All, please accept my apologies for not slowing down enough

 

I don't see any need for you to apologise. You have gone down a path of trying to justify potential problems and trust me when I say, it is so much more refreshing to read your theory, than those that come in here to purposely bash a product, just for the sake of it. Trust me, they stand out like a sore thumb.

Ditto. We didn't think you were a troll etc.. We all honestly just couldn't grasp what the problem was and tried explaining what you saw wasn't  that big of a big deal..   We all screw up, get ahead of ourselves, make assumptions we shouldn't have out of passion. It happens. 

Gents, 

 

Where does one find reliable information about the direction P3D is heading from a fundamental coding point of view?  Paul seems to making comments that might have come from LM and truly I'd like to have the best understanding of the logic behind the performance as it were.  There are comments from folks here but they are getting them elsewhere and I'd like to read those myself.  Perhaps I would have a better understanding of why some things behave as they do, for example the low frame rate, lowish GPU use combination I've seen not only topdown but in high-demand situations.  

 

 

I don't see any need for you to apologise.

 

Thanks, but the apology was heartfelt for not taking the time to look at some of the images Brian took time to post.  I don't think they say everything, but they do support some conclusions however soft they may be.

Before P3Dv2 was released getting any information from LM was not an easy task but they did release some information ( in the blog they did and some other interviews and some forum posts ) which gave me the most basic insight on what to expect from v2 when it was released and some direction they are planning to take. When you combine with other information they have posted since its release it does help with understanding the sim it self.

 

The main issue is that most of the information about how the sim works is fragmented but the dev team do provide some very technical answers and i wont pretend to understand them fully but they are informative

 

But in a few years time as gpu's become more powerful we should truly see the benefit, Im using a GTX760 while not the most powerful card it does perform well in p3dv2 and this has kept me very happy with the sim, if i still had my GTX470 i think i would not be using the sim until i upgraded but at some point we should see nvidia driver improvements for v2

But in a few years time as gpu's become more powerful we should truly see the benefit, Im using a GTX760 while not the most powerful card it does perform well in p3dv2 and this has kept me very happy with the sim

I've done some observing that makes me wonder, as V2 stands now, how much it can benefit from more GPU processing power over a certain minimum level. What's puzzling, and makes me believe SLI will not substantially help in high-demand areas which is where one might hope it would help the most, is that GPU utilization will not go that high in certain types of scenarios that are high demand enough to drive down raw frame rate. One would like to see GPU use go well up towards 100% all the while scaling frame rate up with it, but that is not the case I find. I've tried to make the distinction that a good part of what the strong GPU does is handle DirectX effects more than raw scenery generation as it were. I say this because I've noticed what really hits the GPU is certain types of scenery, for example ocean water maybe because of 3D animation, and certain terrain shadows and shadow quality. Put another way, I can be flying around LA area in the A36 and see a frame rate of say 36 but GPU use may only be at 58% utilization. Tick the terrain shadow checkbox & bump quality to ultra, and now the GPU is up to 99%. Frame rate doesn't change hardly at all in this situation in my machine which is predicted if a big part of the GPU's role is to broker DirectX effects. Now go to FB's KPHX. Frame rate might be down to 28, and once again GPU use may actually only be 45%. So there is clear evidence to me that the strong GPU, once it hits a minimum level of processing power (a relatively high one by today's standards), doesn't really add much to the low frame rate issue in this type of high-demand area. So adding another Titan in SLI for example, doesn't appear to be able to help in this type of situation if you follow the argument. As a side point, I can't really perceive a difference when shadows are at High or Ultra, so that added power to drive that effects doesn't add nearly as much as I would like to see for example by bumping frame rate up from 28 to 38 for example in the KPHX example. I have read that the GPU does play a role in autogen depiction which I'm guessing is why lower end GPUs do struggle more w/ V2.

It's really not even that complicated  :lol:

 

It is still very easy to become CPU limited, no more mystery to it than that.

 

All of this effort to draw conclusions from strangely formed hypotheticals when we already know most of this, and with some significant changes coming with 2.05/2.1, will likely be invalidated anyway.

 

Also, and I really am not being snarky with this, break that up into a paragraph or three next time I beg you  :lol:

It's really not even that complicated  :lol:

 

It is still very easy to become CPU limited, no more mystery to it than that.

 

All of this effort to draw conclusions from strangely formed hypotheticals when we already know most of this, and with some significant changes coming with 2.05/2.1, will likely be invalidate anyway.

 

Also, and I really am not being snarky with this, break that up into a paragraph or three next time I beg you  :lol:

 

Really?  What part?  I made the statement that SLI in a box w/ a strong GPU won't contribute much to what matters, which to me is performance in high demand areas.   While I do apologize for my not looking deeper into your former reply you are truly expert at being snarky, even when you don't try.  I'll make you a gentleman's deal:  please don't respond to my posts and I won't respond to yours, fair enough?

Really?  What part?  I made the statement that SLI in a box w/ a strong GPU won't contribute much to what matters, which to me is performance in high demand areas.   While I do apologize for my not looking deeper into your former reply you are truly expert at being snarky, even when you don't try.  I'll make you a gentleman's deal:  please don't respond to my posts and I won't respond to yours, fair enough?

 

I'm not all that bothered by you not reading my replies specifically...it's more that you ask questions, an answer is provided by myself or others that have replied to you, you skim the answer and then proceed to ask another question that was covered by the answer you couldn't be bothered to actually read.

 

To your request, sure, works for me. Not much point in having a discussion with someone that isn't really paying attention anyway. I'm sure all of your conclusions drawn from pure conjecture will invalidate the real testing the rest of us are doing anyway.

 

Oh wait, was I not supposed to respond to these questions either ?  :unknw:

Your 'test' was flawed because your base assumption is that there is a similarity in how the two sims render.  In short, there is not.  That breaks the entire premise of the test.  It is difficult to explain other than to say that how DX9 renders a sphere is not the same as how DX11 renders a sphere.

 

As time passes, I suspect that Prepar3D will continue to see performance increases (it already has a great deal, with a correctly optimized system)... however FSX will never, ever see performance increases.

 

I work professionally with ESP/Prepar3D v1.4 and Prepar3D v2 and see their potential and limitations both.

 

 

Your 'test' was flawed because your base assumption is that there is a similarity in how the two sims render.  In short, there is not.

 

Very clearly stated thank you Ed--it supports the conclusions I had drawn after reading some of other replies here.