1080ti + R5 1600 = Low GPU Usage

[iiNNeX]

For some reason in a few games (Battlefield 1, GTA V and a few other AAA titles) my 1080ti is only at 50-65% usage (fluctuating). That would normally indicate a CPU bottleneck, correct?

 

I play on 2560x1440 on a 165hz monitor. The R5 1600 is overclocked to 3.9ghz and I assumed it would not be an issue. For example Rainbow Six Siege uses 95% GPU and I get 160+ frames at all times, but a large portion of the others don't. My Firestrike and Cinebench scores are better than the average 1600 + 1080ti system so it should be working as expected, meaning it's probably the cpu...

 

Any input on this?

[Mira Yurizaki]

2 minutes ago, iiNNeX said:

For some reason in a few games (Battlefield 1, GTA V and a few other AAA titles) my 1080ti is only at 50-65% usage (fluctuating). That would normally indicate a CPU bottleneck, correct?

 

I play on 2560x1440 on a 165hz monitor. The R5 1600 is overclocked to 3.9ghz and I assumed it would not be an issue. For example Rainbow Six Siege uses 95% GPU and I get 160+ frames at all times, but a large portion of the others don't. My Firestrike and Cinebench scores are better than the average 1600 + 1080ti system so it should be working as expected, meaning it's probably the cpu...

 

Any input on this?

Low GPU usage alone isn't indication of a CPU bottleneck. You also need the CPU to have a disproportionately high usage or CPU usage that is capped.

[iiNNeX]

3 minutes ago, M.Yurizaki said:

Low GPU usage alone isn't indication of a CPU bottleneck. You also need the CPU to have a disproportionately high usage or CPU usage that is capped.

Hmm here are some screenshots.

 

 

[stealth80]

As above, probably not, especially at 1440P. I would look at drivers and ram usage to start with. Also ensure your ram us running 3000mhz as it will default to 2133mhz on Ryzen 

[iiNNeX]

Just now, stealth80 said:

As above, probably not, especially at 1440P. I would look at drivers and ram usage to start with. Also ensure your ram us running 3000mhz as it will default to 2133mhz on Ryzen 

Ram is running to 3000 with XMP profile enabled, RAM usage is posted above in screenshots. I have 16gb total so it should be fine.

 

Drivers are the latest on a fresh Win 10 Pro copy.

[Simon771]

It would be game engine bottleneck. 

Some games (MMORPGs mostly) will benefit from strong single cores.

 

For example my GTX 1070 usage was 50%, and CPU (Ryzen 1700) was 30%. Yet I got 30fps in BDO.

Now I'm using i5 8600k at 60% load and GTX 1070 working at 80-90% ... getting smooth 60-75fps in BDO.

 

Hard to say what's the cause of your problem, but I'm certain that if you put i5 8600k or 8700k in your system, you would have much higher fps. At least those are my experiences with some games.

 

Overwatch had no problems on R7 1700, delivering me same amount of fps as now on 8600k, so it kinda depends on the game and it's engine.

[Mira Yurizaki]

1 minute ago, iiNNeX said:

Hmm here are some screenshots.

Average CPU usage is too low to be considered a bottleneck. It would also be better to post CPU usage from Task Manager in the "Logical Processors" view.

[stealth80]

1 minute ago, iiNNeX said:

Ram is running to 3000 with XMP profile enabled, RAM usage is posted above in screenshots. I have 16gb total so it should be fine.

 

Drivers are the latest on a fresh Win 10 Pro copy.

try DSR and run BF1 at 4k, see if GPU usage goes up 

[stealth80]

1 minute ago, Simon771 said:

It would be game engine bottleneck. 

Some games (MMORPGs mostly) will benefit from strong single cores.

 

For example my GTX 1070 usage was 50%, and CPU (Ryzen 1700) was 30%. Yet I got 30fps in BDO.

Now I'm using i5 8600k at 60% load and GTX 1070 working at 80-90% ... getting smooth 60-75fps in BDO.

 

Hard to say what's the cause of your problem, but I'm certain that if you put i5 8600k or 8700k in your system, you would have much higher fps. At least those are my experiences with some games.

 

Overwatch had no problems on R7 1700, delivering me same amount of fps as now on 8600k, so it kinda depends on the game and it's engine.

That would be case if GTA5 and BF1 weren't multi threaded. The usage is around 50% so it isn't a bottleneck - BDO is a heavily single threaded game which actually quite rare now in terms of the newer games

[Majestic]

You're monitoring CPU usage, not per core. I'm pretty sure BF1 doesn't utilize 12 threads. 6 threads is more likely. Which corresponds nicely to the 54% CPU usage.

[stealth80]

5 minutes ago, iiNNeX said:

Hmm here are some screenshots.

 

 

also try again in DX12

[iiNNeX]

5 minutes ago, Simon771 said:

It would be game engine bottleneck. 

Some games (MMORPGs mostly) will benefit from strong single cores.

 

For example my GTX 1070 usage was 50%, and CPU (Ryzen 1700) was 30%. Yet I got 30fps in BDO.

Now I'm using i5 8600k at 60% load and GTX 1070 working at 80-90% ... getting smooth 60-75fps in BDO.

 

Hard to say what's the cause of your problem, but I'm certain that if you put i5 8600k or 8700k in your system, you would have much higher fps. At least those are my experiences with some games.

 

Overwatch had no problems on R7 1700, delivering me same amount of fps as now on 8600k, so it kinda depends on the game and it's engine.

I am planning on getting a 8700k system and just transferring my 1080ti over, Ryzen is unfortunately not good enough single thread wise to reach 165frames in most competitive games I play. It's fine on the casuals though.

4 minutes ago, M.Yurizaki said:

Average CPU usage is too low to be considered a bottleneck. It would also be better to post CPU usage from Task Manager in the "Logical Processors" view.

I will split the cpu cores and take another screenie

4 minutes ago, stealth80 said:

try DSR and run BF1 at 4k, see if GPU usage goes up 

Will do

[Majestic]

3 minutes ago, stealth80 said:

also try again in DX12

No, DX12 is horrendous.

 

 

Core scaling takes a significant diminishing returns after 6 threads. Meaning his CPU usage will probably be just a tad over 50% given his 12 threads. It's a CPU bottleneck, but i'm seeing 80-92fps (should be good, no?). If you were aiming for 144hz you should've gotten the 8700K and overclocked it.

[iiNNeX]

DX12 on 4K (almost)

 

 

 

 

Time to order an 8700k and a delid kit I guess (:

[Princess Luna]

Yes it is a CPU bottleneck even if the CPU isn't maxing out, Ryzen is not the solution one should ever buy to pair it with a 1080 Ti for 165hz gaming lol... the person above explained it very well just one note though @Majestic the i7 8700 suffices perfectly for the matter, gaming it always is around 4.4ghz~4.5ghz on the latest IPC that is already way faster than Ryzen 5 1600 at 3.9ghz.

 

He could have gotten the appropriated CPU and still save money.

check the benchmarks down: already set the time:

Spoiler

 

 

[iiNNeX]

1 minute ago, Princess Cadence said:

Yes it is a CPU bottleneck even if the CPU isn't maxing out, Ryzen is not the solution one should ever buy to pair it with a 1080 Ti for 165hz gaming lol... the person above explained it very well just one note though @Majestic the i7 8700 suffices perfectly for the matter, gaming it always is around 4.4ghz~4.5ghz on the latest IPC that is already way faster than Ryzen 5 1600 at 3.9ghz.

 

He could have gotten the appropriated CPU and still save money.

check the benchmarks down: already set the time:

  Reveal hidden contents

 

 

 

This is my temporary PC man, I just put the 1080ti in to see how it will game while waiting to get the i7 kit. Also it has to be the K, 5ghz+ is the only way

 

Cheers anyway, even at 4K the Ryzen CPU has ridiculously high usage lol

[Mira Yurizaki]

I decided to go back and see what the R5 1600 should be getting in DX11. It's around 110-120 FPS for 1080p, and I did find a 1440p benchmark that showed it at ~90 FPS.

 

So... there's no hardware bottleneck per se. The game just isn't designed well enough.

[Princess Luna]

Just now, iiNNeX said:

5ghz+ is the only way

To spend money without need at such high resolution you still face GPU bottleneck regardless but I won't argue, just one note, I'm not a man.

 

Cheers for finding your bottleneck, happy upgrading.

[iiNNeX]

1 minute ago, M.Yurizaki said:

I decided to go back and see what the R5 1600 should be getting in DX11. It's around 110-120 FPS for 1080p, and I did find a 1440p benchmark that showed it at ~90 FPS.

 

So... there's no hardware bottleneck per se. The game just isn't designed well enough.

Game ran perfectly on my 6700k and 1080 SLI system that I had before this, so I guess it's just Ryzen in this instance.

 

1 minute ago, Princess Cadence said:

To spend money without need at such high resolution you still face GPU bottleneck regardless but I won't argue, just one note, I'm not a man.

 

Cheers for finding your bottleneck, happy upgrading.

True however you can achieve those frames with an i7 and slightly more optimized settings, and apologies didn't know  What FPS does your non K CPU get on CPU intensive games compared to 5ghz variants ?

 

Thank you!

[Princess Luna]

Just now, M.Yurizaki said:

So... there's no hardware bottleneck per se. The game just isn't designed well enough.

The software is the "bottleneck" in the sense of not utilizing all the CPU threads well however due to it the CPU with better threads it manages utilizing will see vantage in performance. The upgrade still makes sense... seems OP had the idea all along regardless.

[Princess Luna]

4 minutes ago, iiNNeX said:

Thank you!

I wish I could help you with numbers but my bad/good luck is that I put the computer together and never got to use it because the second hand super cheap killer deal on G.Skill Trident Z memory end up getting me a kit where one of the sticks was faulty, kinda reason why the super good price heheh... I didn't mind it though, shipping from Brazil to USA for the RMA was still cheaper than buying it new and now I'm receiving a brand new replacement kit... worth it for savings despite the hassle and been waiting  a whole month to actually boot up this system.

 

At least it's almost here already...

[iiNNeX]

1 minute ago, Princess Cadence said:

I wish I could help you with numbers but my bad/good luck is that I put the computer together and never got to use it because the second hand super cheap killer deal on G.Skill Trident Z memory end up getting me a kit where one of the sticks was faulty, kinda reason why the super good price heheh... I didn't mind it though, shipping from Brazil to USA for the RMA was still cheaper than buying it new and now I'm receiving a brand new replacement kit... worth it for savings despite the hassle and been waiting  a whole month to actually boot up this system.

 

At least it's almost here already...

Damn sorry to hear, it's lovely piece of kit that ram, I really want a 3600mhz version of it for my new build but it will have issues fitting under the NH-D15 unless I change one of the fans to a 120mm, but then it will cover the ram sticks so i wont see the lovely RGB ...Maybe a reason to go custom water cooling again ? lol

[stealth80]

52 minutes ago, Majestic said:

No, DX12 is horrendous.

 

 

Core scaling takes a significant diminishing returns after 6 threads. Meaning his CPU usage will probably be just a tad over 50% given his 12 threads. It's a CPU bottleneck, but i'm seeing 80-92fps (should be good, no?). If you were aiming for 144hz you should've gotten the 8700K and overclocked it.

I never said it wasn't, I wanted to know if the usage changed by trying it, also I would take that review with a pinch of salt as the review hardware states "DDR3 2133" ...

 

firstly Ryzen is DDR4,

 

secondly, everyone knows it performs terribly on slow ram and considering they pushed they res down to 720p to prove CPU efficiency and then crippled the inifinity fabric in the process doesn't provide very accurate testing results - I'm not saying it should be run on 4000mhz ram, but I think 3000mhz is the norm, mine is running 3200mhz and difference at 1080p from 2133 to 3200 was substantial, it would be even greater at 720P and then to top it off they use 4 dims .... 

 

 

 

 

[Majestic]

Absolute performance is not the intended goal of the test. Even if absolute performance suffered due to the poor memory choice, it wouldn't negate core scaling.

[xg32]

it's a cpu bottleneck, i would not have paired a 1080 or a 1080 ti with ryzen, as it cannot handle 120/144, let alone 165 hz in some games.