Sign in with
Sign up | Sign in
Your question
Solved

If a CPU benchmark test is done on a game, isn't the FPS limited by the GPU as well?

Last response: in CPUs
Share
September 21, 2013 3:20:01 AM

I'm looking at a benchmark between the i7-3770K and FX-8350 (http://) and on Metro 2033 the i7-3770K stock gets 84 FPS. But wouldn't the GPU be bottlenecking it as well? or would they have picked a GPU better than the CPU's?
a b à CPUs
September 21, 2013 3:33:03 AM

kaio37k said:
I'm looking at a benchmark between the i7-3770K and FX-8350 (http://) and on Metro 2033 the i7-3770K stock gets 84 FPS. But wouldn't the GPU be bottlenecking it as well? or would they have picked a GPU better than the CPU's?

Possibly, yes. One way to at least partially alleviate that is to run the benchmark at a much lower resolution, etc, 1024x768, with AA off.
m
0
l

Best solution

a b à CPUs
September 21, 2013 3:34:48 AM

It depends on what card they are using ultimately. In an ideal situation, the card(s) would out class the CPUs by a long shot, so the limits of each CPU would be apparent. In those tests, they were using Crossfired 7970s, so they should have been fine.

Whether or not the GPU is going to be an issue depends mostly on the game. I had a rig with SLIed 560TIs and a FX-6200. At times, frame rates would plummet to 20 to 25 FPS simply because of the CPU, my favorite example being the scene in Africa in the beginning with all the NPCs running around. It was unplayable. One of those cards in with an i5-3470S ran the game like silk. In a case like that, the CPU bottleneck is readily apparent.
Share
Related resources
September 21, 2013 3:46:01 AM

BSim500 said:
kaio37k said:
I'm looking at a benchmark between the i7-3770K and FX-8350 (http://) and on Metro 2033 the i7-3770K stock gets 84 FPS. But wouldn't the GPU be bottlenecking it as well? or would they have picked a GPU better than the CPU's?

Possibly, yes. One way to at least partially alleviate that is to run the benchmark at a much lower resolution, etc, 1024x768, with AA off.


Hmm, OK, thanks! Would you happen to know of any sights that might have a guide on which CPU are par with which GPU's?
m
0
l
September 21, 2013 3:46:20 AM

gbryan101 said:
It depends on what card they are using ultimately. In an ideal situation, the card(s) would out class the CPUs by a long shot, so the limits of each CPU would be apparent. In those tests, they were using Crossfired 7970s, so they should have been fine.

Whether or not the GPU is going to be an issue depends mostly on the game. I had a rig with SLIed 560TIs and a FX-6200. At times, frame rates would plummet to 20 to 25 FPS simply because of the CPU, my favorite example being the scene in Africa in the beginning with all the NPCs running around. It was unplayable. One of those cards in with an i5-3470S ran the game like silk. In a case like that, the CPU bottleneck is readily apparent.


Hmm, OK, thanks! Would you happen to know of any sights that might have a guide on which CPU are par with which GPU's?
m
0
l
a b à CPUs
September 21, 2013 3:54:44 AM

kaio37k said:
Hmm, OK, thanks! Would you happen to know of any sights that might have a guide on which CPU are par with which GPU's?

There was this one from a couple of years ago, but that's outdated now. It's a shame THG doesn't do a new one for 2013:-
http://www.tomshardware.com/reviews/game-performance-bo...

The problem is, different games have different CPU vs GPU ratio's. Eg, Battlefield 3 Single-Player was more GPU limited (and ran 60fps even on some i3's for the most part), but multi-player on the same game was more CPU limited and needed an i5, so there isn't a "one size fits all" for even the same game, let alone all games.

For a i7-3770K, that's as fast as CPU's get and you're not going to be bottlenecked with that. You could drop down to an i5-3570K, and when OC'd to +4GHz, you're still not going to be CPU bottlenecked for most games out there.
m
0
l
a b à CPUs
September 21, 2013 3:57:01 AM

kaio37k said:
gbryan101 said:
It depends on what card they are using ultimately. In an ideal situation, the card(s) would out class the CPUs by a long shot, so the limits of each CPU would be apparent. In those tests, they were using Crossfired 7970s, so they should have been fine.

Whether or not the GPU is going to be an issue depends mostly on the game. I had a rig with SLIed 560TIs and a FX-6200. At times, frame rates would plummet to 20 to 25 FPS simply because of the CPU, my favorite example being the scene in Africa in the beginning with all the NPCs running around. It was unplayable. One of those cards in with an i5-3470S ran the game like silk. In a case like that, the CPU bottleneck is readily apparent.


Hmm, OK, thanks! Would you happen to know of any sights that might have a guide on which CPU are par with which GPU's?


This comes down to so many factors that a single guide would be daunting if not impossible. Primarily, it comes down to what you are doing. For example, BF3 draws more heavily on the GPU than CPU and ran well on my machine that balked at CoD: BlOps II. Some tasks, like Bitcoin mining, are almost entirely GPU dependent, with the CPU there pretty much to keep the computer on while the graphics cards work.

While it's older now, Tom's has an excellent article on bottlenecks I suggest you read.
http://www.tomshardware.com/reviews/game-performance-bo...

This is focused on gaming and not other more esoteric tasks, but the same still holds true.
m
0
l
a b à CPUs
September 21, 2013 3:57:33 AM

And I posted the same link, sorry.
m
0
l
September 21, 2013 4:10:04 AM

BSim500 said:
kaio37k said:
Hmm, OK, thanks! Would you happen to know of any sights that might have a guide on which CPU are par with which GPU's?

There was this one from a couple of years ago, but that's outdated now. It's a shame THG doesn't do a new one for 2013:-
http://www.tomshardware.com/reviews/game-performance-bo...

The problem is, different games have different CPU vs GPU ratio's. Eg, Battlefield 3 Single-Player was more GPU limited (and ran 60fps even on some i3's for the most part), but multi-player on the same game was more CPU limited and needed an i5, so there isn't a "one size fits all" for even the same game, let alone all games.

For a i7-3770K, that's as fast as CPU's get and you're not going to be bottlenecked with that. You could drop down to an i5-3570K, and when OC'd to +4GHz, you're still not going to be CPU bottlenecked for most games out there.


Thanks! I was looking at getting the i5-3570K but wasn't sure if maybe I'd need something stronger.
So "theoretically" speaking, an i5-3770K is a useless upgrade as far as gaming goes (because a 3570K maxes out the best games)?
m
0
l
a b à CPUs
September 21, 2013 4:10:31 AM

gbryan101 said:
And I posted the same link, sorry.

No problem gbryan. It just goes to show how hard it is to find a relatively simple "this GPU for this CPU" chart with recent hardware!

kaio37k - A general simple guideline:-

- For high end i5/i7 chips, just get the best GPU you can afford.

- For mid-range (eg, i3's), go for a 7790/7850 mid-range card.

- For budget Pentium, etc, chips, just get a 7750/7770 budget card.

It can't really be broken down into individual cards for individual CPU's because each game loads at different ratio's. Some MMORPG's can be very CPU heavy even with very weak GFX. Some FPS's can run flawlessly on i3's but have a very heavy load on the GPU.

kaio37k said:
Thanks! I was looking at getting the i5-3570K but wasn't sure if maybe I'd need something stronger. So "theoretically" speaking, an i5-3770K is a useless upgrade as far as gaming goes (because a 3570K maxes out the best games)?

The general consensus is that there are very few games where an i7 runs significantly faster than an i5. The biggest difference is that i7's are better if you do a lot of video encoding / 3D rendering work, and i5's are better for gaming overall (especially if you put the money saved into upgrading your GFX card to the next model up).

Edit: real world example - take ARMA3 a game whose engine has been traditionally CPU heavy:-
http://gamegpu.ru/images/stories/Test_GPU/Action/ARMA%20III%20Alpha/test/arma%203%20proz%20intel.jpg

As you can see, Hyper-Threading (greay bars) makes little difference, and even on an old i7-2600K, the cores (blue bars) are nowhere near maxed out. And that's about 10% slower than the i5/i7 3xxx series you're looking at - at stock speeds.
m
0
l
September 21, 2013 4:12:55 AM

gbryan101 said:
kaio37k said:
gbryan101 said:
It depends on what card they are using ultimately. In an ideal situation, the card(s) would out class the CPUs by a long shot, so the limits of each CPU would be apparent. In those tests, they were using Crossfired 7970s, so they should have been fine.

Whether or not the GPU is going to be an issue depends mostly on the game. I had a rig with SLIed 560TIs and a FX-6200. At times, frame rates would plummet to 20 to 25 FPS simply because of the CPU, my favorite example being the scene in Africa in the beginning with all the NPCs running around. It was unplayable. One of those cards in with an i5-3470S ran the game like silk. In a case like that, the CPU bottleneck is readily apparent.


Hmm, OK, thanks! Would you happen to know of any sights that might have a guide on which CPU are par with which GPU's?


This comes down to so many factors that a single guide would be daunting if not impossible. Primarily, it comes down to what you are doing. For example, BF3 draws more heavily on the GPU than CPU and ran well on my machine that balked at CoD: BlOps II. Some tasks, like Bitcoin mining, are almost entirely GPU dependent, with the CPU there pretty much to keep the computer on while the graphics cards work.

While it's older now, Tom's has an excellent article on bottlenecks I suggest you read.
http://www.tomshardware.com/reviews/game-performance-bo...

This is focused on gaming and not other more esoteric tasks, but the same still holds true.


Thanks! I am looking into it for gaming as well, but for the most part, I like learning about these kinds of things :D 
m
0
l
!