Why so little CPU & RAM usage?

Bill Stuntz

Bill Stuntz

Well-known member
Here's a screen shot of System Explorer's performance page while lazily flying around. I landed and/or changed airports a couple times.

My system is a 4-Core 8-thread Xeon 3.6, with 32G RAM, GTX-760. RF8 is using only about 700MB RAM & only 2 cores are showing much action. RF7.5 looks about the same except that the 2 active cores are different ones. Why is my hardware so under-used?
 

Attachments

  • System Explorer.jpg
    System Explorer.jpg
    146.9 KB · Views: 13
Bill Stuntz said:
Here's a screen shot of System Explorer's performance page while lazily flying around. I landed and/or changed airports a couple times.

My system is a 4-Core 8-thread Xeon 3.6, with 32G RAM, GTX-760. RF8 is using only about 700MB RAM & only 2 cores are showing much action. RF7.5 looks about the same except that the 2 active cores are different ones. Why is my hardware so under-used?
Click to expand...

RF mostly run on the GPU (GTX-760).

Edit The CPU and ram don't matter to much except for loading content from the Harddrive/ssd to the GPU
 
Last edited:
RF8 is based on a fairly old piece of Software, and it probably doesn't make a lot of use of multiple cores on todays systems and it really doesn't need a lot of power to run. It can run on 10+ year old computers just fine. It is also a 32bit software so it is rather limited in how much ram it can take.

Load a software Like EVGA Precision, and it can tell you how much of your video card memory is being used. I use about 700mb of my total Video ram when I run RF8. Compare that to my RF-X memory usage which is more....a lot more. about 6gb on average, sometimes more, sometimes less.
 
So basically, RF sends minimal raw data to the video card - which effectively does all the physics calculations & rendering? And the CPU doesn't calculate anything? Can't the CPU's multiple cores calculate the physics & let the GPU render what it's calculated?

Why does everyone seem to want to spend big bucks upgrading to the newest, hottest CPU & M/B if it won't improve performance & frame rates? Wouldn't they be better off going with the newest, hottest video card instead, and leave their already under-used CPU alone? Is it because they play games that are more CPU intensive than RF?
 
Bill Stuntz said:
So basically, RF sends minimal raw data to the video card - which effectively does all the physics calculations & rendering? And the CPU doesn't calculate anything? Can't the CPU's multiple cores calculate the physics & let the GPU render what it's calculated?

Why does everyone seem to want to spend big bucks upgrading to the newest, hottest CPU & M/B if it won't improve performance & frame rates? Wouldn't they be better off going with the newest, hottest video card instead, and leave their already under-used CPU alone? Is it because they play games that are more CPU intensive than RF?
Click to expand...

The physics is done on the CPU and that takes a lot of horsepower for a complicated program, RF-8 isn't but RF-X is. Plus the CPU gets the data from the HD or SSD and stores it in the GPU and when the GPU needs another set of data the CPU fetches it and stores it in the GPU. In a demanding game the CPU and GPU work hand in hand and if you don't have a fast CPU and GPU your performance goes down a lot. RF-X needs both a Fast CPU and Fast GPU to perform the way you want it too, that's why a good gamming computer has the latest CPU and GPU. They both work together and you need both fast or things slow down at different times for different reasons when one or the other gets overloaded. Like many have said RF-8 is very old technology so it doesn't need that much for it to run fast. One why to think of it is, the CPU gets the data, calculates the physics and then hands it off to the GPU to render so the game will only run according to the fasted element (CPU/GPU).

One common way of benchmarking is to set the graphics settings as LOW as they will go and the FPS result of the benchmark is said to be the CPU speed. Then turn the graphics setting as HIGH as they can go and run the same test again to get the GPU speed. Like this.

CPU Speed

640x480 - 16 bits per pixel - NoAA - NoAF

GPU Speed

Highest Graphics Resolution - Highest Pixel Setting

Like 3840x2160 - 32 bits per pixel - Full AA - FULL AF

But if you need to skimp anywhere do it with the CPU since most games these days (or sims) are heaviest on the GPU. But.. if you have a SLOW CPU you will still get bad performance so CPU next to Top - GPU Top.
 
What's with the PhysX setting in the nVidia control panel? It's on auto select, using the GTX760. Would changing that to CPU be likely to improve frame rates by putting more load on the CPU & less on the GPU? I find it hard to believe that the bottleneck would be the CPU - it's a Quad Xeon 3.6. And RF is installed on my SATA3 SSD. I guess I have some experiments to run.

With my set-up, would VR be a waste of cash? I ALMOST bought a Lenovo VR for $200.
 
Well, I tried to do a little testing as suggested. On "Highest" graphics settings I get about 125fps. On "Low" I get about 225. The physics & graphics frame rates displayed in RF are within a few frames of each other. Changing the nvidia control panel PhysX setting to CPU doesn't seem to change anything.

But I can't try different resolutions & settings because I can't get into full screen mode.
I get an error: DS3DERR_INVALIDCALL DXMesh.cpp line 1681.
I wonder if it has something to do with my dual monitors. I run RF on my primary monitor at 1920x1080. My secondary monitor is 1440x900
 
Bill Stuntz said:
Well, I tried to do a little testing as suggested. On "Highest" graphics settings I get about 125fps. On "Low" I get about 225. The physics & graphics frame rates displayed in RF are within a few frames of each other. Changing the nvidia control panel PhysX setting to CPU doesn't seem to change anything.

But I can't try different resolutions & settings because I can't get into full screen mode.
I get an error: DS3DERR_INVALIDCALL DXMesh.cpp line 1681.
I wonder if it has something to do with my dual monitors. I run RF on my primary monitor at 1920x1080. My secondary monitor is 1440x900
Click to expand...

I'd call support on that error. I run RF-8 on dual monitors, primary 1920x1200 and secondary 1920x1080 with no problem. So dual monitors are okay but the 1440x900 is an odd size so maybe something's going on with it. You might try unplugging your secondary and make sure Windows is set for a single monitor and try it then.

Was your testing on the Monitor? Remember VR has to render Two Separate views at the same time so that would certainly change your FPS.
 
Last edited:
Talked to tech support & they had me reset RF to defaults after they confirmed that my video drivers & DirectX were OK. I'm not sure exactly what changed. Looking at the settings, I don't see anything that jumps out as different except that it had changed the resolution from 1920 to 1280 & set quality to medium. But even after I reset it back to 1920x1080 & highest, the frame rates jumped from ~125 to ~175 at Joe's Garage HD. And ~350fps at Carl Henson Field. I hadn't checked frame rates at photo fields. Full screen mode works too.
 
You must log in or register to reply here.
Back
Top