Skip to content
View in the app

A better way to browse. Learn more.

The AVSIM Community

A full-screen app on your home screen with push notifications, badges and more.

To install this app on iOS and iPadOS
  1. Tap the Share icon in Safari
  2. Scroll the menu and tap Add to Home Screen.
  3. Tap Add in the top-right corner.
To install this app on Android
  1. Tap the 3-dot menu (⋮) in the top-right corner of the browser.
  2. Tap Add to Home screen or Install app.
  3. Confirm by tapping Install.

Autogen - CPU or GPU dependent?

Featured Replies

I recently purchased FTX Norway and in order to mimic the preview pictures I turned my autogen slider to maximum and I noticed loss of FPS. Usually I fly heavy planes like PMDG with autogen density of normal to avoid OOMs, but occasionally I fly smaller aircraft and then I can really crank up the heat. Apparently not, since its taking a bite of my FPS. I "only" have a GTX 760 so I wondered if I were to experience any increase in FPS if I purchased a better graphics card? Im planning on switching to P3D when PMDG release the NGX, but for now Im stuck with FSX.

38.jpg

Brynjar Mauseth 

  • Author

Already using DX10 with OCed 4770K to 4,4 ghz  B)

38.jpg

Brynjar Mauseth 

I have a very simple rule...

 

If I notice a performance impact in a specific scenery location, I dial back the settings until the performance is ok again.

 

Most of the time, I fly with 30 fps locked and steady, but there are places near detailed airports where my fps take a hit... Dial back lod, traffic, or autogen, whatever breaks the logjam..

Bert

You can create different graphics profiles in FSX: Airliner, GA, daytime, nighttime and so on. Save them and reload them according to the scenery and the aircraft you're flying. No need to move many of those sliders anymore.

What happened to AVSIM

An easy way to find out if your gcard is the bottleneck is to use a tool such as MSI afterburn (which I highly recommend) to see if it the autogen density setting is maxing out your gcard, i.e. its load is running at 100%.  That being said, I've found that high autogen setting are more taxing on the CPU than on the GPU, and that's true with DX10 as well.  My guess is that your 760 should handle the dense autogen without problem, provide you are not using dense cloud cover, excessive AA settings, ground structure shadows, and/or high end water effects.

CPU: AMD 9800X3D PBO MB +200 CO -25| Motherboard: MSI MAG X870e Tomahawk WiFi | GPU: MSI RTX 5090 Ventus 3X OC | RAM: G.Skill 2x32GB DDR5 6000 cas 30 | M.2 SSDs: Samsung 990 EVO Plus 2T, WD Black SN750  M.2 1T | Hard Drive: WD Black HDD 6T 7200 | Optical Drive: LG Bluray writer, internal | Cooling: Thermalright Phantom Spirit 120 EVO | Case: Fractal Design Focus G | PSU: NZXT C1200 1200W

Win 11 Pro 64|HP Reverb G2 revised VR HMD|Asus 25" IPS 2K 60Hz monitor|Saitek X52 Pro & Peddles|TIR 5 (now retired)

 

 


An easy way to find out if your gcard is the bottleneck is to use a tool such as MSI afterburn (which I highly recommend) to see if it the autogen density setting is maxing out your gcard, i.e. its load is running at 100%. 

 

How would I do that ? I have msi afterburner.

 

Thanks

Autogen requires a considerable amount of drawcalls which hits both the GPU and CPU hard. 1 texture rendered onto 1 object costs 1 drawcall, without autogen there are no autogen drawcalls hence the huge performance difference. If you imagine thousands of objects being rendered which sometimes even contain further multiple drawcalls for different elements, this very quickly starts to hurt. The PMDG T7 VC on it's own for example has around 1400(?) drawcalls so if you imagine the huge amount of objects that are rendered as autogen, performance quickly goes out of the window.

 

I have learnt through experiment that FSX is very dependent on a complex balance between the CPU and GPU where one can very easily bottleneck the other. Many are quick to say that FSX is CPU dependent but this is not strictly true. There is a hard limit where your CPU and therefore upgrading it is irrelevant due to the GPU quite simply not being fast enough to deal with the amount of 'passed on' requests. The GPU is what renders the scene and each GPU has a hard limit of what it can do on a frame by frame basis. Remember though, this doesn't change the fact that the CPU sends the data in the first place and at it's own independent rate which is where the balance between the two comes from.

 

If the CPU is slow it will send data to the GPU VRAM slowly meaning the GPU goes through periods of not being pushed. You would think this means more FPS but FPS is reduced because the CPU cannot send data to the GPU fast enough. You can tell when things are in balance by setting an FPS limit, if the FPS limit is being met, both the CPU and GPU are managing just fine, whether you are happy with what is on the screen however is another thing.

 

In my experience, if you are running an i5 CPU or less which is also not running beyond 4GHz, it is highly likely your CPU is bottlenecking the GPU. If you are running a higher-end CPU beyond 4GHz, it is probably the GPU. There is no real proven way of testing what is happening beyond judging what happens when you change things like sliders and graphics settings.

Lawrence Ashworth

  • Author

Autogen requires a considerable amount of drawcalls which hits both the GPU and CPU hard. 1 texture rendered onto 1 object costs 1 drawcall, without autogen there are no autogen drawcalls hence the huge performance difference. If you imagine thousands of objects being rendered which sometimes even contain further multiple drawcalls for different elements, this very quickly starts to hurt. The PMDG T7 VC on it's own for example has around 1400(?) drawcalls so if you imagine the huge amount of objects that are rendered as autogen, performance quickly goes out of the window.

 

I have learnt through experiment that FSX is very dependent on a complex balance between the CPU and GPU where one can very easily bottleneck the other. Many are quick to say that FSX is CPU dependent but this is not strictly true. There is a hard limit where your CPU and therefore upgrading it is irrelevant due to the GPU quite simply not being fast enough to deal with the amount of 'passed on' requests. The GPU is what renders the scene and each GPU has a hard limit of what it can do on a frame by frame basis. Remember though, this doesn't change the fact that the CPU sends the data in the first place and at it's own independent rate which is where the balance between the two comes from.

 

If the CPU is slow it will send data to the GPU VRAM slowly meaning the GPU goes through periods of not being pushed. You would think this means more FPS but FPS is reduced because the CPU cannot send data to the GPU fast enough. You can tell when things are in balance by setting an FPS limit, if the FPS limit is being met, both the CPU and GPU are managing just fine, whether you are happy with what is on the screen however is another thing.

 

In my experience, if you are running an i5 CPU or less which is also not running beyond 4GHz, it is highly likely your CPU is bottlenecking the GPU. If you are running a higher-end CPU beyond 4GHz, it is probably the GPU. There is no real proven way of testing what is happening beyond judging what happens when you change things like sliders and graphics settings.

 

Thanks for a comprehensive answer!

 

Im guessing that my i7 4770K OCed to 4,4 ghz should be capable of handling alot of information. Maybe my GTX 760 is being a little bit of a bottleneck, dont you think? Lockheed Martin has stated that due to tons of new rendering options and graphics settings in Prepar3D, a strong GPU is required to run with desirable eyecandy.

38.jpg

Brynjar Mauseth 

Thanks for a comprehensive answer!

 

Im guessing that my i7 4770K OCed to 4,4 ghz should be capable of handling alot of information. Maybe my GTX 760 is being a little bit of a bottleneck, dont you think? Lockheed Martin has stated that due to tons of new rendering options and graphics settings in Prepar3D, a strong GPU is required to run with desirable eyecandy.

 

Did you follow the manual that should have come with Norway (all other ORBX regions come with one) that has suggestions on how to set autogen?  Like MS and their philosophy of building versions of flightsim, ORBX also makes their scenery to the level that you can't necessarily run it at full settings with current hardware.  They basically state this in the performance tuning section of their manuals.  Based on my experiences with my 4790K running at 4.4Ghz in their NorCal region and a GTX780 I'd say you are still running into a CPU limit if you have cranked the autogen to max.

AMD Ryzen 9950X3D | Asrock X870E Taichi | Gigabyte Gaming OC 4090 w/EK waterblock | Full Custom Loop Cooling | Corsair Dominator Titanium DDR5-6000 | Samsung & WD NVME/SSDs | Phanteks Enthoo 719 | Seasonic Vertex Gold 1200W | Keychron Q5 Max | Corsiar Scimitar Elite SE Wireless | Honeycomb Alpha and Bravo | Logitech Pro Flight Pedals | VKB Gladiator Pro NXT L&R handed | MiniCockpit MiniFCU | Alienware AW34DWF | Asus PG279Q | Win 11 Pro

Thanks for a comprehensive answer!

 

Im guessing that my i7 4770K OCed to 4,4 ghz should be capable of handling alot of information. Maybe my GTX 760 is being a little bit of a bottleneck, dont you think? Lockheed Martin has stated that due to tons of new rendering options and graphics settings in Prepar3D, a strong GPU is required to run with desirable eyecandy.

 

More than capable, I don't really think that upgrading your CPU would yield any real tangible performance difference. This is especially true in Prepar3D where most of the new changes are graphically related.

Lawrence Ashworth

Create an account or sign in to comment

Account

Navigation

Search

Search

Configure browser push notifications

Chrome (Android)
  1. Tap the lock icon next to the address bar.
  2. Tap Permissions → Notifications.
  3. Adjust your preference.
Chrome (Desktop)
  1. Click the padlock icon in the address bar.
  2. Select Site settings.
  3. Find Notifications and adjust your preference.