duo core PC question

[sagamol 0]

I read that fsx FPS can be dramatically increase, if you use both cores on the processors, and that by default only one core runs fsx. me being a total dumb-a... when it comes to things of this nature, I would like to hear a "simple" explanation as to how I get to use both cores on my processor.

any help would be appreciated...

Tom

[Snave 466]

You read wrong. There are only advantages to having FSX on a dedicated core, and routing all other applications to another. FSX core engine is simply NOT multi-core compliant, although it is multi-core compatible:-

http://blogs.msdn.com/tdragger/archive/200.../21/711152.aspx

and later:-

Originally Posted by Phil Taylor

FSX SP1 News:Intel quote

Today Intel announced their new 2.93G QX6800 Quad Core processor here.

As part of this press release, Aces Studio participated with a quote about our improved multi-core support in SP1. The quote stated

“The latest version of Microsoft Flight Simulator* X, Service Pack One (SP1), due out later this month, is a great match for the extreme multi-core processing delivered by the new Intel Core 2 Extreme quad-core processor”

and

“Flight Sim X SP1 greatly increases multicore utilization and will scale as more threads are available leading to reduced load times as well as frame rate improvements and greater visual complexity during flight. The Flight Simulator team at Microsoft is pleased to work with Intel to provide our end users with a great gaming experience."

And I wanted to take some time and explain what this means for FSX and FSX flyers.

Our multi-core support will take advantage of both 2 and 4 cores today, and more cores in the future when they become available via a config setting. This is for both Intel and AMD processors.

At load time, we run the terrain loading on threads across the cores. This can result in reduced load times, the actual percent reduction can vary but it could be reduced by as much as 1/3.

At render time, we run the terrain texture synthesis on threads across the cores. During flight on multi-core machines, as terrain and terrain textures are loaded you will notice significant multi-core usage. As all tiles are loaded, the multi-core usage will fall off, this is expected. As the terrain is re-lit, approximately every minute, you will see multi-core usage increase. As you bank and load terrain tiles, or as you fly forward and force a load of more terrain tiles, you will see the multi-core usage increase.

At render time, we now perform more extensive batching of objects to reduce our API Draw calls. The batch rebuilds are also performed on a thread and scheduled on cores.

For those of you on single core machines, we did do some performance work to benefit you as well. The batching work will help even on a single core. Plus we made some changes to animations to make them perform better and we modified how we draw trees to reduce API SetTexture calls.

So we really tried to hit all the wickets with our performance work. I still don’t want to make any final FPS gain claims, and will stick with my conservative 20%. We expect it to be more, but we need to get Beta2 out and see where we are.

As far as the release date for SP1, we need to get Beta2 out and see what remains. The last day of April is still possible, but any significant Beta2 bugs will cause us to fix them and delay – we want SP1 to be right for the community and are not tied to a date.

Note - hyperthreaded is not multi-core. Our current plan is to treat HT machines as single-core since we noticed extensive collisions between threads which caused stutters.

When you read through the PR BS, what this means is that they have optmised the engine to make best possible use of dual- or multi-core, but that is a LONG way from actually making the engine suited to multi-core application.

Any fps gain is small, the loading time reduction only of consequence at startup or re-set.

[Robert S 75]

Judging by real world experience, I'd like to disagree with Snave here.

On a general level, applying SP1 to FSX not only enables multiple-core utilization (where before FSX only hit one core), but it does render FPS increases as well which can be significant. The FPS gains you might see will vary depending on your computer configuration, but as Phil Taylor states a 20% increase should be expected...some people have seen higher gains, averaging about 30%.

So, to answer your original question sagamol, the simple way to as to how to get to use both cores is to apply SP1, available from www.fsinsider.com

Post your results here after you finish! Hopefully, you will see an improvement that will make a difference to your flightsim experience.

Also, as Snave points out, your load times should decrease as well.

[sagamol 0]

first of, I have to say that my problem was, I forgot to install sp1 after a complete re-install of fsx. now that I did that, my load time is cut almost in half and my frame rates have improved. the seattle example was after sp1 install, before, the fps were only 8.5 to 10 at the most. also the autogen is much much smoother then before sp1. so overall.....HUGE DIFFERENCE for me....

thank you all for getting me on the right track....love to learn as I go... :roll:

[Snave 466]

Judging by real world experience, I'd like to disagree with Snave here.

On a general level, applying SP1 to FSX not only enables multiple-core utilization (where before FSX only hit one core), but it does render FPS increases as well which can be significant. The FPS gains you might see will vary depending on your computer configuration, but as Phil Taylor states a 20% increase should be expected...some people have seen higher gains, averaging about 30%.

So, to answer your original question sagamol, the simple way to as to how to get to use both cores is to apply SP1, available from www.fsinsider.com

Post your results here after you finish! Hopefully, you will see an improvement that will make a difference to your flightsim experience.

Also, as Snave points out, your load times should decrease as well.

As SP1 includes a raft of fixes designed explicitly to improve fps, I don't see how you can add 1+1 and get five? For sure, SP1 DID improve multi-core compatibility, and I have provided the sources that explain that conclusion - but that's a LONG way from being able to assume that the improvement in SP1 is attributable to multi-core.

In fact, I ALSO see a 20% improvement ni fps in SP1 - and I don't have a multi-core processor...

In the light of that, perhaps you could explain how you have arrived at your conclusion, which seems to fly in the face of what ACES themselves tell us? Do you know something ACES don't..?

[Robert S 75]

No Snave, I (as well as anyone else who has been reading Phil Taylor's blog) seem to know something that YOU don't.

From Phil's blog (ref: http://www.fsinsider.com/tips/Pages/SP1How...InstallIt.aspx)

Performance

Addressing persistent issues with FSX performance was the driving force behind SP1. How did we do that? With a lot of hard work! :) We had two goals: *add* multi-core support and address the most pressing performance hot spots.

So here Aces is ADDING multi-core support to FSX with SP1, not [and I quote] "improve multi-core compatibility" as you put it. This then, renders your suggested best practice:

 There are only advantages to having FSX on a dedicated core, and routing all other applications to another.

as utterly moot.

Now, by ADDING multi-core support in SP1, ACES was able to increase performance by (again, reference the same SP1 blog from Phil Taylor as above):

Moved DEM loading to threads.

Moved terrain texture synthesis to threads.

Moved Autogen batch rebuilds to threads.

whereby BEFORE SP1 these tasks were relegated to a single core, creating a performance bottleneck.

Now on to your next inspired insight: You state that loading time is only improved "at startup or re-set". By adding SP1 this also is not true. Once again, from Phil:

During loading, we run the DEM loader on threads. You'll see good balanced usage across all cores, as well as about 1/3 faster load times on average.

So you're only partly correct, because WAIT! There's more:

During flight we spawn threads for Autogen batch rebuilds as well as the terrain texture synthesis. The terrain texture work tends to be a bit bursty; as an area gets generated the load reduces true. But as you fly forward, as you bank, and as the terrain is lighted ( once a minute ) threads are spawned. The terrain grid system is radial around the current viewpoint, and, depending on level of detail, radius can be up to 4.5 tiles in either direction, something like 64 tiles. So there is plenty of work to go around. Autogen is more constant, with a 2km extent being batched.

Even given the bursty nature of the core usage when flying, when there is load, its pretty balanced across the cores. And we got rid of as much of the stutters as we could by going to a lock-free synchronization style. It’s solid work that we are deservedly proud of.

So here you can see that there are "load threads" in-flight as well, which realize an improvement from multi-core utilization as well, not just "at startup or re-set". :roll:

And finally, from Mr. Taylor:

Conclusion

With all that said, the Draw and SetTexture API call reductions and Autogen size reductions are probably as important for FPS improvements; the multi-core work really shines for load balancing and reducing stutters and blurries. And both are critical for better scaling as CPUs and GPUs get better.

So Phil is saying that multi-core support specifically for FSX "really shines", and is critical for scaling as CPUs get better (read: quad and beyond cores). Finally he goes on to say:

We think SP1 is going to deliver the goods for most users, and will reward users with better hw the most. We expect that, except in the very,very low end hw, all users should see a 20% gain. Some scenarios will see 40%, and some will see a bit more. Its really going to depend on a lot of variables. We hope this enables users to either fly at the same settings with greater FPS, or to bump the sliders up 1 or 2 ticks and still get the same FPS you had.

Phils says "all users" should see a 20% gain. This explains your 20%, Snave as you admittedly do not have a multi-core CPU. Note that the 20% gain as Phil projects is NOT dependant on having a multi-core, as you suggest. For those that do have multi-core CPUs as well as other modern hardware, Phil projected (and many have indeed realized) improvements of up to 40%.

So to conclude, - no: I don't know anything Aces doesn't know. I do know how to read :wink: though. And my math is also good, as I believe I've properly added 1+1 and come up with 2.

Now it has been my sincere objective not to come across as curt and condescending as you have appeared at times in responding to other users on this good forum...if I have in any way, then forgive me for stooping to that level.

Edit - You should also re-read the very same text you reference from P.T.; I believe you misunderstood that text. It is the press release from Intel stating that: “Flight Sim X SP1 greatly increases multicore utilization", not Aces saying that. But remember that by this point Aces has only just added it to their FSX code, but it is still in beta stage. SO, to put it this way: "You read wrong".

In any case, hopefully this has turned out to be an informative thread for many people 8)

Oh, and sagamol, - I'm glad that a simple application of SP1 has improved your flight simming experience!

[Snave 466]

Sorry, but you simply place too simplistic an interpretation on what was being said. I warned you to read past the PR BS...

It is one thing to iterate - many times - the advantages of `balanced use across the cores`. THAT, my dear boy, is a LONG way short of stating that FSX takes advantage of multi-core processes. It does not, and will not until the core of the sim engine is changed. The obvious conclusion is derived from the fact: HOW do I get a 20% increase in fps post SP1, when the same increase is there for multi-core users..? Obviously - and it truly is blatantly obvious if you had but eyes to see - what has happened is that the developers have `farmed out` some of the processes onto the second core in order to act as a sop to gullible non-technical persons who now see the gauge showing lots of use of both cores and think that MS ACES have somehow given them more performance by accessing the second CPU. They haven't, the processing has been altered slightly to cater for the availability of a co-processor, but that remains some considerable way from making FS a multi-core application. Add the second core and you add a second pipeline, yes. But that doesn't mean that twice as much gas flows down that pipe. You have to be able to open the taps more to get more flow. And no FSX patch is going to achieve more than a small extra turn on the wheel. You need a new pumping house to maximise the gas through both pipes...

If that were not the case, then the jump between two and four cores would result in a performance increase equal to or greater than, the jump between single and dual cores. If that were not the case, then a second identical installation of FSX, shared over a network, would ALSO show the same multi-core advantages. `Greatly increasing multi-core utilisation` is a long way in PR-speak from stating `Takes full advantage of multi-core utilisation`... Add a second, equally powerful processor - and performance jumps by a maximum (that's maximum) of 40%. Add a quad core processor, and the figures do not increase by the same amount, Do the math, boy, that's NOT full utilisatioon of multiple core or else the gains would be greater, and growth proportional to the number of processors. Do think about what is being said, not just believe it because it is...

Look at the very careful choice of words, and now apply an equally careful choice of understanding and you will see that yes, they have improved multi-core processing. But not taken full advantage of it.

I had an almost-empty glass of beer. I added a spoonful of beer to that glass. I now have increased the amount of beer in that glass by one hundred percent, "massively increasing the amount of beer available to drink," in PR-speak... Now, I get handed a second pint glass, and I drip one of those spoonfulls into the second glass. I am now `taking advantage` of two pints of beer. But nobody with any sense would claim I was going to be able to get ratarsed from it. For that, I need to buy more beer from the bar.

The FS bar is not open, and now, post-SP2 never will be. Come back in two years when FSXI starts the pumps, and then I have no doubt that multi-cores will be fully supported and we can have two pints or more on the go at the same time.

Bless your naivety, but please spare us from it.

[Kofi 0]

I have to agree with Snave in this case.

Bottom line, quick and dirty explanation is that FSX with SP1 takes advantage of multi-cores but it isn't a pure multi-core application.

[Robert S 75]

Kofi I agree with your assesment and that of Snaves...but that was not the point of this thread! I never argued with him about HOW multiple cores were being used but rather that using more than one core provides benefits.

Kofi, you say that you agree with Snave and wrote: "FSX with SP1 takes advantage of multi-cores..."

Snave says: "THAT, my dear boy, is a LONG way short of stating that FSX takes advantage of multi-core processes. It does not..."

Remember, this discussion began with me disagreeing with Snave when he wrote the following advice: "There are only advantages to having FSX on a dedicated core, and routing all other applications to another".

I simply disagreed with THAT particular statement and pointed out that in addition to FSX now using more than 1 CPU core, there were additional FPS increases as well.

Snave then went on to reply: "Any fps gain is small, the loading time reduction only of consequence at startup or re-set"

I pointed out that both of these statements can be inaccurate, as a minimal FPS gain of 20% and up to 40% should not be considered small and my using more than 1 core there are additional improvements beyond just startup or re-set.

Besides the fact that Snave seems to be completely dodging his original advice of running FSX on a single core (the inverse of this logic: there are no advantages to using more than one core) he is now employing a clever deflection by arguing that FSX is not TRULY using multi-core but rather it simply: "Farmed out some of the processes onto the second core..."

Actually, he is proving my point (and I have forgotten to thank you, Snave for doing so) - Since SP1 FSX now uses more than one core (while that is not EXCLUSIVELY responsibile for all FPS performance improvement) it certainly will provide a user with a multi-core CPU MORE improvement than a single core user.

Snave, please quit trying to educate everyone on the difference between TRUE multi-core utilization and simply farming out processes onto other cores...I never argued that difference so I can't seem to understand why you are so keen on harping on about it. My original opinion still stands: FSX will see additional improvements (above & beyond other tweaks in SP1) by using more than one core...

Furthermore, the more cores you have the better the gain. I also never stated that 4 cores would double the performance of 2 cores, but the more processes you can farm out across multiple cores the better.

Here is a video of Phil Taylor explaining how this works by demonstrating FSX running on an 8 core machine:

So, Snave, I'll simply directly ask you after all of this: regardless of how cores are utilized by FSX (not the original point of this thread btw...) would you still stand by this statement:

"There are only advantages to having FSX on a dedicated core"?

[Snave 466]

I will answer equally simply: No, unless you have other processes that can be run on the others...

SP1 provides the performance gain from utilising code refinements, NOT adding true multi-core compatiblity. In so doing, ACES provided a way to achieve a small measure of improvement in the apparent usage of the second, or further cores. Anyone who thinks a 20% improvement shows multi-core compliance is not playing with a full deck. That's a woeful increase in performance for the use of a complete, separate chip of equal value in performance terms to the first. One core gets 100% of the performance. TWO cores gets 120% - and I'm supposed to be happy about multi-core `gains`..?

Throw the second chip away and buy a faster single chip, and even post-SP1 the gains are greater for a 25% increase in single core speed, than they are from adding a second, third, or eighth core. On that basis my conclusion stands, My advice stands, and your theory is nothing more than wishful thinking.

HOWEVER, if you add outside apps into the original core app - say by adding on an external ATC program to FSX, or running an off-game weather package, real-time texture updater or navigation program, then yes, multi-core is the way to go. But not because the SP1 offers increased multi-core compatibility, simply because the other items drag FS down less. A subtle - yet utterly vital - difference when considering upgrade paths.

[Robert S 75]

Snave, I think perhaps the key in our difference of opinions is in the definition of "performance improvements". You are quite clearly stating that improvements from SP1 are the result of code refinements, and NOT adding true multi-core compatibility.

I have been saying all along that multi core usage will yield benefits. Let's take one more look at the following quote from Phil Taylor:

With all that said, the Draw and SetTexture API call reductions and Autogen size reductions are probably as important for FPS improvements; the multi-core work really shines for load balancing and reducing stutters and blurries

Examining this statement, it would seem that the first half certainly supports your point of view. However, FPS is not the only consideration for improving the flight simming experience; obviously in flight blurries and stuttering contributed to people's dissatisfaction with FSX. What Phil is saying is that multicore usage reduces THOSE annoyances. It also seems to suggest that those problems would remain without multicore usage...also, we are leaving out load times, which you did in fact refer to. Those will be improved as well, but not by code refinements but rather by the multi-core usage capabilities introduced in SP1.

So based on the whole picture, my theory is much more than wishful thinking...so you may buy a faster single core processor which will increase FPS but you will STILL have longer load times when starting, blurries and stuttering when flying.

[Snave 466]

And again, you include a blanket statement that is incorrect, just as I thought you were beginning to see the point I was making.

Double the chip speed on a single-core processor, add more RAM, and you get GREATER increase in fps, and a GREATER decrease in load times, than from multi-core use. Yes, even after SP1. From a financial perspective, that may be a far cheaper upgrade path for a simmer who understands the criterion for selection.

`Load balancing` really is nothing more than PR bullshine for spreading the jam more thinly, not adding more jam. As I have said all along, you just have to read very carefully what is being said - and what is NOT being said.

The educated consumer needs to balance the cost:value equation based on their personal performance requirements. That is why being a smart consumer is so important. And why listening to this kind of blanket advice is utter drivel. FSX requires the fastest processor with the greatest throughput and a shed load of RAM to manage it all. While a multi-core processor may be one solution to the question of distributed computing, it is NOT a true multiple-core application that it is supporting, so any gains are diminished by the untainted fact that FSX is not, never has been, and will never be in this incarnation, a multiple-core process application...

So although Intel would like to sell you a multiple-core CPU, on the promise of increased performance of your favourite sim, (a fact which is, of course, inevitably although only vaguely true...) and MS, their trading partner would readily concur, it's simple: Depending on your individual use of the sim, a faster SINGLE core chip may actually be much faster. Add more RAM (assuming your OS will accept it) and the gains are greater than multi-core CPU's.

And finally:

Of course, this applies solely in the simplistic environment that one wants a computer to fly FS, and FS alone.

Should the computer be used for other things than just flight simming using the ACES product, then there can be considerable gains in a multitasking environment to using a multi-core processor. What ACES have done is facilitate the routing of individual FS processes to avoid congestion, and that results in worthwhile gains in the situations I describe above. Buit for the simmer using a virgin install of FSX SP1 with no outside apps, faster, better, quicker results are still obtained from a faster single core, not from multiple core CPUs.

It is futile to continue this discussion as I have proved the answer, and I do not expect this to change during this sims lifespan. I will go further and state unequivocally this will not change during this sims lifespan! By all means do your own experiments and prove it to your own satisfaction.

However, the ground-up redesign of FSXI, (which it had better be!), should address these issues and allow FS to follow more closely the current developments in computing. ACES have admitted on numerous occasions, and with open and honest candour, that they got it wrong with FSX. They have attended to what they can within the framework of what they have to deal with in the FS engine, but FSX has now become a meaningless `gap year` product (like FS2002) that doesn't offer retro-compatibilty, and yet doesn't take sufficient advantage of new hardware and software techniques.

The lesson learned from SP2/Acceleration is clearly that they cannot `fix` anything without `breaking` something else... the core has gone as far as it can go.

Time for a new one - and therefore the need to spend NOW on hardware to run FSX is pretty much meaningless and worthless. Vista is unnecessary. DX10 is no use, all those expensive graphics cards are a waste of money, multiple core processors are pretty much a waste, and no matter what you throw at it, FSX is a cow dressed up as a racehorse. And with no details on FSXI and Vista still the biggest lemon ins MS history NOW is not the time to be throwing good money after bad. IF ONE IS TALKING ABOUT FSX IN ISOLATION.

[Kofi 0]

Nevermind...

[Mo 0]

I totally agree with Simon's statement. I have got a single core (3.4) and I am running the FSX pretty reasonable and the FS9 runs absolutely fantastic with full setting.

Microsoft have not developed the FSX correctly and now they want to move to the FS11 and I am just wondering how the development of the third party are going to be able to catch up with their product.

Mo

[Infin8lives 3]

How about 256 Cores in FSXI? http://fr.youtube.com/watch?v=5wKe3V1l6Dk Phil Taylor speaks.

[Aerosoft Team [Inactive Account] 51558]

I have to agree with Snave in this case.

Bottom line, quick and dirty explanation is that FSX with SP1 takes advantage of multi-cores but it isn't a pure multi-core application.

The proof is in the pudding... A quad core machine will run many tasks in FSX a heck of a lot faster then a single core machine. In fact it makes the difference between being able to run it and between not being able to run it.

Knowing some of the CPU processes of FSX I am not sure how to interpret 'pure multicore'. I see some processes being handled by one core while others do other things. Perhaps it is not 100% streamlined from the start of the development, but again, the proof is in the pudding, if you see the cores being used they are doing 'something' and lets assume they do something useful. If not somebody in the MS FS team is pretty good in writing code that keeps 3/4 of my machine busy doing nothing. If I assign those cores to other tasks (like FSMap... hint hint) however I see in the 60% drop of fps that the cores WERE working hard on getting me more frames.