I just spent the day playing with the X-plane 10 demo 64 bit. It is impressive. I could not max it out and get a fluid experience at the same time and it did make my system crash due to overclocking so I Know it was pushing all the hardware to its limit. I might buy it after I build my new system latter this year but there is in my opinion a certain something that FSX has that makes it a richer experience. I still believe that P3D is where the FSX fans will ultimately go not to X-plane not to say that it is not very impressive and it has its following but FSX fans again will go after better hardware to get what they want out of FSX and not switch to X-Plane especially since they will need that next gen hardware either way!

I strongly disagree, but this opinion is based on third party developers moving to XPlane, and this remains unsure at this time. This is the sensible pathway because then, and only then, will future hardware development (increasing parallelism) be in sync w/ software leaving lots of headroom for future development. Supercomputers are based on this, not whipping a few cores into a frenzy. And of course, the software programs running on supercomputers are designed to exploit this. FSX will forever be stuck in this, and getting to 6Ghz for example, a stretch unless something gives in a big way, is only going to get you another 20% over 5Ghz. Think how long it's been since meaningful increases in overclocking have occurred. 3-5 years? When you load up the NGX at KSEA w/ all sliders right in PNW scenery and you're seeing 21FPS, you will get to a whopping 27fps. That is good enough I would admit, but that becomes the end of the line for increasing complexity. You will never be able to get to high resolution close up ground textures for example at 6Ghz, only w/ spreading the processing load over 6, 8, 16 cores, etc. Maybe P3D will offload enough to the GPU to add some more complexity and depth--we'll see.

 

I had to respond in a similar fashion re: parallelism in another thread, so I'll just re-post it here:

 

Making software run faster via parallelization is a non-trivial problem. Not all tasks are able to be executed in parallel to other tasks due to data dependencies. Of those tasks which can be parallelized, not all of them stand to gain any performance from it. I'm as much for advancement as the next guy, but single thread performance is still critical and likely always will be.

 

As for X-plane migration, too many people have far too much invested in MSFS to even contemplate such a move, hence why it always seems to hit a brick wall in marketshare even with new releases.  

Even if add-on developers produce similar quality add-ons for X-plane as what's available in FSX, it will take years to build up that catalog and convince anyone with a serious investment to migrate.  There are still lots of FS2004 users out there that won't even touch FSX due to their current software investment, let alone an entirely different platform.  X-plane will always be an "also ran" because of this.  In fact, unless P3D makes significant strides in performance while also maintaining add-on compatibility, or some 3rd party does the same, we'll probably all still be on FSX 5 years from now.  

 

Don't get me wrong, I desperately want something better than FSX to come out, but I'm not excited about the prospect of essentially throwing away thousands of dollars of software I've invested in for FSX so I'll hold out until something comes along that can preserve at least some level of compatibility.  

 

 

Pros: 16 physical cores (2-Socket WS Mobo) and 128GB of RAM take your productivity to entirely new levels. If your workstation is a *tool* and not a toy, these 8-core E5 CPUs decimate even the vaunted 3960X 6-core units by an order of magnitude in C4D and AE/Premiere. If you're a video and/or 3D motion graphics pro, just get two of these via any means possible -- sell blood, superfluous appendages, excess children, or whatever you need to do to get these. You'll make enough back in productivity gains to buy them back later.

 

I use C4D and Adobe PP CS6 ... both products will use as many CPU cores as you toss at them, so you should expect to see 16 cores out perform 6 core 3960X in these applications.  I use a 64 Core AMD "workstation/server" today to do rendering but it doesn't run FSX very well  

 

Haswell isn't going to do much for FSX, sadly :(

 

 

Don't get me wrong, I desperately want something better than FSX to come out, but I'm not excited about the prospect of essentially throwing away thousands of dollars of software I've invested in for FSX so I'll hold out until something comes along that can preserve at least some level of compatibility.

 

This is VERY unlikely, you've kinda put yourself between a rock and a hard place.

prospect of essentially throwing away thousands of dollars of software I've invested in for FSX

 

I hear you, however the other side of this very issue is how much more do you want to invest in this platform? I think FTX England was my last major purchase as I have a little sense I'm throwing good money after bad. And in the end, were I to get rid of addon's I don't use, the $1000's of investment would likely shrink back greatly. I only fly 4 birds any longer: QW757, Super MD80 Pro, NGX & Turbine Duke. As for sceneries, I rarely fly down under, but do enjoy the US regions & a few airports. I'd rather invest in something w/ much more headroom for future development and discard the FSX stuff if something, XP or whatever comes along. Moreover, how about how much hardware people pay for just to eek out a few more frames for FSX? Upgrades every 2-3y? People will clamor to keep their $1000 worth of software, while buying $1,500 - $3000 worth of hardware every 3 years.

 

As for parallelism, I'm only going by LR's advertisement, i.e., that their simulator exploits this much better. To be honest, I don't really know if it does. If that is accurate, and their team is as small as I've heard, then the challenge of programming for better use of parallelism is certainly on the table, especially in light of the fact they program for 2 or 3 completely different OSs (!).

I use C4D and Adobe PP CS6 ... both products will use as many CPU cores as you toss at them, so you should expect to see 16 cores out perform 6 core 3960X in these applications.  I use a 64 Core AMD "workstation/server" today to do rendering but it doesn't run FSX very well  

 

Haswell isn't going to do much for FSX, sadly :(

 

 

 

 

This is VERY unlikely, you've kinda put yourself between a rock and a hard place.

 

There have been some exciting developments recently discussed on this forum, both for the future of P3D and a 3rd party project in the works.  I'm not counting on the latter, but the former is very likely to bear fruit and as soon as this year, according to the most recent information.  

 

I hear you, however the other side of this very issue is how much more do you want to invest in this platform? I think FTX England was my last major purchase as I have a little sense I'm throwing good money after bad. And in the end, were I to get rid of addon's I don't use, the $1000's of investment would likely shrink back greatly. I only fly 4 birds any longer: QW757, Super MD80 Pro, NGX & Turbine Duke. As for sceneries, I rarely fly down under, but do enjoy the US regions & a few airports. I'd rather invest in something w/ much more headroom for future development and discard the FSX stuff if something, XP or whatever comes along. Moreover, how about how much hardware people pay for just to eek out a few more frames for FSX? Upgrades every 2-3y? People will clamor to keep their $1000 worth of software, while buying $1,500 - $3000 worth of hardware every 3 years.

 

As for parallelism, I'm only going by LR's advertisement, i.e., that their simulator exploits this much better. To be honest, I don't really know if it does. If that is accurate, and their team is as small as I've heard, then the challenge of programming for better use of parallelism is certainly on the table, especially in light of the fact they program for 2 or 3 completely different OSs (!).

 

X-plane does indeed exploit parallelism to a much greater degree, but what doesn't when you compare to FSX?

 

 

X-plane does indeed exploit parallelism to a much greater degree, but what doesn't when you compare to FSX?

 

I rest my case ;o)

until know nothing compares to FSX!!!

Just to clarify, Parallelism (in terms of programming) doesn't mean multi-threaded. Also, Multi-threaded can work on a single CPU. Need to keep this in mind in case someone is trying to sell you these terms . Parallelism is more commonly associated with the repetition of the SAME process over many processing devices (CPU, networked computers, etc. etc.) -- works well for Seti, Rendering/Compression, but doesn't work so well for 3D interactive games/simulations. Parallelism is a bit to "generic" of a term to be meaningful in the context of a flight simulations.

 

For flight simulation you are basically trying to portrait a 3D space on a 2D plane (your monitor or monitors). The basic problem is that many "different" activities need to happen at the same time and be finalized on that single 2D plane. For example, calculation of your air speed vs. the current position of the virtual world vs. the current position of AI aircraft vs. location of a cloud etc. etc. ... these are all very different processes that need to "come together" and produce the final frame. Computationally, these processes will never finish "at the same time" as they have different calculation requirements, so a "process cop" (thread manager) needs to be in place to ensure everything is done in order to process that frame.

 

To do this efficiently, you HAVE to build the flight simulation from scratch with this concept in mind - neither FSX, P3D, nor XP10 do this. Not end of the world -- there are still benefits to taking smaller chunks of different processing and threading them uniquely, but you'll still need a thread manager to make sure that AI aircraft hasn't travel 15X further just because that thread was able to calculate it position in 15 times more frequently than 5 cloud layers on a stormy day. "Time" is the thread manager.

 

Time -- and that's a KEY concept to distinguish ... C4D and Adobe PP rendering is done when it's done, could be 2 minutes could be 2 hours. In other words, these processing tasks don't really care about time. On the other hand, 3D interactive simulations, TIME is a critical element of the illusion

 

FSX/P3D is multi-threaded, so is XP10 ... put FSX, XP10 and C4D or (Adobe PP) or some video compression software (that is multi-threaded) under a process monitor and look at their CPU utilization over time. C4D and Adobe PP will pretty much max out CPU utilization across all available CPUs (if setup to do so) ... FSX, XP10 will not (even with Affinity settings). Why, because C4D and Adobe PP tasks fit a much easier processing requirement of a the same task being repeated over and over and over ... hence much higher CPU utilization. Where-as FSX, XP10 their processes aren't all the same and lots of waiting is introduced before a correct final frame is displayed.

 

My 2 cents.

Just to clarify, Parallelism (in terms of programming) doesn't mean multi-threaded. Also, Multi-threaded can work on a single CPU. Need to keep this in mind in case someone is trying to sell you these terms . Parallelism is more commonly associated with the repetition of the SAME process over many processing devices (CPU, networked computers, etc. etc.) -- works well for Seti, Rendering/Compression, but doesn't work so well for 3D interactive games/simulations. Parallelism is a bit to "generic" of a term to be meaningful in the context of a flight simulations.

 

For flight simulation you are basically trying to portrait a 3D space on a 2D plane (your monitor or monitors). The basic problem is that many "different" activities need to happen at the same time and be finalized on that single 2D plane. For example, calculation of your air speed vs. the current position of the virtual world vs. the current position of AI aircraft vs. location of a cloud etc. etc. ... these are all very different processes that need to "come together" and produce the final frame. Computationally, these processes will never finish "at the same time" as they have different calculation requirements, so a "process cop" (thread manager) needs to be in place to ensure everything is done in order to process that frame.

 

To do this efficiently, you HAVE to build the flight simulation from scratch with this concept in mind - neither FSX, P3D, nor XP10 do this. Not end of the world -- there are still benefits to taking smaller chunks of different processing and threading them uniquely, but you'll still need a thread manager to make sure that AI aircraft hasn't travel 15X further just because that thread was able to calculate it position in 15 times more frequently than 5 cloud layers on a stormy day. "Time" is the thread manager.

 

Time -- and that's a KEY concept to distinguish ... C4D and Adobe PP rendering is done when it's done, could be 2 minutes could be 2 hours. In other words, these processing tasks don't really care about time. On the other hand, 3D interactive simulations, TIME is a critical element of the illusion

 

FSX/P3D is multi-threaded, so is XP10 ... put FSX, XP10 and C4D or (Adobe PP) or some video compression software (that is multi-threaded) under a process monitor and look at their CPU utilization over time. C4D and Adobe PP will pretty much max out CPU utilization across all available CPUs (if setup to do so) ... FSX, XP10 will not (even with Affinity settings). Why, because C4D and Adobe PP tasks fit a much easier processing requirement of a the same task being repeated over and over and over ... hence much higher CPU utilization. Where-as FSX, XP10 their processes aren't all the same and lots of waiting is introduced before a correct final frame is displayed.

 

My 2 cents.

@, hi there!! I saw your specs and since you are running a 6-core processor I have to ask is you test your system on FSXMark11? My question (out of the topic, I know Sorry!!), if it is posible to achieve the same FSX performance with your specs comparing with a I7 quad processor also OC to 4,8 GHz just considering the processor diference, lets say we use same RAM, same GPU? you can post here if you find it better http://forum.avsim.net/topic/404598-fsx-overclocked-45-or-higher/

 

 

Where-as FSX, XP10 their processes aren't all the same and lots of waiting is introduced before a correct final frame is displayed.

 

Sounds all plausible thanks for those insights.  Given your nice overview of what's relevant in this discussion, if you could quantify how well XP10.10 exploits multiple cores/threads compared to FSX, do you have any idea of their relative differences?  For example, if you assign a value of 1.0 for what FSX does, what value would you assign XP10.10?  Probably pretty hard to say w/o really examining code and processes running, yes?  Intuitively, I sensed the degree to which code could be optimized for multithreading would be very very complex & challenging.  Again, how well does XP10 do this do you think?

I rest my case ;o)

 

To be fair though, even with X-plane's inherent parallelism (or multi-threading, if you prefer) it can still bring even the fastest current machine to it's knees by cranking up the weather and traffic settings (much like FSX).  

To be fair though, even with X-plane's inherent parallelism (or multi-threading, if you prefer) it can still bring even the fastest current machine to it's knees by cranking up the weather and traffic settings (much like FSX).  

Even w/ dual Xeon 8-core 3.8Ghz processors, 32Gb (64Gb?) of DDR3-1800, and a 6Gb Titan GTX?

Its very simple. I am overdue a rebuild. Its going to be a Haswell, a titan or 780 and 16 GBs of fast RAM all overclocked to the max. At that point I will buy a used copy of X-plane 10 on Ebay and hopefully an XP10 scenery package from ORBX or equivalent and something like REX to replace the XP10 Horrible Sky. (You don't hear to much about that: the sky in XP10 is Horrible). Then I will compare the two. I anticipate that I will much prefer FSX but my mind is open! If I Prefare XP10 I will keep it and wait to compare it to P3D V2 on the same Hardware. Hopefully in the next year I will have decided on the flight sim future for me and until then I am not buying anymore Addons. I have plenty. Oh wait a minute, there is the mail man in the evening!, It must be my flight1 mustang that I got on sale for $19.99. LOL! Anyway I am sure that many enthusiasts purse strings for FSX Addons are tightening as we speak at least if they are like me. So someone better spill the beans soon.

However I intend to spend up to a few $1000 to build the mother of all FSX rigs this year.

Even w/ dual Xeon 8-core 3.8Ghz processors, 32Gb (64Gb?) of DDR3-1800, and a 6Gb Titan GTX?

 

No one around here is running such a system but I'd be willing to bet the answer is yes.  Turn up the weather settings and get enough AI aircraft involved and watch the FPS plummet.  

No one around here is running such a system but I'd be willing to bet the answer is yes.  Turn up the weather settings and get enough AI aircraft involved and watch the FPS plummet.  

Well then XPlane 10.10 isn't exploiting multiple cores as they advertise they do.  I for one would build that system in a heartbeat if a simulator could use it.  From LR's website:

 

 

Recommended System Specifications

For the best experience, we recommend the following system specs:

• a 3 GHz, multi-core CPU (or, even better, multiple processors),
• 4 GB of RAM,
• a DVD-ROM, and
• a DirectX 10-capable (DX11 preferred) video card with 1 GB of on-board, dedicated VRAM.

Other Notes

X-Plane will take advantage of as many cores or distinct processors as you can afford. Having 16 cores split among 4 CPUs is not required by any means, but Version 10 would be able to use every one. No more than 4 GB of RAM is necessary, but the more VRAM you have, the better–X-Plane 10 can easily use 1.5 GB of VRAM at the maximum settings.