EVGA GTX 1070 ti underperforming

[Qupter]

Hello I am experiencing some heavy underperforming from my 1070 ti When I compare my heaven benchmark results to others. It gets beaten by a GTX 1070 with the same CPU (

)

 

What could  the problem be?

 

Specs:

I5 7500 3.2ghz (tops at 56°c)

MSI b150m bazooka

8gb ddr4 balistix tactical 2666mhz

EVGA GTX 1070 ti black edition (tops at 69°c ( ͡° ͜ʖ ͡°))

seasonic focus plus 650w

120gb SSD (13gb free)

1tb wd blue (80gb free)

 

i am using it in a open air Testbench so thermal limiting are out of the question.

 

thanks in advance

[Princess Luna]

DDU in safe mode and reinstall drivers fresh new.

 

Make sure CPU is not bottlenecking it, it's not a powerful enough chip for multi-tasking along side the latest games.

[birdys91]

From my experience with i5 6600, I will say it's probably the CPU bottlenecking it. Similar thing happened to me when there are some background apps running when I had my i5 6600

[Zeitec]

7 hours ago, Qupter said:

 

Well you're being beaten by 3 fps in the video you showed. That's basically margin of error. What does that mean? You're hard CPU bottlenecked by dude. A quad core CPU at 3.6ghz ain't gonna cut it for a 1070ti. You're basically running a Ryzen 3 2200g at 4ghz with a 1070ti. If it was a 7600k at 4.8ghz it would probably be fine, but alas, you'll need to upgrade your CPU. You can either look for a relatively cheap 7700 or 6700 on ebay, or get something like a Ryzen 5 2600x/i5 9600k and build a whole new PC.

[Qupter]

8 hours ago, Zeitec said:

Well you're being beaten by 3 fps in the video you showed. That's basically margin of error. What does that mean? You're hard CPU bottlenecked by dude. A quad core CPU at 3.6ghz ain't gonna cut it for a 1070ti. You're basically running a Ryzen 3 2200g at 4ghz with a 1070ti. If it was a 7600k at 4.8ghz it would probably be fine, but alas, you'll need to upgrade your CPU. You can either look for a relatively cheap 7700 or 6700 on ebay, or get something like a Ryzen 5 2600x/i5 9600k and build a whole new PC.

The person in the video uses a GTX 1070 not a 1070 ti, and still beats me how? We are using the same CPU.

[RAM555789]

18 minutes ago, Qupter said:

The person in the video uses a GTX 1070 not a 1070 ti, and still beats me how? We are using the same CPU.

Because the CPU is your bottleneck it's holding you back.

[Sett]

Err... One thing's for sure, and that's that you are not bottlenecked by somewhat modern mainstream CPU in old GPU benchmark, that uses about 10% of it.

I'd start by making sure you are set to "prefer maximum performance" and not "Optimize compute performance" disabled for Heaven in NVCP, and that you don¨t have a browser open in the background.

If that doesn't help, you might want to upload the benchmark run with some metrics shown (via e.g. Afterburner), namely GPU & CPU utilisation and temperatures nad RAM usage.
 

[Qupter]

2 hours ago, RAM555789 said:

Because the CPU is your bottleneck it's holding you back.

I know but explain how he beats a 1070 ti with a 1070

[GabeThePCHelper]

Silicon lottery maybe, but i'm going to have to say the bottleneck of a non-K processor is holding the 1070ti back.

[GabeThePCHelper]

18 minutes ago, Qupter said:

I know but explain how he beats a 1070 ti with a 1070

>Bottleneck man

 

Your non-K processor is having trouble with the 1070ti, and since the 1070 is a less of a bottleneck it could be helping performance. It could even be your ram, since you're only running 8gb's of the stuff.

[RAM555789]

5 hours ago, Qupter said:

I know but explain how he beats a 1070 ti with a 1070

I do suggest trying other benchmarks and actual gaming and compare them. Limiting yourself to one single kind of test can easily produce skewed results that don't seem to show the full picture.

[Zeitec]

9 hours ago, Qupter said:

The person in the video uses a GTX 1070 not a 1070 ti, and still beats me how? We are using the same CPU.

You're losing by 3 FPS. That's margin of error. In other words, you're getting the same performance. Why's that? Because your 1070ti isn't runing at full speed because your CPU isn't powerful enough to give your GPU as much work as it's capable of. If you ran that benchmark at 4k to put more of the load onto the GPU, your 1070ti would come out ahead.

 

In other words, you're both running at the limits of your i5 7500, not your respective graphics cards. Your graphics card COULD run faster if it had a better CPU, but it doesn't so they perform the same. It's similar to why a 9900k at a Ryzen 3 2200G perform identically when playing games at 1440p on an RX 580 at ultra settings. The 9900k absolutely could put out way more frames than the Ryzen 3, but the GPU can't keep up with the 9900k, so they perform the same. In that case the GPU bottlenecks the CPU. In your case, the CPU is bottlenecking the GPU. 

 

edit: To put it in really simple terms, it doesn't matter how wide the pipe is if the terminating end is only 3cm diameter. A 10cm diameter pipe terminating in a 3cm diameter nozzle will push through precisely the same amount of water as a 1m diameter pipe terminating in a 3cm diameter nozzle because the limiting factor isn't the rest of the pipe, it's the nozzle.

[Zeitec]

9 hours ago, Sett said:

Err... One thing's for sure, and that's that you are not bottlenecked by somewhat modern mainstream CPU in old GPU benchmark, that uses about 10% of it.

I'd start by making sure you are set to "prefer maximum performance" and not "Optimize compute performance" disabled for Heaven in NVCP, and that you don¨t have a browser open in the background.

If that doesn't help, you might want to upload the benchmark run with some metrics shown (via e.g. Afterburner), namely GPU & CPU utilisation and temperatures nad RAM usage.
 

Not true. I ran a 3570k at 4.5ghz (at least as fast as that 7500 at 3.6ghz) and was bottlenecked reasonably hard in Heaven with my 1080 (basically an overclocked 1070ti) at 1440p max settings. It's entirely possible he's bottlenecked on his 7500.

 

edit: By the way topping at 69 in the benchmark is pretty low for that card, indicating lower than maximum usage, indicating a CPU bottleneck.

[SpookyCitrus]

9 hours ago, Qupter said:

I know but explain how he beats a 1070 ti with a 1070

When a CPU bottlenecks a GPU it lowers performance of the GPU, So in other words your CPU is hindering the performance of your GPU and causing it to under perform. So your CPU may be the same as his but he has a less powerful GPU, the CPU you have is a bigger bottleneck with your GPU, your CPU paired with a 1070 is a somewhat more balanced system so it's performance is better with the hardware it uses. Your 1070ti needs to be paired with a more capable CPU that would create less of a bottleneck and thus maximize performance and the ability for your hardware to work better together. Your CPU is about a 25% bottleneck on your 1070ti anything over 10% is going to cause some major performance drops. Even with his 1070 the 7500 is a 17% bottleneck but still that's marginally better than yours. I'd suggest upgrading to an I7 that is compatible with your motherboard or get a new config with a better processor and more ram. Ryzen would be a decent alternative if you're on a budget, a 2700x which is incredibly good for multi-threaded performance is only $299 on Amazon currently and you can even get it paired with a motherboard for around $350 if the deal is going on. And DDR4 ram has gone down a lot in price, 16GB 2x8GB kits used to be around $200 or even more, now they're down to about $110 I believe for corsair LPX 3000mhz.

[Sett]

Just because GPU doesn't run at 100% load doesn't mean CPU is the problem, especially if it's more pronounced at higher resolution. CPU load does not scale with resolution. On the contrary; because the GPU is puting out fewer FPS, there is less work for CPU preparing them.
In HB, @80-ish FPS I get around 6-8% utilisation of my i7-8700k meaning, worst case scenario, about 50% use of one 4,8GHz core. I'm positive I5-7500 can put out more than that.

It COULD be, however, that the OP still gets CPU bottlenecked in "easier" (for GPU) scenes, where the FPS can spike to well over 300 on that card and that might just be too much for a 3,2GHz CPU. Might be some lost points there, but probably not a problem in general.
 

Seeing as temps are not a problem, i'd suggest overclock (at least with "enhanced turbo" or whatever) and see if it makes any difference..
PS: on ARK the i5-7500 is listed as 3,4/3,8 GHz, not 3,2?