First observations with GTX 970 - puzzled

[chatman 29]

I recived my GTX970 today (EVGA SC) and was eager to try it with p3d 2.4.

Turned everyting to max (except shadows) and in "normal" scenarios I get great frames. I then tried a hardcore-scenario (New York, bad weather) and only got single digit, max 13 frames/s.

What makes me wonder is that I had GPU-Z running in the background and it showed a maximum GPU use of only 22% :blink:

Also max. memory use was at about 2 GB.

 

I thought that the GPU would run at 100% and therefore give me better frames. Did I miss something?

 

Res is 1920x1080, Win 7 64, i5 2500K, 8GB RAM

 

Thanks!

 

 

[HamSammich 17]

Simple. You've just discovered that P3D, like FSX, is CPU-bound.

 

That means the weakest link, for all of us, is processor speed. Until Flight simulators get massively rewritten to run more code, when possible, on the GPU, we're all stuck with it.

[chatman 29]

Well, I was aware of that but I thought with all the comments that P3D is much more dependant on GPU-Power that this would be a massive improvement with the GTX970. I mean - 20%??? - Would be interesating to see which figures other people with this card get!

[Manny 1,293]

 

 

Well, I was aware of that but I thought with all the comments that P3D is much more dependant on GPU-Power that this would be a massive improvement with the GTX970. I mean - 20%???

 

Hmmm... Interesting. I too was thinking the same thing.. and I was even wondering if I should move from my current 670 to the new 980.

[Kuragiman 227]

I think FSX, which is the base platform that P3D is built from, followed the same WRONG assumption that companies like Sony did when developing games like Everquest 2.   They assumed that processors would get faster and faster and didn't even consider the fact that instead of faster, manufacturers would just add more cores.  So they based their original code around faster processors coming instead of multi-cored processors coming.  As a result, this CPU dependent simulator still laughs at us when we upgrade to the latest super massive awesometastic video card which does almost nothing for us frame rate wise.  It's infuriating.   I spent $5000 on my PC build, specifically to run FSX (this was before I knew the facts about the simulator with all the tweaking guides), only to see dismal performance on a system that was chewing up modern video games on a triple monitor setup.   It still gets me bent to think about it but what's done is done.  Until they completely scrap ESP and write a new x64 multi-core base code in a simulator we will always have this struggle of FPS vs. Eye-candy wants.

[chatman 29]

Hmmm... Interesting. I too was thinking the same thing.. and I was even wondering if I should move from my current 670 to the new 980.

Don't get me wrong, the card is fabulous! I switched from a GTX 570 and according benchmarks the new card is about 3 times faster. I just though I would benefit a little more in low frame situations. On the other hand I maxed out almost all sliders. So turning them down a little when flying in dense environments should ease matters

[chatman 29]

Here are 2 screenshots of my systems  workload:

 

 

Green line is Core 1 of my CPU and is the only one always occupied

 

And this is a screen of the GPU utilisation:

 

 

For this test I used the same scenario (New York) all sliders maxed out (except shadows at medium)

 

Question is: which settings will make use of the unused GPU capacities without lowering performance? So for example how much AA and Anisotropic filtering can the card handle? How about texzure size (actally using 1024 - would 2048 or 4096 be possible without reducing frames?

[Adam T Lutley 12]

Just a minor point that may affect the performance of your new GPU. I'm assuming that you're running a Z 68 PCI 2.0 mainboard of some sort. The 970GTX being PCI 3.0, might be a tad handicapped. I know the performance hit shouldn't be big 10% max, but it is a factor. Regards

[chatman 29]

Yes Adam, you're absolutely right - Z68 board 

I didn't even know that there is a 3.0 already

 

But a quick search on the net confirmed your assumption - they don't differ too much in performance.

 

Thanks !

[Fi5kuS 57]

This is whats keeping me from upgrading my GTX 760 to a 980. Doesn't seem to gain much FPS on this.

[SuperCell 10]

This is why I still have my 7850... my 4.2Ghz 3570k does all the work.

[rototom 198]

P3D has a lot of evolving to do. That's why I'm not even looking that way. I also fear that that process of change will create a lot of growing pains.

 

You might find that if you use something like Active Sky / REX that increased cloud coverage will create spikes in GPU and graphics card memory usage. I found that my FSX system performed better in weather after a graphics card upgrade. However, I can't explain why.

[Anderscn 0]

How is your AA setting? Would be interested to see how the GTX 970/980 handle 8xSGSS in Inspector with DX10.

 

BR,

Anders

[MarkW 1,490]

How is your AA setting? Would be interested to see how the GTX 970/980 handle 8xSGSS in Inspector with DX10.

 

BR,

Anders

 

I have 8xSGSS and in Fixer have gone as high as 16cQCSAA but I can't see any difference in image quality or framerates.  Card never goes above 50% usage.

 

I actually think my framerates are lower with the card but the experience seems smoother, no flashing, no shimmering, occasional pause but I had a lot of pauses with the 560Ti.

[Johnny19 344]

 

You might find that if you use something like Active Sky / REX that increased cloud coverage will create spikes in GPU and graphics card memory usage. I found that my FSX system performed better in weather after a graphics card upgrade. However, I can't explain why.

That's what I noticed too, since I fly in both FSX and P3D, ok it's a ROG gaming laptop so it won't perform miracles, but in FSX i get steady 25-30 FPS with ASN heavy weather, in P3D performace drops to teens.