Fps Drops while running two Steam titles at once

[HongBi]

 

Hello Everyone,

The games I run together are Dota 2 and Rocket League I like to play RL in between D2 games so I leave RL on in the background. When RL is on my fps from D2 goes from about 100 average to about 50 average.

My setup:

4690k Stock
Msi gaming 5 z97
8gb corsair vengence @ 1600
gtx 970 sli
rog swift 144hz 1440p monitor along side a 1200p monitor.

While gaming with both games on have about 70% cpu usage and 70% ram usage. 

 

I'm wondering if I am being limited by my hardware or if i'm getting fps drops because of poor steam game utilization.

 

 

[dragoon20005]

of course you game will become slow

 

you are running two games which are sharing your PC resources

 

esp CPU and RAM

[manikyath]

well.. if you're running two games, expect both of them to only perform half as well...

[Enderman]

youre getting fps drops because youre trying to run two games at once...

 

what did you expect would happen lol

[Ophidio]

Try not running two games at once... yeah... that'll work... your hardware is fine and steam is fine it's just you trying to run multiple games at once. .-.

 

Two CPU demanding games nonetheless... 

[Curufinwe_wins]

Making sure both are in full windowed will probably help.

 

With a i5, running two games is really cpu stressful even with one gpu. I 100% would never ever recommend sli on an i5.

 

The 0 thread is just getting pounded too hard.

[Kinda Bottlenecked]

I can actually run League and Blade and soul at the same time though... Both 60fps. But RL might just be very demanding. 

[MblaZe7run]

Obvious solution is obvious

 

Run one game, get good fps. Run two, get half fps

 

/thread

[Samfisher]

Making sure both are in full windowed will probably help.

 

With a i5, running two games is really cpu stressful even with one gpu. I 100% would never ever recommend sli on an i5.

 

The 0 thread is just getting pounded too hard.

What..? SLi performance has nothing to do with an i5..

[Curufinwe_wins]

What..? SLi performance has nothing to do with an i5..

Actually it has everything to do with an i5. SLI has an insane amount of overhead and i7's show massive improvements even in "GPU bound" environments over i5s. 

 

Fuck, x99 shows huge improvements over z series i7s under sli.

 

Just a small example video.

 

 

 

Not the best thing in the world since most people/reviewers don't even consider it being a possible issue, but even in single gpu situations Digital Foundry has shown massive microstutter improvements to going to an quad core i7, and given the known fact that 2-way gpu considerations have much higher overhead (and SLI having higher cpu overhead than CF) it shouldn't be any surprise that said effects manifest themselves.

[matto97]

maybe turn all the Graphics and res down for Dota to see if that helps the perf of both your games

[Samfisher]

Actually it has everything to do with an i5. SLI has an insane amount of overhead and i7's show massive improvements even in "GPU bound" environments over i5s. 

 

Fuck, x99 shows huge improvements over z series i7s under sli.

 

Just a small example video.

 

 

 

Not the best thing in the world since most people/reviewers don't even consider it being a possible issue, but even in single gpu situations Digital Foundry has shown massive microstutter improvements to going to an quad core i7, and given the known fact that 2-way gpu considerations have much higher overhead (and SLI having higher cpu overhead than CF) it shouldn't be any surprise that said effects manifest themselves.

That probably has more to do with the game using more CPU cores rather than an SLi test.  Both GPUs are at 90%+ so the only difference is the CPU, which means it's a CPU bound game, and not indicative of any SLi performance.

[Curufinwe_wins]

That probably has more to do with the game using more CPU cores rather than an SLi test.  Both GPUs are at 90%+ so the only difference is the CPU, which means it's a CPU bound game, and not indicative of any SLi performance.

First off, it's 4k ultra settings on very graphically demanding titles (arguably 3 of the top 10 most demanding out there today), there literally is nothing less gpu bound for that configuration.

 

GPUs are in both cases effectively tapping out (or as much as gpu's can tap out in SLI, you basically never see 99/99 even in a GPU only synthetic you rarely will see even 99/97).

 

It is "gpu bound" in the conventional sense that both processors tap the gpus and you still see performance increases from increasing gpu core/memory speeds.

 

Additionally, if you wanted to claim the benchmarks given are too cpu bound then you are basically saying no processor yet made can keep up with even two way sli GM200 on the most demanding titles, at the highest available resolution to consumers, which is lunacy. 

 

So many people wrongly assume you won't see a difference in performance by changing cpus even when "gpu bound". 

 

 

The additional cores (and much larger cache likely) allow the sli overhead to be spread more evenly (unlike DX11 main threads, the sli overhead is fairly parallel) which in turn allows the normal DX11 loading to hit threads more efficiently (although the lower overhead load on thread 0 is the biggest difference).

 

 

 

The point is once you start running sli, you see these issues (regardless of what kind of "boundness" you want to claim) prop up on every game and every setting. And if it always happens then clearly it's an issue you as a consumer should address.

 

Again GTA V example (because apparently russians are the only ones who actually do this type of testing and apparently they love GTA V)

 

https://youtu.be/ew5MIgXfHL0?t=299

 

@6:48 you can see W3, 8:25 is BF:4, 10:02 is crysis 3.

 

 

Here is a great example of classic CPU bottle-necking of SLI systems (note low gpu load and high differences in load in comparison to the original video I linked by the same person). (near matching frequency i5/i7 oc'd and oc'd 980 sli)

 

 

 

 

So really I guess the statement is, even if you claim that really it's a cpu limitation and not "indicative of sli" then that I show significant performance reductions even with 970 sli using a 4690k over a 4790k should really explain why like I said from the beginning "I 100% would never ever recommend sli on an i5."

[killcomic]

Oh good lord. Don't you know that when you run 2 games at once the little elves that live inside your computer have time management issues causing everything to slow down?

 

To solve this, you need to sacrifice a goat surrounded by black candles and chant the second paragraph in page 12 of the Necronomicon.

Once Satan has been summoned, you sell him your entire family's soul in exchange for a cardboard box.

Now, this is the important bit, you get your computer and place it inside that box, put it outside your house and write on the side with virgin's blood "I don't have a clue how computers, or electronics in general, work."