PC Upgrade for v4 - Thoughts?

[niyoko 7]

10 minutes ago, Lenny777 said:

Niyoko what app are you using to test?

 

 

I used MSI Afterburner, then made the charts with Excel. I had to manage Afterbuner's log file per run. I also have to change the file extension from .hml to .csv to easily import into Excel.

[Lenny777 299]

On 6/16/2017 at 8:00 AM, eXpress48 said:

Hey guys

My PC, on which I use Prepar3D v3, has the following Specs:

i7 4790K (OC: 4.6 GHz)
4x8 GB RAM (DDR3 / 2400 MHz)
MSI GTX 970 (GDDR5 4GB)
256 GB SSD Samsung 840 Pro (Windows)
1 TB SSD Samsung 850 EVO (Data + P3D)

I'm currently getting a good Performance out of it, but I have most of the sliders in the Lower to Medium Section, due to the beloved VAS Limit. This finally changes with v4 as we all know. I'm currently holding off on v4, until all my core addons are fully compatible

In the meantime, I wanted to identify the Parts of my Setup, which I could improve to get a superb Experience with v4. From my Point of View the Processor + RAM should be enough, or in other words the Upgrade to Kaby Lake + DDR4 won't really make that much of a difference. The SSD should also do its job.

I think the Graphic Card would be the bottleneck in my setup and I'm currently thinking about upgrading to a 1080 or even a 1080 Ti.

Do you guys have any other suggestion or has anyone made the jump from a 970 to a 1080/1080 Ti?

Thanks :)

What's your mobo? I think the Haswell are stuck with the LGA 1150 socket - not compatible with Kaby Lake. So a new processor - a new MOBO. I'd get the 1080, regular or ti.

[Lenny777 299]

1 minute ago, niyoko said:

I used MSI Afterburner.

Thanks

[MehdiAli 5]

56 minutes ago, niyoko said:

@eXpress48
OK. I sat down and did some testing. I ran a test at night with the GTX 970 and 1080 since most have been saying that the dynamic lighting is an impact on their FPS. I also included a day/night test with 4xSSAA and 8xSSAA on the 1080.

Mods
PMDG 777-200 Base, FsDreamteam KCLT, FlightBeam KSFO-HD, ORBX Global, ORBX Northern California, ORBX Trees.

(All runs used Prepar3D's Fair Weather preset.)

Averages (2xSSAA)

GPU Usage: 64.47
Framerate: 29.46

Averages (2xSSAA)

GPU Usage: 39.79
Framerate: 30.43

Averages (2xSSAA)

GPU Usage: 39.75
Framerate: 31.38

Would there be a way to increase GPU usage on the 1080? there isnt much point as its only using about 40% of its capability. 

[niyoko 7]

1 minute ago, MehdiAli said:

Would there be a way to increase GPU usage on the 1080? there isnt much point as its only using about 40% of its capability. 

From my base  settings, I saw a significant increase in GPU usage when I went from 2xSSAA to 8xSSAA. I also have played around with Dynamic Reflections increased my memory usage, but it's not factored into the charts that I posted. Both settings also had impacts on Framerate.

With 8xSSAA I saw a noticeable difference in the night time lighting. I didn't notice much of a difference during the day. At night, the street lights were clearer and more like pin-points of light. Their was also less shimmering in the night lighting.

[MehdiAli 5]

3 minutes ago, niyoko said:

From my base  settings, I saw a significant increase in GPU usage when I went from 2xSSAA to 8xSSAA. I also have played around with Dynamic Reflections increased my memory usage, but it's not factored into the charts that I posted. Both settings also had impacts on Framerate.

With 8xSSAA I saw a noticeable difference in the night time lighting. I didn't notice much of a difference during the day. At night, the street lights were clearer and more like pin-points of light. Their was also less shimmering in the night lighting.

How was the FPS when using 8XSSAA ?

[eXpress48 20]

@niyoko Thank you very much for taking the time to put this together. Interesting that the GPU Memory Usage never got close to the LM suggested 8 GB. You also got pretty constant 30 fps with the 1080. This is a 1080 Non-Ti correct?

[Lenny777 299]

2 hours ago, MehdiAli said:

How was the FPS when using 8XSSAA ?

I just did some tests on another thread and 8XMSAA will be my new choice. Virtually no differences visually between that and 8XSSAA. And using the airport,plane etc. for a longer panoramic view of my plane and airport in that test GPU usage for 8XSSAA is at 100% for external views and 8X MSAA at 70%. The FPS I tested showed a roughly 50% FPS difference in favour of 8XMSAA. Even a an external closer view to the plane and airport was showing 100% GPU for 8XSSAA.

 

Using 4xAF on both tests.

[niyoko 7]

8 hours ago, MehdiAli said:

How was the FPS when using 8XSSAA ?

The FPS was little lower with 8xSSAA, so not that huge as seen on the graph, but it pegged the GPU to 100%.Like Lenny777 is wrote, I also personally didn't feel like there was that much of a reward. I'd like to put the extra resources to other settings.

I think I might go back and see the difference of 2xSSAA and 8xMSAA.

6 hours ago, eXpress48 said:

@niyoko Thank you very much for taking the time to put this together. Interesting that the GPU Memory Usage never got close to the LM suggested 8 GB. You also got pretty constant 30 fps with the 1080. This is a 1080 Non-Ti correct?

I was using the Gigabyte GTX1080 G1.