How Do You Think The New 3D XPoint Technology Will Effect P3D Developement?

[shivers9 1,174]

I assume this is why many people on flightsim forums are desperately always looking for a new silver bullet hardware technology instead of software fixes. The bottleneck in Prepar3d is not today's hardware. Not the memory, not the IO, not the storage, nothing.​

 

Just to be clear...I for one am not looking for a silver bullet as I have had Prepar3D running very well for a good long time now. Yep some people have problems but then any given software is never going to perform on every computer out there. Their are people out there who write nasty remarks about FSX and P3D because it won't run on their IPhone for gods sake.

[Lorby_SI 2,038]

I assume this is why many people on flightsim forums are desperately always looking for a new silver bullet hardware technology instead of software fixes. The bottleneck in Prepar3d is not today's hardware. Not the memory, not the IO, not the storage, nothing.​

 

I beg to differ...

 

"P3D is old, ineffective code": until Microsoft releases the ESP source code into the public domain, this is nothing but conjecture. IMO if anyone knew how to implement a flightsim efficiently, it would have been the guys who did this for 20 years already, and on limited hardware no less.

I know of the argument that ESP does not make use of the modern technology. That is very easy to say, but very hard to prove.

Actually, P3D doesn't even need to. For what it has to do as a simulator, raw processing power is required. Fact is, that in the past 10 years, this has not

progressed at the same rate that it used to. Maybe if the code gets rewritten and compilers make use of the updated CPU instruction sets, some

improvement can be seen. Although recompiling it with a current version of Visual C++ (FSX SE) didn't do anything major for it.

If it has to do the same things, it will run roughly the same. Proof? Example: all real flight simulator platforms show about the same performance.

 

I think 3D XPoint sounds great, and it will have a major performance impact, not only on P3D, but on every software out there.

The application of this tech will not be limited to replacing RAM and mass storage in computers. Next we will get the talking fridge (need a lot of space for good sound samples)

 

SSD: "You will only notice better load times": I have been reading this more often lately, wonder where that came from.

What does "only" mean? P3D is constantly reading stuff from the disk, if this gets faster, your sim will benefit. Are you saying that many a simmer worrying about fragmentation and disk access times was wrong, that it was irrelevant even?

I think one has once again to distinguish here, what kind of PC you are running this on and how your Sim is set up.

On my system the difference between HDD and SSD was very noticable, in fact it was so massive that I bought a second SSD and controller right away, just for running the scenery. There are some things that can go wrong, like running the sim from the SSD and the scenery from a HDD on the same SATA controller, wrong SSD alignment, wrong handling of the SSD (like defragmenting it...).

This is what I figure: if your system already struggles to run the sim on the settings you chose, the SSD won't do much good. But if you have a powerful system, built exclusively for the purpose of running the sim, and it actually runs so well that the disk access becomes your bottleneck, then the SSD will help a great deal - as it did with mine.

 

Comparing simulator software to 3D games is no valid experiment. Their inner workings are very different. Games are more like movies really, they simulate very little but do quite a lot for show. Food for thought: imagine you would expand GTA with something simulated: give every 'living being' a heartbeat and a set of lungs. These should be roughly simulated, so that they get out of breath, react to heat and cold, get a heart attack... Your own avatar's inner workings have to be simulated in greater detail of course. And give the NPCs a purpose, so that they do not wander aimlessly (like AI flightplans...). The GPU won't be able to help with this, it is all on the CPU. Wonder how that would work out in terms of performance.

 

As I see it, one of the biggest problems in these discussions is, that there is no clear baseline to define "better" or "worse" when running a sim. We are always shooting at moving targets here, as people are constantly changing settings and twiddling with the setup. More often that not, "truth" is siphoned from statistics - "if I feel a lot of people are posting the same thing, so it must be true". Mathematically speaking, if you don't know relation of your count to the total number, the statistic is meaningless. And how often does it happen, that when someone buys a new component or whole system, the first thing he does is to push them sliders to the right, expecting the new system to so massively outperform the old one, that not only the same settings are improved, but he/she can UP them right from the start. This was true up until 2005 I think, but not any more. So we end up with diverging points of view, and discussions about open platforms like they were closed boxes like game consoles. Imagine everyone building their own Playstation from arbitrary components - how reliable would their experience with the same software be to each other?