DX10 & Bufferpools

[RJC68 823]

Couldn't wait until tomorrow lol

 

Have a nice dust free rig, just uninstalled the Nvidia drivers. Time to take the dogs for a walk then break open that little beauty sitting on the dining room table

[RJC68 823]

Woo Hoo

[Jetblst 23]

 

 

TextureMaxLoad=30   This one has a similar purpose to TBM, i.e. it's a multiplier used by FSX in the algorithym used to calculate how much data it can, or should send to the GPU. Bojote first brought it up a few years back, as a tweak purported to be better than using TBM: Michael Moe more recently was playing around with it, and so did I in some measure, but nothing conclusive came out of my sessions so I've removed it again. I believe the normal setting is 6.

 

No Kidding!  Well that is this is all about.  

 

If you had no real measure good or bad I think I'll leave it alone.  Thanks Paul!

[Ron Attwood 3,499]

 

 

Thanks Paul!

Paul has left the building. :(

[Lucass 3]

Poul, Jet

at first, I would like to thank you for replies and help, but... before start testing I have to clarify some things...

 

@Poul

1. what does "s/b" mean in your proposed fsx.cfg?

2. what is ideology of testing with default plane (as you wrote C172) if I mainly fly "heavy"  PMDG planes?

3. Does VSync 1/2 RefreshRate in Inspector is necessary? I don't fly in full screen mode. Besides, if I'm not wrong, VSync decreases frames...am I right?

4. Is it necessary to limit frames in FSX? When I do it in Inspector, I have few more frames (there is smoother)

5. TextureMaxLoad=30 is a relic of fsx.cfg DX9 optimization ^_^

6. You wrote "slow proc"?     – when I bought it, it was one of the most recommended CPU for FSX  in approachable price  ^_^  (also GTX560) – I'm very disappointed see these words :mellow:

(by the way - what do you recommend (CPU and GPU) in approachable price, provided DX10 flights?)

7 ...and Poul - could you drop again CFG and Inspector setting on "dropbox" again? (actually there is no possibility to view them  :(  )

 

@Jet

1. what BMB is?

2. Please confirm:

- RT is RejectThreshold?

- BP is BufferPools? If YES, I have problem with understanding it... because I thought that this is name of section ([bufferPools]) but not parameter name. But you tell me that I have to change them (BP=X). I don't understand. I know three parametres in [bufferPools] section:

Usepools=
PoolSize=

RejectThreshold=

 

 

3. Let me describe how I understand testing procedure:

 

set TBM=80, delete TextureMaxLoad=30

 

now, section [bufferPools]:

a. set UsePools=1

b. don't set PoolSize

c. BufferPools=1

d. RT=128000

...testing. If stutters and low fps are visible...

e. set RT=262144

and so on

When I reach RT=20971520 and stutters and low fps are still visible I should rise BufferPools=2 and back RT=128000 ..and testing. If no results, rise RT=262144...

and so on.

Right?

So when should I set PoolSize?

 

Best regards,

Lucas

[Jetblst 23]

Poul, Jet

at first, I would like to thank you for replies and help, but... before start testing I have to clarify some things...

 

@Poul

1. what does "s/b" mean in your proposed fsx.cfg?

2. what is ideology of testing with default plane (as you wrote C172) if I mainly fly "heavy"  PMDG planes?

3. Does VSync 1/2 RefreshRate in Inspector is necessary? I don't fly in full screen mode. Besides, if I'm not wrong, VSync decreases frames...am I right?

4. Is it necessary to limit frames in FSX? When I do it in Inspector, I have few more frames (there is smoother)

5. TextureMaxLoad=30 is a relic of fsx.cfg DX9 optimization ^_^

6. You wrote "slow proc"?     – when I bought it, it was one of the most recommended CPU for FSX  in approachable price  ^_^  (also GTX560) – I'm very disappointed see these words :mellow:

(by the way - what do you recommend (CPU and GPU) in approachable price, provided DX10 flights?)

7 ...and Poul - could you drop again CFG and Inspector setting on "dropbox" again? (actually there is no possibility to view them  :(  )

 

@Jet

1. what BMB is?

2. Please confirm:

- RT is RejectThreshold?

- BP is BufferPools? If YES, I have problem with understanding it... because I thought that this is name of section ([bufferPools]) but not parameter name. But you tell me that I have to change them (BP=X). I don't understand. I know three parametres in [bufferPools] section:

Usepools=

PoolSize=

RejectThreshold=

 

 

3. Let me describe how I understand testing procedure:

 

set TBM=80, delete TextureMaxLoad=30

 

now, section [bufferPools]:

a. set UsePools=1

b. don't set PoolSize

c. BufferPools=1

d. RT=128000

...testing. If stutters and low fps are visible...

e. set RT=262144

and so on

When I reach RT=20971520 and stutters and low fps are still visible I should rise BufferPools=2 and back RT=128000 ..and testing. If no results, rise RT=262144...

and so on.

Right?

So when should I set PoolSize?

 

Best regards,

Lucas

 

Very Close!

 

Bufferpools = 1 is how we say yes to the compiler.  0 = no.

 

When Paul was telling us that we didn't need to define the pool was because the particular pool size in question is the DEFAULT pool size so no need to define.  IF by chance you get to moving up to the next pool size (After testing the RTs as you now understand) then you would need to define that BufferPoolSize with the line Poolsize =X, using one of the Bufferpool sizes Paul posted about.

 

What I found that is a little faster way to getting down to the proper BP/RT ratio is to simply test the various BP's with the medium RT values.  That way you can get into the right "ballpark" if you will quicker.  Then if you feel you are near what you need for your system you know you have the right bufferpool defined it now comes down to the RT.  Again this worked for me.

 

So yes, you have the testing procedure down very well!

[Jetblst 23]

Also to clear up the format question.  For now just ignore Paul's comment on not needing to define the bufferpool.

 

You'd do well to always have your Bufferpool section to look like this.

 

[bUFFERPOOLS]

UsePools=1

PoolSize=838608

RejectThreshold=524288

 

First we define that we are using BufferPools

 

[BUFFERPOOLS]

UsePools=1

 

Note: UsePools=0 is the same as not using Bufferpools at all more or less allowing your GPU to do the work.

 

Second we define the poolsize.

 

PoolSize=X, where X is the size of the Bufferpool.  Recommended sizes given in the above posts.

 

Third we tell the our system to REJECT at a certain point.

 

RejectThreshold=Y.  Where Y is the reject point where the CPU decides to send the rest of the processing to the GPU.

 

 

Hope that helps.

[Lucass 3]

Ok, so:

[bufferPools]

UsePools=1

RT=128000

...testing. If stutters and low fps are visible...

RT=262144 ..testing...

and so on

 

When I reach RT=20971520 and stutters and low fps are still visible I should add:

PoolSize=10MB (as bytes of course)

and RT back to 128000 ..and testing. If no results, rise RT=262144...

and so on.

Right? :rolleyes:

 

But I don't need to enter PoolSize=8MB (as bytes) because this is default value, yes? (by the way in example above there is mistake in PoolSize=838608, it should be 8388608 - am I right?

[Jetblst 23]

On Testing Theory.

 

Speaking for myself here.  But what I am finding is the best way to test ANY changes to the system, configs, etc, that are done in efforts to increase performance is to always do this.

 

Max Load Aircraft

 

For example PMDG aircraft.  Anything with a lot of glass and complex Systems.  If you test these changes out on a stock 172 and get things working like you want them to with a stock 172, you are going to be in for a real shock when you jump into your favorite megaplane as the demands on the system are going to be HIGHER.  So you want the HIGHEST demanding aircraft for testing.  Think about it.  Everything lower than that is going to give better performance which is why we are all here in the first place.

 

Max Load Scenery

 

Max Load Weather

 

Again load up your system once you have your FSX.CFG configured the way you want to BEGIN testing.  REMEMBER the less tweaks the better. ESPECIALLY if you do not really know what you are changing, why you are changing it, and what the outcome will be with that change.  If you make a change, only make ONE change at a time and then test it.  Do not become a tweakoholic like I was.  (lol).  

 

Once you have your FSX.CFG configured to test, THEN go for an initial test flight or flights as Paul described above.

 

Once you have the best performance you can get with the above conditions, then TUNE FSX via the IN SIM or IQ sliders.

 

Once you have that down, leave it alone and go fly.  Your are probably at the best solution for your given system.

 

If done correctly every airplane from then on should be smooth to even smoother.

 

KEEP THIS IN MIND

 

Simply stated this whole process is about a point to start from, then a point to which you personally are happy with your given system.  The reason people get frustrated is because what works for some does not work for others.  Hence Nick N's term...  

 

YOUR MILEAGE MAY VARY!

 

Have Fun!  

Ok, so:

[bufferPools]

UsePools=1

RT=128000

...testing. If stutters and low fps are visible...

RT=262144 ..testing...

and so on

 

When I reach RT=20971520 and stutters and low fps are still visible I should add:

PoolSize=10MB (as bytes of course)

and RT back to 128000 ..and testing. If no results, rise RT=262144...

and so on.

Right? :rolleyes:

 

But I don't need to enter PoolSize=8MB (as bytes) because this is default value, yes? (by the way in example above there is mistake in PoolSize=838608, it should be 8388608 - am I right?

 

YEP!

 

Complicated but the process was well worth it for me.

[Lucass 3]

For example PMDG aircraft. Anything with a lot of glass and complex Systems. If you test these changes out on a stock 172 and get things working like you want them to with a stock 172, you are going to be in for a real shock when you jump into your favorite megaplane as the demands on the system are going to be HIGHER. So you want the HIGHEST demanding aircraft for testing. Think about it. Everything lower than that is going to give better performance which is why we are all here in the first place.

That’s why I was a little surprised when Poul suggested tests with Stock 172. I also think that tests with PMDG are better solution.

 

Once you have the best performance you can get with the above conditions, then TUNE FSX via the IN SIM or IQ sliders.

 

What do you mean: „IQ sliders”?

 

 

REMEMBER the less tweaks the better.

Because my present FSX.CFG file was modified by venetubo and later changed according to Poul’s How-to.doc , I decided to generated fresh CFG file and made modification as Poul suggested… and of course I will start BufferPools testing. That’s my planned way…

 

* * *

 

Jet, Poul… one more question. I also have one issue – slow ground textures sharpening in external view (default textures + UTX). When I move camera to view another area (example from left to right side of aircraft), I have to wait 7-8 (or sometimes more) sec. to textures become sharpness. What can I do with this?

[Jetblst 23]

 

 

What do you mean: „IQ sliders”?

 

The main FSX Sliders.  Once you are into FSX.  

 

 

That’s why I was a little surprised when Poul suggested tests with Stock 172. I also think that tests with PMDG are better solution.

 

Keep in mind.  This is how  I test.  Not the only way or the right way, it just the way I like to test as I have found that when I do this every other aircraft in my hanger runs as good or better and then I do not have to set things up for every aircraft.

 

Just so you know, Paul has moved on from AVSIM, so he will not be answering your posts.

 

The other questions you have there I wont be able to help as I don't use that software.  Hopefully someone else. 

[Lucass 3]

 

Just so you know, Paul has moved on from AVSIM, so he will not be answering your posts.

 

It's very sad.

 

Jet, are you able to answer for my 2 questions to Poul:

 

1. Does VSync 1/2 RefreshRate in Inspector is necessary? I don't fly in full screen mode. Besides, if I'm not wrong, VSync decreases frames...am I right?

 

2. Is it necessary to limit frames in FSX? When I do it in Inspector, I have few more frames (there is smoother).

 

I'll be thankful.

 

And what about with this:

slow ground textures sharpening in external view (default textures + UTX). When I move camera to view another area (example from left to right side of aircraft), I have to wait 7-8 (or sometimes more) sec. to textures become sharpness. What can I do with this?

 

How it looks in your sim?

 

 

 

Regards,

Lucas

[MrRoper 27]

It's very sad.

 

Jet, are you able to answer for my 2 questions to Poul:

 

1. Does VSync 1/2 RefreshRate in Inspector is necessary? I don't fly in full screen mode. Besides, if I'm not wrong, VSync decreases frames...am I right?

 

2. Is it necessary to limit frames in FSX? When I do it in Inspector, I have few more frames (there is smoother).

 

I'll be thankful.

 

And what about with this:

 

How it looks in your sim?

 

 

 

Regards,

Lucas

 

 

Hi Lucas,

 

I will attempt to help a little,

 

1)

If you run in windowed mode then you can leave V-sync at "application controlled" The nvidia Control Panel V-sync settings only work in full screen mode.

 

2)

Bear in mind I don't run DX10 currently (to many issues with it to use it full time) however you have to realise that when running the FSX limiter at unlimited you 'may' see blurry textures in certain situations. using the FSX limiter dedicates more time to the rendering engine hence the loss in FPS.

 

You can gain these back by lowering your FIBER_FRAME_TIME_FRACTION  setting. You need to use this setting to balance the loss of FPS vs Blurring of ground textures.

 

The benefit of using this and the FSX limiter is that FSX will allocate processor resources to texture rendering when your frame rate exceeds the limit you have set.

[Pe11e 812]

After I upgraded my PC, from C2Q @ 3.5ghz to i7 3770k @ 4.5ghz, and complete and clean installs of Win7 and FSX, I realised that BUFFERPOOLS section is completely not mandatory at the moment. Completely smooth performance with the NGX and pay ware airports. So now I wonder what the BUFFERPOOLS section is all about, and I used it a lot.  :lol:

[Lucass 3]

@Chris

Ad 2. I was not preciselly, sorry. I know that limit fps is necessary, but which way is better: in fsx options or via Inspector? Many people says: via fsx options.. but in my opinion via Inspector is smoother effect.