Any improvements to low CPU/GPU usage planned?

[odai 4]

Hello

I have been messing around with FSW and comparing it to my experience in P3D v3, and one of the most noticeable differences is that the GPU usage tends to be a lot lower in FSW when compared to P3D in similar situations. Performance overall seems worse than P3D. 

When looking at CPU usage, it seems there is still the issue of very low utilisation on most threads on a i7 6700K, with just the one thread pegged pretty much at 100%. In terms of both CPU and GPU usage, there doesn't appear to be any improvement from vanilla FSX. P3D on the other hand seems to load the GPU considerably more, and from what I can tell from very early reports, P3D v4 also seems to improve significantly on the issue CPU usage with far more even utilisation of multiple CPU threads.

Does DTG plan on making improvements at some point to both CPU and GPU usage?

Odai. 

 

 

[SamYeager 244]

DTG have said, on Steam, Twitch and probably other places, that their initial priorities are optimisation and bug fixing.

[Alphafoxsierra 166]

On 2017-5-30 at 5:27 PM, odai said:

Does DTG plan on making improvements at some point to both CPU and GPU usage?

We sure are. Optimization is a ongoing part of our Early Access process. We have already made slight improvements as part of last week's update!

[fsxkitty 62]

I think some more multithreading in the sim will do miracles for performance. As the OP said, there's currently one thread that's always working at 100% while the others aren't doing all that much.

Also, I really hope that the load from rendering AI traffic can be moved to the GPU from the CPU. There was an addon for FSX that did this (UT2, I think?), and it performed spectacularly well.

[odai 4]

That very issue is actually the exact reason I ended up giving up on FSX and then P3D when I first tried it a year ago. I was using MyTraffic X and found that GPU loads were good and I was getting great visuals with excellent performance on the base installation (bar the OOM issues) until I started adding AI traffic.

It was at that point the CPU became a bottleneck with one thread being stuck at 100%. Prior to using AI traffic, performance was very smooth and it was actually my GTX 1070 that was the bottleneck. 

 

[Jiri Kocman 20]

I do not see much more difference in usage of CPU between FSW and P3D, sure latest v4 with texture size exp=10 makes all thread workers do more work than without that option, but it do not increase perfromance, even decrease it slightly as that usage is only becouse it reads much more data from disk, Without that option, with mostly default scenery I have similar CPU usage on both.

[Colonel X 1,428]

On 30.5.2017 at 6:27 PM, odai said:

In terms of both CPU and GPU usage, there doesn't appear to be any improvement from vanilla FSX.

These observations are so telling. DTG's strategy - throw some new paint on 10 year old software and sell it at a premium. It's horrible and it's our duty as flight simmers to avoid this product at all cost. 

[Chock 17,769]

6 minutes ago, frontendrob said:

These observations are so telling. DTG's strategy - throw some new paint on 10 year old software and sell it at a premium. It's horrible and it's our duty as flight simmers to avoid this product at all cost. 

How does giving it away free to people who bought Flight School, or selling it on Steam for less than 20 quid qualify as 'sell it at a premium'? That's less than the cost of a mediocre add-on for FS and in any case, most people got it for 'free' from having bought Flight School in a Steam sale for ten quid.

More to the point, if FSW ends up not being well optimised and doesn't run well, people certainly won't be inclined to slow it down even more by purchasing add-ons for it, they'll go and get XPlane or AeroFly or Prepar3d, so it is critical that DTG optimise it because they intend to make money from it by selling add-ons for it for literally years to come. It already cranks up faster than FSX right now, and they haven't even released the optimisation update for it yet.

Any observations of its performance thus far, are observations of a non-optimised as yet alpha version of the software, so one can hardly come to any concrete conclusions about it at this point, and I would contend that if flight simmers had a 'duty' at all, it would be to encourage developers to produce flight sims.

[Colonel X 1,428]

I think last time I checked, FSX was 4.99$ in the budget corner. FSW is 4 times that to parttake as beta tester. Of course it's a drop in the bucket considering the overall cost of this add-on infused hobby. But my point remains. They're trying to sell a new, full price (after early access) program, which is really just FSX with new menus and a great raindrops-on-windhsield effect. Why weren't the assets (textures, autogen, vectordata ect.) updated? Why are they using a decade old graphics engine (well documemted by now). LOL do you really believe in "optimizations down the line"? How much performance did LM squeeze out of the FSX code in 5 years time? Did they at all? How many "I get 12 FPS at LAX" posts does it need for people to realize that the FSX code was and is hopelessly slow. Now here's DTG trying to make the most profit from their investment (that's the whole point - why not write their own sim?). And of course their hope lies in naive, unimformed simmers. Noone else would buy a 10 year old sim in disguise. 

[vortex681 806]

23 minutes ago, frontendrob said:

And of course their hope lies in naive, unimformed simmers. Noone else would buy a 10 year old sim in disguise.

Seems to have worked well enough for LM with P3D! I'd hardly call their customers naive or uninformed. My personal experience is also that XP11 doesn't seem to perform much better than FSX either.

[Colonel X 1,428]

I was thinking the same when I wrote those lines. But there are differences. When P3D came out, it represented the logical "upgrade path", especially considering the add-ons. Now it's 2017, and noone waited for DTG to release yet another FSX version, there's no demand for it nor a USP.

[SmokeDiddy 262]

27 minutes ago, frontendrob said:

...naive, unimformed simmers. Noone else would buy a 10 year old sim in disguise...

I resemble that remark!  Not only am I naive and sim ingnorant, I'm old to boot.  hahaha...Seriously though, I'm not that naive nor ignorant, but if you look over the sim market, you'll find what is perceived to be better sim engines and performance, but they lack in so may other areas, so if you're looking for a perfect sim, you'll still be looking years from now.  I know you're not asking for grandfatherly advice, but I'll give you some anyway, please throttle your expectations.  Creating a new sim from scratch will take several years and for people like me, I don't have the time to wait.  So, I'll give FSW a chance and P3D I believe is the most "complete" sim offered at the moment - given "new gen" 64bit platform.  My 2.6875 cents

[Chock 17,769]

1 hour ago, frontendrob said:

LOL do you really believe in "optimizations down the line"?

Nope, I'm not assuming anything with any certaintly, I'm waiting for DTG to come up with the goods. The ball is in their court, and whilst it is, it costs nothing (literally) for me to await the outcome of that. On the contrary, it will however cost DTG a good deal of time, money, effort and indeed credibility if they don't manage to do that.

Now, I'm presuming since they are aware of this - particularly since they are not selling FSW for much, and in many cases not charging anything for it at all - that their accounts department, their product planners and business strategy people will have looked very carefully at the possibility of whether that investment is worth it. Since they appear to be going ahead with it, one presumes they have determined they can pull it off, because amongst the things they will also have examined, is the fact that nobody is going to buy add-ons for FSW if it offers nothing new over FSX-SE, and since they already own that, it would seem a strange thing to do, to waste time developing it further if they did not believe it could be done, especially in creating a platform which itself, is earning them little to no revenue, but is intended to be a means to create additional revenue through DLC.

But short of being sat at a desk in DTG, or Lockheed Martin for that matter, neither you nor I know what is in store, however two things I do know are: One -  Nothing is impossible if one has the determination to do it, and: Two, All the people at DTG and Lockheed Martin know a crapload more about computer programming and optimisation than us lot. Because I strongly suspect a company which managed to make an aeroplane which flew at Mach 3 at an altitude 80,000 feet well over fifty years ago, can pull off getting a PC to move a few pixels around the screen a bit faster if they put their minds to it. And I'm fairly certain they knew they'd be able to achieve that when they bought ESP, as I suspect, did DTG too.

[Colonel X 1,428]

Deleted

[Colonel X 1,428]

25 minutes ago, SmokeDiddy said:

I resemble that remark!  Not only am I naive and sim ingnorant, I'm old to boot.  hahaha...Seriously though, I'm not that naive nor ignorant, but if you look over the sim market, you'll find what is perceived to be better sim engines and performance, but they lack in so may other areas, so if you're looking for a perfect sim, you'll still be looking years from now.  I know you're not asking for grandfatherly advice, but I'll give you some anyway, please throttle your expectations.  Creating a new sim from scratch will take several years and for people like me, I don't have the time to wait.  So, I'll give FSW a chance and P3D I believe is the most "complete" sim offered at the moment - given "new gen" 64bit platform.  My 2.6875 cents

I didn't want to throw in age into the equation. That's a very sensitive topic here on AVSIM, just like politics (that's even prohibited I believe), or P3D criticism (also prohibited, I just came off a 4 day ban for violating that). I am 36, and let me tell you I am always in for some grandfatherly advice, so keep em coming! I think the perfect sim will always be 10 years in the future, so don't worry, everyone will still be complaining long after you've checked out.