A Question About Bottleneck

[goodie]

Hi,

 

I have a question about bottleneck and I couldn't find an answer to that. 

 

There are some gameplay videos on Youtube which shows the performance of a very powerful card along with 2 powerful CPUs. 

 

Let's say the components are GTX1070, i5 6500 and i7 6700 and the game is GTA 5 which is considered as a cpu intensive game. 

 

GTX1070 + i5 6500 gets around 70 fps and the cpu load doesn't hit 100% and gpu load stays at 100%. 

 

Then there is the second situation where they use GTX1070 + i7 6700. The cpu load doesn't hit 100% again but the fps is actually higher. 

 

Can we say that there is a bottleneck here?

 

Cpu load on i5 is generally higher. Is it the cause of this situation? 

 

I assume cpu load should be close or equal to 100% to talk about a bottleneck situation. 

 

Thanks in advance and sorry for my English.

 

 

[don_svetlio]

i5 6500 will bottleneck the 1070 at 1080p. i5 6600K will reduce it but it still bottlenecks.

[FloRolf]

Yeah, the i5 is the bottleneck in this situation.

High CPU load is generally bad.

[Godlygamer23]

 It depends on the how the load is spread. The 6700 has a higher clock speed, including a higher boost, and hyper-threading, and therefore will perform better.

Edited July 23, 2016 by Godlygamer23

[goodie]

Thanks for the answers. Gpu load is always 100%. This is where I'm lost. Shouldn't it be lower if there is a bottleneck?

[HKZeroFive]

Just now, goodie said:

Thanks for the answers. Gpu load is always 100%. This is where I'm lost. Shouldn't it be lower if there is a bottleneck?

It's not a bottleneck if that's the case. GTA V takes advantage of more threads and higher clocks, resulting in a higher framerate.

[goodie]

7 minutes ago, HKZeroFive said:

It's not a bottleneck if that's the case. GTA V takes advantage of more threads and higher clocks, resulting in a higher framerate.

So, getting a lower fps doesn't necessarily mean there is a bottleneck. This is what I wanted to learn by starting this topic.

 

But if we compare it to i7 6700, there is a bottleneck in theory, right? I mean we get lower fps by using an i5. Maybe its weaker architecture causes a bottleneck.

[Godlygamer23]

Just now, goodie said:

So, getting a lower fps doesn't necessarily mean there is a bottleneck. This is what I wanted to learn by starting this topic.

 

But if we compare it to i7 6700, there is a bottleneck in theory, right? I mean we get lower fps by using an i5. Its weaker architecture cause a bottleneck.

The architecture is the same between the two CPUs. The difference is the clock speeds, cache, and the number of threads supported concurrently.

[HKZeroFive]

Just now, goodie said:

So, getting a lower fps doesn't necessarily mean there is a bottleneck. This is what I wanted to learn by starting this topic.

 

But if we compare it to i7 6700, there is a bottleneck in theory, right? I mean we get lower fps by using an i5. Its weaker architecture cause a bottleneck.

Technically, yes, the i5 is a bottleneck in games that utilise more threads.

[goodie]

Just now, Godlygamer23 said:

The architecture is the same between the two CPUs. The difference is the clock speeds, cache, and the number of threads supported concurrently.

Yes, they are both Skylake cpus. I meant the other specs that you mentioned here.

[goodie]

2 minutes ago, HKZeroFive said:

Technically, yes, the i5 is a bottleneck in games that utilise more threads.

Thank you and others who responded. I understand now. Have a nice day.

[Enderman]

CPUs can be a bottleneck wtihout being at 100%

You can see this also from benchmarks of the FX8350.

CPU usage is usually around 50%, but it gets worse fps in most cases than an i5 or i7.

[goodie]

1 minute ago, Enderman said:

CPUs can be a bottleneck wtihout being at 100%

You can see this also from benchmarks of the FX8350.

CPU usage is usually around 50%, but it gets worse fps in most cases than an i5 or i7.

It makes sense. Thanks.

[agent_x007]

I don't see why you want "xx%" on GPU/CPU, when the only thing that matters is actual FPS in game.
If GPU would be at 25% and CPU at 35% (but you would get that 60FPS minimum), would you call that a bottleneck ?
BTW : Above situation is possible for an old game, with new hardware and V-sync switched on.

Think about it.

[DragonTamer1]

The i5 is bottle necking the 1070, however the 1070 is still running over 60 FPS which is the mark we are going for. Therefor the bottleneck during game play most likely won't be noticed (except for maybe spikes).

 

I only consider a bottleneck as a major issue if it is noticeably causing the FPS to drop below the monitors refresh rate. Vsync and Freesync help hide it but it can still be noticed.

[goodie]

14 hours ago, agent_x007 said:

I don't see why you want "xx%" on GPU/CPU, when the only thing that matters is actual FPS in game.
If GPU would be at 25% and CPU at 35% (but you would get that 60FPS minimum), would you call that a bottleneck ?
BTW : Above situation is possible for an old game, with new hardware and V-sync switched on.

Think about it.

You're right. I won't be upgrading my cpu to get 10 fps more because I already play games well above 60 fps but I just wanted to learn the reason behind this i5 vs i7 situation. 

[SkilledRebuilds]

You also DON'T have to have a CPU @ 100% to get a CPU bottleneck...

Dropped frames can happen without 100% CPU utilization on any core.

 

My only concern with bottlenecking, is MIN-FPS values on slower CPU's, if my MINFPS are fine (IMO, we all vary with whats acceptable),... and my averages are close to where they should be, I'm happy.