Is the wait going to be 3 years? When does 5-7 GHz become common without OC

 

Clock speed hasn't changed appreciably for 4-5 years, in fact it's become worse at least in the Intel line secondary to what they did w/ Ivy Bridge & Haswell's thermal interface material leading to higher core temps and hampered overclockability.  Your SB processor probably overclocks a little better than a new Haswell would.   Apparently obtaining progressively higher clock speeds isn't as easy to accomplish as it once was. 

 

The real change I am waiting for is possible today, without needing to overcome the apparently inherent problems w/ increasing clock speed.  And that is to design the simulator engine to be maximally multithreaded, utilizing as many cores & physical CPUs as one can put into the hardware.  More use of large system and video ram amounts and modern graphical APIs like DX11 as well since these are available right now.   FSX doesn't exploit all of the power already available in modern high end systems.  I designed my current rig to run a better simulator very well, while running FSX 'good 'nuf'.  XPlane 64 tries to solve some of this, and I'm hoping Prepar3D goes even farther.

hopefully by then that will all be ironed out, or alternatively the great guys at ORBX will give us spectacular terrain,  mesh etc that it will look like real life scenery and we will be still using FSX. 

Not as long as Austin thinks the current system "Rocks" as he stated in the AVSIM teamspeak conference.

So long as FSX is a 32bit OOM factory it doesn't matter how much faster the chips get.  We have hit a wall already.

Clock speed hasn't changed appreciably for 4-5 years, in fact it's become worse at least in the Intel line secondary to what they did w/ Ivy Bridge & Haswell's thermal interface material leading to higher core temps and hampered overclockability.  Your SB processor probably overclocks a little better than a new Haswell would.   Apparently obtaining progressively higher clock speeds isn't as easy to accomplish as it once was. 

 

The real change I am waiting for is possible today, without needing to overcome the apparently inherent problems w/ increasing clock speed.  And that is to design the simulator engine to be maximally multithreaded, utilizing as many cores & physical CPUs as one can put into the hardware.  More use of large system and video ram amounts and modern graphical APIs like DX11 as well since these are available right now.   FSX doesn't exploit all of the power already available in modern high end systems.  I designed my current rig to run a better simulator very well, while running FSX 'good 'nuf'.  XPlane 64 tries to solve some of this, and I'm hoping Prepar3D goes even farther.

 

Clock speed hasn't gone up appreciably because for most people, 3Ghz is "good enough". What they care about is power usage and battery life, and that's what you see in Haswell. Not much faster, but by all accounts a lot better in battery life.

 

Fully utilizing multiple cores is a hard, hard problem.

 

Cheers!

 

Luke

 

 

Fully utilizing multiple cores is a hard, hard problem.

 

Oh, yeah. I will always remember lashing two mainframe "cores" together, and they both provided less performance than one. Shared memory is a killer if the cores don't work and play well together. Use a primitive spin-lock and burn cycles waiting for the other processor to release the resource? Inadequate granularity for shared memory operations? Start lock convoys when everyone is waiting in line? And my personal favorite: CPU 1 is bored - I think I will steal the workload from CPU 0.

 

Flashback...

I am wondering how dramatic the impact of 5-7 GHz processors will be for FSX? Because right now even 4 GHz doesn't allow for the ultimate experience. At 5-7 GHz will this sim get a new lease on life? I cannot see how it wouldn't. Right now I have an i5 2500k with a 5850 and feel like I am just waiting for an amazing CPU, at which point I will get the next gen GPU to go with it.

 

Is the wait going to be 3 years? When does 5-7 GHz become common without OC?

 

When FSX was being developed, Intel was touting its Netburst architecture, engineered to eventually reach 10 GHz. The Pentium 4 topped out at 3.8 GHz in 2005. Unfortunately, the laws of physics got in the way and around 2006, Intel switched to the Core architecture and started focusing on lower power usage and multi-core efficiency.

 

Since then, the focus has never been on clock speed. In fact, while some Sandy Bridge CPUs could reach 5 GHz, it has become *harder* rather than easier to achieve this with the newer Ivy Bridge and Haswell, that typically top out at 4.5 GHz or so.

 

Traditional CPUs made of silicon wafers will probably never move beyond ~4.5 GHz. Even if a 7 GHz CPU would come out, the difference would not be dramatic. 7 GHz is 55% higher than 4.5 GHz. So if you get 15 FPS at a complex payware airport with a complex payware aircraft, the 7 GHz CPU would still give you only 23 FPS.

 

The bottom line is that FSX is written for a type of architecture (single-core, high clock speed) that is no longer being actively pursued by the industry. To achieve significant improvements, the sim would have to be re-architected for modern CPUs and GPUs. This would result in huge gains even on current systems. This is the main reason why P3D 2.0 is so exciting.

 

 

This is the main reason why P3D 2.0 is so exciting

 

Are you saying this because we have some evidence this is their trajectory, or because we hope this is their trajectory?  I hope it's the former!

 

 

This is the main reason why P3D 2.0 is so exciting.

 

But P3D will still be 32bit so... don't expect too much of it.

Are you saying this because we have some evidence this is their trajectory, or because we hope this is their trajectory?  I hope it's the former!

 

Eveidence. From Wesley Bard, Lockheed Martin:

"We are now updating our rendering system to support DirectX11 and development is progressing nicely. With DirectX11 support, Prepar3D can leverage features such as hardware instancing, shaders, GPU-based particle systems, tessellation and multi-threaded rendering."

 

But P3D will still be 32bit so... don't expect too much of it.

 

The move to 64-bit will come with LM feels the sim and surrounding eco-system is ready. For now, they are concentrating on removing legacy code and modernizing the engine. This will make the transition to 64-bit easier. The optimizations should also reduce VAS usage, which will be an adequate short-term solution.

 

 

Fully utilizing multiple cores is a hard, hard problem.

 

Luke, I've heard this same comment re fully utilizing multiple cores especially in something like a simulator engine.  Even so, isn't this where to put the focus if you were building a new simulator engine, as difficult as it would be?  I still think the very fact there are whole companies and vendors dedicated to supporting 7 y/o FSX is a tacit indication that there is certainly a market.   I guess 'fully' might be the real challenge, but 'much more' leveraging of multicore could be meaningful.  Apparently FSX uses some extra cores for various tasks and so this can be exploited more and maybe P3D will go there.  I've heard XPlane uses add'l core by dedicating air traffic to other available cores.

The move to 64-bit will come with LM feels the sim and surrounding eco-system is ready. For now, they are concentrating on removing legacy code and modernizing the engine. This will make the transition to 64-bit easier. The optimizations should also reduce VAS usage, which will be an adequate short-term solution.

 

 

Did someone from LM say this or is this a personal guess? From what I've always understood going to 64 bit requires a total rewrite. Looking at how long they are working on version 2.0 (DX11) you would think going to 64 bit will take even longer... 

One cannot use clockspeed in itself as a measure of performance. The only time you can use clockspeed to compare performance between two cpus is when the cpus are of the same architecture. You could have a CPU that runs at 1ghz that would outperform a CPU running at 2.5ghz.

 

https://en.wikipedia.org/wiki/Megahertz_myth

Can get 30 (locked at 30) FPS at KSEA with everything maxed out now that I have FTXG. I can also get 50-55 fps at FSDT's KLAX (FPS unlocked) with the MS CRJ700 but if I drop in the NGX or Aerosoft A320X it drops to the mid 20's.

Did someone from LM say this or is this a personal guess? From what I've always understood going to 64 bit requires a total rewrite. Looking at how long they are working on version 2.0 (DX11) you would think going to 64 bit will take even longer... 

They've made the statement, that they would like to go to 64bit, but because FSX still has some legacy assembler code dating back many versions they have to take this slow. They either have to rewrite the assembler code to C/C++ routines is possible, or convert the assembler code from 32bit to 64bit. That's not an easy process, and will take time to do it right. So I don't think we're going to see a 64bit version anytime soon. If the efficiencies of DX11, is what is claimed, and LM takes full advantage of them in 2.0, going 64bit, may not even be necessary, at least for a while.

Time it comes out, so will Win 9 or 10 which will be so FSX unfriendly, nobody will be using it.

 

 

Microsoft....greedy merchants.

 

How do you figure, FSX on Win8-64 seems to be better than my Win7 and certainly better than on my XP machine.