Posted this on the Lockheed-Martin P3D forum, didnpt get a response.

Does anyone know if V5 addresses the core 0 bottleneck? (That is, that core 0 always runs at 100%, and hence the single thread performance of the processor has a big impact on overall performance.)

Lack of a response om the LM forum suggests the answer is no, which would be bad news. 

Edited April 9, 20206 yr by silentsage

Not everything can be threaded out to different cores nor would it always benefit to do so. It's not really a core 0 bottleneck... 

41 minutes ago, silentsage said:

the single thread performance of the processor has a big impact on overall performance

I can't remember the details exactly but somebody from the software gurus on these boards had explained why all flight simulators would always be very much dependent on a single thread performance, it's like a law of the universe. 

Edited April 9, 20206 yr by Dirk98

Not all simulators are bound by a core 0 bottleneck. 

It is not a bottleneck... what threads don't run on core 0 run on other cores.  It's not logical to expect all cores to be evenly loaded, after all they are clocked and doing something even if not processing your thread.

I have seen all 6 cores of my 8700k being used in almost perfect balance on flight simulators before. Never P3D though. 

LM never wont give you an answer to core0 bug.

The problem with core0 has been in the graphics engine since 10/2006, when FSX was released. And wasn´t fixed with SP1 and SP2.

• When LM released P3Dv1 in 10/2010 they could have re-build the FSX graphics engine. They didn´t: Problem with core0
• When LM released P3Dv2 in 10/2013 they could have re-build the FSX graphics engine. They didn´t. The same problem with core0
• When LM released P3Dv3 in 09/2015 they could have re-build the FSX graphics engine. They didn´t. The same problem with core0
• When LM released P3Dv4 in 05/2017 they could have re-build the FSX graphics engine. They didn´t. The same problem with core0

I purchased V1 and found the core0 problem. I waited 7 years to purchase V4...and still the same problem.

In the Release Notes (https://prepar3d.com/home-slider/2020/04/115906/) you can see that the word “fixed” appears 34 times. That is to say that V5 it is not a new simulator or a new version of V4, but rather an update of V4 with the corresponding "fixes" in its failures.

Please, take a look at teaser trailer (https://www.youtube.com/watch?time_continue=1&v=Jz6KE8dflBc&feature=emb_logo) between 00:34 and 00:38 seconds. Don´t you think you are seeing FSX again?. Again and again.

Why do you think this time, with V5, will be different?.

But, of course, you are free to purchase it again. Again and again.

Happy flights in core0!!!!

Edited April 10, 20206 yr by mauri2007

Mauri2007 - 

You're exactly right.  I'm disappointed it may not be fixed.  It is an absolute barrier to making good use of multi-core processors. 

Like you, I've been aware of this since FSX was first released. There have been hundreds, possibly thousands, of threads about this on various forums over the years.  And countless suggestions for mitigating actions (i.e., playing with affinity masks and assigned add-on threads to other cores).

The requirements for flight sims in this area are no different that in other games (e.g., first-person-shooter games and driving simulators).  The only thing unique to flight sims is the size of the field of view surrounding the eyepoint (possibly a few hundred miles at altitude).  It's the size of the FOV that makes the ability to efficiently use multiple cores to parse the FOV and feed the GPU directly so important.

My guess is that some folks will continue to have stutters and pauses.  I hope the use of DX12 helps reduce these issues.     

15 hours ago, Dirk98 said:

I can't remember the details exactly but somebody from the software gurus on these boards had explained why all flight simulators would always be very much dependent on a single thread performance, it's like a law of the universe. 

And this is the problem with those "gurus". They are sometimes totally wrong and people who even try to question some of their "truths" are immediately attacked by a huge fan base.

If you ever see someone say:

"all flight simulators would always be very much dependent on a single thread performance"

My recommendation is to downgrade that person's guru status immediately.

There sure are some bad legacy software that, in their current state, would never benefit much from multiple cores without being rewritten from scratch. This is absolute not the same thing as "all flight simulators would always ...". Still, this "truth" among others is repeated over and over almost daily in all those never ending core0/hyperthreading/affinitymask threads.

So with a performance gain ( FPS ) from 50-100%, depending on the amount of detail you set V5 you are still worried about core 0....  😄

If you use the same settings in V5 than in V4 and you set Unlimited / 30 hertz then you will notice that core 0 is less occupied ..

LM has done a lot to take away threads from the main core and put it on other cores. And that was with every release...

Now with V5 it even gets way better.

😎

Ah, another thread with people pontificating about graphics engines. It's good to see that we have some coding experts here who fully understand how many lines of code that P3d5 shares with FS9.

Interesting to see the complex world of multi-threading tasks, something that has plagued software development for years and still do, especially gaming, is now being described as a bug.

Always so many experts on these forums telling LM, and in the past Microsoft, what they are doing wrong.

I can't understand why they are not invited onto the development teams, when they have so many brilliant ideas to contribute. Especially when LM and MS  are so short of software designers with their skills. 🤨

24 minutes ago, Biggles2010 said:

Always so many experts on these forums telling LM, and in the past Microsoft, what they are doing wrong.

I can't understand why they are not invited onto the development teams, when they have so many brilliant ideas to contribute. Especially when LM and MS  are so short of software designers with their skills. 🤨

Heh, have to agree there, all cooks, no chefs 🙂

My P3D config seems to manage expanding across cores pretty well.

This pic is @ 30 FPs vsync'd and core use is pretty spreed. I guess it's all in the configuration.

Cheers

 

2 hours ago, mauri2007 said:

LM never wont give you an answer to core0 bug.

I can answer for you: the "bug" sits in front of his monitor, loading tons of addons, pushing sliders to the right beyond the capabilitys of his hardware and ist then wondering why the hell Core0 of his CPU gets hammered.