I will say this, to run FSX without DX10 is doing yourself (and FSX for that matter) a disservice.

 

That is not the case for everyone.

It is upsetting to see some have issues with DX10 but then, DX10 was never finalized by Microsoft.  But I have found the DX10 Fixer has fixed most of the anomalies.  It has been proven that DX10 provides an extra 10% performance boost.  Why not?  It's better technology.  But, DX10 is no different than DX9 technology in FSX other than you'll see some cockpit shadows which provide for more immersion and you will see waves in the ocean, lakes and rivers that matches the weather.  So, if the weather is nasty, you'll see nasty animations in the waters.  You will not get this immersion in DX9.  Many think they need special configurations to run FSX.  I don't.  I just flip the DX10 Preview switch, restart FSX, and I'm in DX10 with graphics just as great and mostly better as DX9.  That's the way Microsoft intended DX10 to be used.  The problems occur when individuals try new and different configurations than they used in DX9 and/or they use products that were not developed originally for FSX.  I think that if you own a product like FSX you need to use the highest level of technology available to that product and DX10 is the highest level.  Is there any difference in the scenery or aircraft textures?  No.  But, with Steve's DX10 Fixer, shadowing has been enhanced and you should be getting better performance.  I'm hoping everyone will at least try DX10 especially if they do not plan to move upward and forward and purchase P3D (which uses DX11 technology).

 

 

I've heard that DX10 mode greatly helps prevent OOM's and increases performance all while upping the visuals (Beautiful Water with no FPS Hit!).

So now that DX10 rendering in FSX has been fixed by various methods (payware/freeware) does the inclusion of FSX in DX10 mode put FSX on a more level playing field with P3D?

Especially graphical and performance wise?

Thoughts?

 

DX10 definitely will not reduce OOM's.  That's a problem with virtual address space (VAS) and there's no pluses or minuses having DX10 to stop OOM's.  That means it will get no better in DX10 than it did in DX9 in regards to OOM's.  Increasing the sliders to get better visuals is a major mistake because you are not going to get them.  As I stated above, you should keep the exact same settings as you have for DX9 like I have been doing for the past two years.  It's not that great of a technology that it will allow you to change your DX9 settings.

 

Of course, DX10 is not as good as DX11 so there is no way it can be compared to P3D.  For instance, in DX10, the VC shadows on some aircraft can hardly be seen.  In DX11, the VC shadows show bright sunlight shining on the panels and make it even more realistic.  The shadowing is more pronounced. 

 

Best regards,

To be honest, the superior VAS management would be enough reason to use the DX10 Fixer alone.

 

 

 

I have just finished a flight across the Alps using the F1 Mustang, FSDreateams Geneva and Aerosofts Venice scenery, tons of traffic from MTX 5.4b, ASN, Radar Contact ect.

Finished the Flight with 1650Mb VAS left. this using DX9

What Are you guys doing that depletes your VAS to the point it becomes a problem ?

@up it is easy - spawning NGX on drzewiecki EPKK, with autogen to very dense and rest maxed out except for water. Than weather loads from Rex 4 with 32 bit clouds, Lod 5.5, traffic from MT2013 set to 30%, accufeel on, ftx global, vector and fsglobal mesh loaded, GSX start to serve my aircraft, fs2crew copilot helps me set things up and shade is giving some nice overall look. Bloom increases immersion (limited to 60%). I have no road traffic, no shadows but ASN has enhanced overcast on as I like to use this tool to FSX limits and for the sake of best image I chose this function.

 

At that point I am already 2.9 GB of VAS. With dx9 I would be higher and no chance to finish my flight.

 

Jets are harsh on VAS. NGX itself is using 800 mb so go figure. Vector and mesh add more, traffic was my main problem because mt2013 at 30% is more than default FSX on 100.

 

Anyway - my bank account will tell you best what did I do to hit vas limit

DX10 definitely will not reduce OOM's. That's a problem with virtual address space (VAS) and there's no pluses or minuses having DX10 to stop OOM's. That means it will get no better in DX10 than it did in DX9 in regards to OOM's.

 

But(!) VAS and/or memory is/are handled differently in DX10 than under DX9 -and THIS ist the point!

Many users - including myself - have reported that they did not run into an OOM in DX10 whereas it happend to them in DX9.

I do not have sufficient technical background to explain why DX10 "handles" VAS differently than DX9 does, but there are several infos about it on the web. Just mentioning two here as an example:

 

Post #2, here taken from around the corner at the AVSIM-forums:

http://forum.avsim.net/topic/433879-thinking-of-doing-a-second-attempt-switching-from-dx9-dx10/

 

... or - as outlined by a user in #72 at this (very long, but most informative) thread here at the ORBX forums:

 

http://www.orbxsystems.com/forum/topic/65134-new-dx10-fixer-feedback/page-4?hl=%20moonman%20%20dx10%20%20vas

 

[...] The reason for the less OOM is apparently DX9 requires mirroring the video memory in FSX's virtual address space which is limited to 4GB and fills up fast with high powered scenery and aircraft. DX10 does not require that mirroring and I think I read may be more efficient overall in its memory use in the sim.  I'm not sure about that part I just seem to remember reading it.  The video memory mirroring is for sure accurate [..]

 

And from my personal (maybe less technical but pragmatical ^_^) perspective:

I did several testruns over Aerosoft's PANC installed with ORBX's Southern Alaska region altogether in PMDG's great 777 for instance.

I did not use real world wheater, just set visibility to 30nm and kept all the settings (including fuel, route, gate and parking position, time and date and all that) the same way in DX9 - as DX10 some time later on.

The flight was scheduled to take me from PANC to FSDT's CYVR - now guess what:

I ran into an OOM some time prior to the scheduled T/D everytime in DX9.

I just could not finish any flight.

But(!):

That never (i guarantee that) happend to me in DX10 - not even with real world weather provided by OPUS in that case then and REX cloud textures enabled.

And furthermore, since switching to DX10 i never encountered an OOM anymore - at all kind of various scenerious my FSX never ran into one since i left DX9 behind.

 

Now as mentioned several times (really sorry for this repeating pattern, but just in case):

DX10 also is not a 100% guarantee to avoid OOM's all the time due to FSX's 32 bit architecture - that's true - but fact also is - and that's the point:

DX10 lowers the chances of running into an OOM really, really noticeably due to its different handling of VAS and that's a proofed and valid point i'd dare saying.

So to sum it up:

To significantly lower the chances of running into an OOM within FSX, DX10 is certainly worth to be a given a serious try!

No more, but also no less.

:wink:

[...] The reason for the less OOM is apparently DX9 requires mirroring the video memory in FSX's virtual address space which is limited to 4GB and fills up fast with high powered scenery and aircraft. DX10 does not require that mirroring and I think I read may be more efficient overall in its memory use in the sim. I'm not sure about that part I just seem to remember reading it. The video memory mirroring is for sure accurate [..]

 

Thats not exactly true, since WDDM 1.1 and vista

 

http://support.microsoft.com/kb/940105

 

However the Gist is along the right lines and as we dont know what FSX is doing with its memory (i.e. marking the buffers as lockable or not you will get a benefit from using dx10

 

It shouldnt be gigabytes though on a modern 64bit operating system

I have both FSX and P3D and in my opinion DX10 FSX far outshines P3D at this moment in time.

 

 

 you will get a benefit from using dx10

 

It shouldnt be gigabytes though on a modern 64bit operating system

 

In my experience, it is an improvement of about 10% over Dx9 which, added to the fact that in Dx10 FSX will continue to function right up to the 4095th MB of VAS, is usually enough to keep it going when in Dx9 it would have stopped.

In my experience, it is an improvement of about 10% over Dx9 which, added to the fact that in Dx10 FSX will continue to function right up to the 4095th MB of VAS, is usually enough to keep it going when in Dx9 it would have stopped.

 

Yep 10% would make sense, which means that 10% of memory being allocated is marked as lockable and being virtualized,

 

 

DX10 definitely will not reduce OOM's

 

That is simply untrue.

 

I have conducted tests in several scenarios rebooting the computer after each test.  In every case, the same flight in DX10 used less VAS than in DX9.

 

I received exactly the same outcome on both of my computers, and this is the main reason that I use DX10 exclusively on both of them now.

 

Dave.

Thats not exactly true, since WDDM 1.1 and vista

 

http://support.microsoft.com/kb/940105

 

However the Gist is along the right lines and as we dont know what FSX is doing with its memory (i.e. marking the buffers as lockable or not you will get a benefit from using dx10

 

It shouldnt be gigabytes though on a modern 64bit operating system

 

First of all: Thank You very much for sharing the link above.

Really appreciated.

As pointed out though - i am no PC technical expert at all and hence i mentioned the links in my previous post.

And as i am aware of the fact that various factors may come into play i also added at least two resources and not just one.

But ... isn't the aspect of video memory "mirroring" or "copying of resources" also outlined in the link kindly priovided by You?

I apologize in advance if i should mess things up here as i just took a quick look at the article - so maybe it could be clarified again by a person more technically advanced here.

 

Nevertheless - and for whatever specific reason, which i sadly can't explain in a detailed way:

DX10 and VAS "interact", "work" or however i should say differently than this is the case under DX9 and the chance of crashing into the 32bit given 4GB "wall" in FSX is lowered noticeably - hence lowering the chance of running into an OOM.

Now as i can't really contribute anything new or anything else of which i consider to be useful to this interesting thread other than i did have already mentioned - i'd say ... be it from my side for now.

 

In my experience, it is an improvement of about 10% over Dx9 which, added to the fact that in Dx10 FSX will continue to function right up to the 4095th MB of VAS, is usually enough to keep it going when in Dx9 it would have stopped.

 Sounds interesting.

I run both P3Dv2 and FSXDX10 and feel that yes, DX10 brings FSX closer to P3Dv2. But it's not quite there and never will be because P3Dv2 provides further enhanced features as described in the P3D forum. I will say this, to run FSX without DX10 is doing yourself (and FSX for that matter) a disservice.

 

1. Performance wise, if used correctly, I get better performance with P3D.

 

2. Graphically? Again, if used correctly, I get better graphics and shadowing with P3D, more autogen, volumetric clouds that I fly through, minimal and sometimes zero autogen popping.

 

But regarding those two points above, I could also easily state that performance and graphics can be worse (and here is the mistake many make) simply by pushing the P3D sliders too far to the right.

 

So in summary, if you are considering purchasing the DX10 addon, I would say it is the best thing you can do to FSX to give a little more life in the old girl, but she can only travel so far on her old worn out tyres.

 

PS You will end up changing the wheels - eventually. 

 

Same for me, and summed up rather nicely  :good:

First of all: Thank You very much for sharing the link above.

Really appreciated.

As pointed out though - i am no PC technical expert at all and hence i mentioned the links in my previous post.

And as i am aware of the fact that various factors may come into play i also added at least two resources and not just one.

But ... isn't the aspect of video memory "mirroring" or "copying of resources" also outlined in the link kindly priovided by You?

I apologize in advance if i should mess things up here as i just took a quick look at the article - so maybe it could be clarified again by a person more technically advanced here.

 

Nevertheless - and for whatever specific reason, which i sadly can't explain in a detailed way:

DX10 and VAS "interact", "work" or however i should say differently than this is the case under DX9 and the chance of crashing into the 32bit given 4GB "wall" in FSX is lowered noticeably - hence lowering the chance of running into an OOM.

Now as i can't really contribute anything new or anything else of which i consider to be useful to this interesting thread other than i did have already mentioned - i'd say ... be it from my side for now.

 

 Sounds interesting.

 

Hi, Firstly you are correct that DX10 manages memory in a totally different way than DX9 and that DX10 reduces the VAS usage due to DX9 requiring a "Virtualized" copy of video memory.

 

Quoting the document directly

 

 

To address this problem, Microsoft is changing the way that the video memory manager maintains the content of video memory resources. This change is being made so that a permanent virtual address range does not have to be used for each virtualized allocation. With the new approach, only allocations that are created as "lockable" consume space in the virtual address space of the application. Allocations that are not created as "lockable" do not consume space. This approach significantly reduces the virtual address space that is used. Therefore, the application can run on large video memory configurations without reaching the limits.

 

So memory allocations that are not marked as lockable are not copied. My point being that a 1:1 copy of video memory is unlikely to be maintained in the VAS unless all the allocated memory is marked as lockable.. This can be checked I with Intel GPA but its not something I have done as yet.

[...]

So memory allocations that are not marked as lockable are not copied. My point being that a 1:1 copy of video memory is unlikely to be maintained in the VAS unless all the allocated memory is marked as lockable.. This can be checked I with Intel GPA but its not something I have done as yet.

 

Okay, now i think i am getting to understaning it better!

Thanks a lot for the info again!

 

The main problem is what do MS mean by "lockable" as I don't think you can mark Vertex Buffers as "non Lockable" in DX9! what we need is two identical installs of fsx, one in XP664bit and one in windows764bit and then measure the VAS usage in both to be sure that the hotfix is doing what it said it is!