than that are stable frame times but that's a whole other topic in itself. (Which again you could learn from both of the tech channels linked above or battlenonsense on YT)

[BTGbullseye]

On 9/20/2019 at 12:47 PM, Nord said:

For example starcraft 2 has an utterly crappy single core based engine, so you wont have high FPS in that game, ever. In fact, you can be happy if you stay at 60 at all times and nothing you can put in your PC will change that.

So I must have the best PC ever... I can keep it at 120FPS constant at maximum settings at 1080p with a Ryzen 3600 and a 5700 XT.

[Nord]

7 hours ago, BTGbullseye said:

So I must have the best PC ever... I can keep it at 120FPS constant at maximum settings at 1080p with a Ryzen 3600 and a 5700 XT.

 

You indeed must have the best PC ever

Or you are simply playing 1v1 and end the games once the 5 pool did it's job, because once it gets into the mid to late game or some of the things start happening that the engine cant handle (like baneling exploding into bios, mothership + cloak effects on units, certain particle effects...) , this is what you get:

 

 

^13% cpu usage, 19FPS (and yes, thr guy is using a i5 2500, but still, with that utilisation, what does it matter.)

 

 

 

Also as a fun sidenote, Blizzards "Heros of the storm" uses the same engine and struggels equally as sc2 does.

But hey, I’m always up to being proven wrong. Setup an OSD and record your stable 120FPS.
If blizzard actually fixed the shitty performance I might even try the game again. Maybe we can even duo up, you any good?

 

[BTGbullseye]

I don't do multiplayer competitive strategy games... I don't like being forced into rushing everyone all the time just to have the slightest chance of winning.