Testing CPU bottlenecks using multiple CPUs and a GTX 1080ti

[(G{in}[AK)TION]]

My curiosity is making me want to test for how much of a CPU bottleneck there is with certain CPUs paired with a Geforce GTX 1080ti.

So i bought an Asus Strix Geforce GTX 1080ti to do the testing myself. Here is a list of hardware to be tested.

Specifications of Hardware to be tested


AMD FX-8350
ASUS M5A99FX PRO R2.0
Asus nVidia Geforce 1080ti
G.Skill 16GB of DDR3-1866Ram to be tested at 1600mhz instead
Corsair H100v2 (2 fans in push)

Intel Core i7 3770k
Asrock Extreme4 z77
Asus nVidia Geforce 1080ti
G.Skill 16gb of DDR3-1600 memory
Noctua NH-D14

Intel Core i7 7700k
Gigabyte Z270X-GAMING SOC
Asus nVidia Geforce 1080ti
G.Skill 16GB of DDR4-3200
Corsair H100v2 (2 fans in push)

Intel Core i7 8700k
Asus ROG Strix z370-E GAMING
Asus nVidia Geforce 1080ti
G.Skill 16GB of DDR4-3200
Corsair H100v2 (2 fans in push)

AMD Ryzen 2700x
Asus Crosshair VII
Asus nVidia Geforce 1080ti
G.Skill 16GB of DDR4-3200
Corsair H100v2 (2 fans in push)

Intel Core i9 7900x
Asus Rampage Apex VI
Asus nVidia Geforce 1080ti
G.Skill 16GB of DDR4-3200
Corsair H150i Pro RGB (6 fans on the rad in push pull config)

results posted in the link below.

Asus Strix Geforce 1080ti results
https://docs.google.com/spreadsheets/d/19Bhuf6Z5mizXWo6P3YrJB5bAmadN-7mdEyG1MU9ENAM/edit?usp=sharing

MSI GAMING Geforce 1070
https://docs.google.com/spreadsheets/d/1sOdWGJfcLlDLNkSXZ0WdiU3tHUe5IUkzfLh4wOy6G1c/edit?usp=sharing

If theres anything you disagree with, then enlighten me. What can i do better?

 

[UltraTaco]

This is interesting! Taco subwoofered!
Since mine is slightly above that c2q, it'll be great to see the results(I'm at 1366)

 

[(G{in}[AK)TION]]

installing doom, Rez, and Nier Automata.

@UltraTaco

glad you like this thread. any programs or game you would want to use on a 1080ti?

 

[UltraTaco]

I don't know yet since I don't have one, but I read conflicting results where some write it'll be bnecked by older cpu and others say it's perfectly fine..

 

[EarthDog]

Its been done before and the results will vary wildly by title, resolution, settings. 1080p and a 1080ti most titles will be held back by a 10 year old cpu. The higher the res, the less of a role the cpu plays.

Id stick to 'real' games though.

 

[Kenrou]

Off the top of my head any of the Total War series will be a good test of your hardware (especially Total War: Warhammer 2). Deus Ex: Mankind Divided, Mass Effect: Andromeda, MechWarrior Online, Battlefield 4 (i think tests CPU in large maps).

 

[EarthDog]

Off the top of my head any of the Total War series will be a good test of your hardware (especially Total War: Warhammer 2). Deus Ex: Mankind Divided, Mass Effect: Andromeda, MechWarrior Online, Battlefield 4 (i think tests CPU in large maps).

Cpu heavy titles which some will show more differences than others. Cant really test MW online because its online. If there is sp or an i integrated bench however... bf4 doesnt have a built in bench so its hard to repeat.

q9650? Why not pick up a cheap 775 p4 or p4 celly for like 2-3$ on ebay. The results would be neet to see.

Something even slower? Ick. Can a p4 even run a 1080ti without the pcie bandwidth getting in the way already?

 

[(G{in}[AK)TION]]

@EarthDog

I know that it's been done before, but I figured I try testing with some of the games I like to play along with some other titles to get a much more better idea of how they would run across different cpus.

And by 'real' games do you mean just games only? No graphics bench marks like 3dmark?

 

[EarthDog]

Yeah, in this case, I would stay away from synthetics honestly and just test actual games where you can. The synthetics (in particular 3dMark and anything with CPU speficic tests), along with being done before, will also skew results quite a bit due to most having specific CPU tests which these old processors will choke on.

 

[UltraTaco]

I'm baffled that a quad-core could act as a bottleneck. Even if clocked over? How can a game consume 4 cores to 100%?

 

[EarthDog]

I'm baffled that a quad-core could act as a bottleneck. Even if clocked over? How can a game consume 4 cores to 100%?

Easily. Many moden games do. A 4c/4t processor is OK now, but will show signs of struggling in several titles that are able to use over 4t...

Welcome to modern gaming... quads are just now becoming what most would (should) consider to be a minimum spec for gaming. 2c/4t chips are long in the tooth in some titles as well. In most it will be fine, but, as I said earlier, as time goes on they will age a lot faster.

 

[(G{in}[AK)TION]]

@EarthDog

kind of reminds me of the ryzen effect.

https://lanoc.org/review/cpus/7755-amd-ryzen-7-2700x-and-ryzen-5-2600x

"suddenly six" LOL

 

[EarthDog]

Sorry, TL : DR, what was your point there?

Are you saying Ryzen had an effect on core count needed in games or something??

 

[Bill Dimwit]

Something even slower? Ick. Can a p4 even run a 1080ti without the pcie bandwidth getting in the way already?

Probably not, even at pice 2.0 speeds. Still would be cool to see. I recall seeing folks with p4's back when the gtx 470/480 were new and it was a laughing stock even then. I can't imagine how bad it would be with today's games and with a 1080ti.


EDIT:
http://www.overclock.net/forum/69-nvidia/793791-gtx-470-pentium-4-a.html

 

[EarthDog]

If its PCIe 2.0, then there is 1-2% off the top. I was thinking it was PCIe 1.0 back then which really puts a hurting on things.

 

[Bill Dimwit]

If its PCIe 2.0, then there is 1-2% off the top. I was thinking it was PCIe 1.0 back then which really puts a hurting on things.

ohh, yeah if OP has a older 775 board it would be pcie1.0.
Seeing he has a q9650 its likely a newer board with 2.0. But it could be a p35 board with PCI-e 1.0, most p35's support the q9650.
Could be a G or Q or 975X chipset board too, some have support for the q9650 and have PCI-e 1.0 or 1.1


EDIT:
OP if you don't mind, nab one of these.
http://www.cpu-world.com/CPUs/Celeron_D/Intel-Celeron D 320J - B80547RE056256 (BX80547RE2400C).html
It would be funny to see the results.

 

[Kenrou]

Cpu heavy titles which some will show more differences than others. Cant really test MW online because its online. If there is sp or an i integrated bench however... bf4 doesnt have a built in bench so its hard to repeat.

How about Crysis ? it still +- taxes modern hardware and it has a benchmark (or you download it from TPU, don't recall which) ? sadly MW:O doesn't have a benchmark or SP but MW:5 might when it comes out, it's based off Unreal Engine 4.

 

[(G{in}[AK)TION]]

@EarthDog

That, and also because intel decide to follow AMD with a 6c/12t Coffeecake processor.

@Kenrou
I just read your post right after i posted results. will be testing crysis 1 and 3. I do not have crysis 2 in my games library unfortunately.

 

[UltraTaco]

Dear sir [garbled username],
You could use fraps software. It has benchmarking hotkey and records minimum/maximum/average FPS in simple to use text file! Really helps keep track. You just push the button and it begins and when you push it again, it stops. Like a hotkey, for example F12 to turn benchmarking on/off

 

[(G{in}[AK)TION]]

Dear sir [garbled username],
You could use fraps software. It has benchmarking hotkey and records minimum/maximum/average FPS in simple to use text file! Really helps keep track. You just push the button and it begins and when you push it again, it stops. Like a hotkey, for example F12 to turn benchmarking on/off

ill give it a go then.