does FS9 benifit from duo cores?

[Jure 36]

Full screen. I usually start in windowed mode simply to drag whatever sub-panels I choose to a secondary monitor on the main computer, then go to full screen, but that's got nothing to do with the secondary computer. It's a completely separate unit and FS add-ons on it communicate with the FS computer via WideFS. Regards,Jure

[monnie 1]

"I couldn't spell nurd a year ago and now I are one according to friends."I'm sorry, Nes...I just couldn't let this go. I probably have Kool-Aid overdose.Isn't it spelled, "N-E-R-D"?? I don't really know if it makes a difference becuase I've noticed that fooling aroud with this game and reading the forums has the side effect of teaching me more about computers than I ever thought I would know.And just so I don't get accused of not contributing to this thread, I have to agree that when I got my duo core I notice what for me a big difference In my performance, but I also have more RAM. Is it possible that more RAM is just as responsible? I don't know.Best regardsJan1,KINDWhen I push the button and it works, I'm happy:-)HP Media CenterPentium D 2.80 GHz Duel core2.00 GB RAMNVIDIA GeForce 6200SEConexant Falcon II NTSCXP SP2

[Guest over and out]

That's very cool

[wrenchca 0]

To say that it isn't being helped directly by the second CPU is a bit misleading. FS9 on my Core 2 Duo rig shows both cores running about 50-55%. So both cores are working together to run it.FSX shows one core at max and the second at less than 10% so really, FSX isn't really helped by the second CPU.

[Bob Scott 7,714]

>To say that it isn't being helped directly by the second CPU>is a bit misleading. FS9 on my Core 2 Duo rig shows both>cores running about 50-55%. So both cores are working>together to run it.>>FSX shows one core at max and the second at less than 10% so>really, FSX isn't really helped by the second CPU.No, not misleading at all. What you are seeing is a single FS9 thread being swapped back and forth very rapidly between the two cores. But only one core is running the FS9 thread at a time, so no, the second core isn't helping run FS9 directly. In fact the overhead of doing that thread swapping is working against you.RegardsBob ScottATP IMEL Gulfstream II-III-IV-VSantiago de Chile

[wrenchca 0]

No, not misleading at all. What you are seeing is a single FS9 thread being swapped back and forth very rapidly between the two cores. But only one core is running the FS9 thread at a time, so no, the second core isn't helping run FS9 directly. In fact the overhead of doing that thread swapping is working against you.

I find that hard to believe personally. Just went into FS9 and checked my performance tab. Both cores show around 55%. Over to processes and set the affinity so that the second core is not being used and the first core usage went to about 98%. If they were swapping threads back and forth, neither one would be using more than the original 55% so a single core shouldn't be maxed if it tries to handle the entire process load.It's a good theory Bob, but in practice, it doesn't hold water very well.:-beerchug

[Guest TomOOO]

HiYes, FS9 and FSX can benefit from multi core due to the following...FS9 and FSX are not designed explicitly to use multi core using"fine grain" parallel code - as you note. They can however usethe resources of a multi-core system since they are complex programswhich require many resources. 1) The require compute such a scene 3D calculations, collision calculations etc 2) The programs require lots of data of disk, so need to carryout data fetch from disk and RAM. 3) need to send data to GPU 4) On board sound - oops...These resources can be seen as different from the OS, so the OScan implement course grain parallel methods.Data fetch in particular can result in a block to the CPU, socan data transfer to the GPU. So the different processors cando the data management, while calculating. So if you see both processors working over 100% then you are gettingmore than 100% of one processor. I get up to 180% in some cases.If the process was swapping between CPU then you would get less than100% as time slicing would result in wasted time.As for using multiple computers - there is no problem with thisas long as what you want on the "secondary" computer does not needto send to much data across the network - of you will end up witha wait for data. Ie the data transfer rate through a networkis really slow compared to anything within a single computer.I use a second computer (really old laptop) to view maps andlearn the instrumentation while flying. I have not tried anything else more integrated. RegardsTom

[Bob Scott 7,714]

Yeah, I just write advanced panels without knowing anything about operating system behavior and performance. I'll go throw my university diploma in Computer Science away as soon as I finish posting this... Too funny.Two CPU cores running at ~50% is equivalent to a single core at 100%. One half plus one half equals a whole, if I remember that lesson from 2nd grade correctly. In one case the single FS9 thread runs for a few microseconds on one CPU, then the thread is swapped by the WinXP executive to the other CPU, then back etc etc. While the FS9 thread runs on one CPU, the other is near idle. The net effect is both CPUs showing an average load of ~50%, which is approximately equivalent to one CPU running the thread at 100%. In the second case, the FS9 thread is constrained to run on a single CPU, keeping its usage up close to 100%, with the other near 0%. Of course there is some overhead added in the process of swapping the thread between CPUs plus the load of services running distributed across both, so the two cores show slightly above 50% when the process is time-shared across CPUs.Sorry you can't believe it, but that's what's happening.RegardsBob ScottATP IMEL Gulfstream II-III-IV-VSantiago de Chile

[Guest TomOOO]

Be careful when you present your qualifications, you never knowwho you are flaming :-)I tried to avoid being technical. Although FS9 and FSX are not parallel code which means that thecode is not specifically designed to run the same function (letssay collision detection), each of the "background" tasks...generate and place terrainload terrain, mesh, scenery addonetcare done as threads. Many computer languages support threads, andall interactive programs must use threading. How do we know - wellyou are flying in FS9, and you want to pick a control on that niceVC. You must do the picking in a thread as the flight sim is busyflying your plane. If not implement in a thread then you could notinteract with the program while it was running. Since FS9 and FSX are threaded, then it is up to the OS to decide where to run the threads, and it you have hyperthreading + multi-corethen the OS has a choice. Thread blocking is a common problem withhigh data throughput - so even hyperthreading can help on a singleCPU. This is known as coarse-grain parallelisation which isdifferent from fine-grain prallelisation that can be done if thecode is specifically written for multi-core.You will not get 200% on 2 cores and 400% on 4 cores (where I definea base line of 100% as a single core), since any one process runninga number of thread will have to wait for the slowest thread to complete before doing the next task as it may suffer a datacoherence/integrity problem (look it up). Ie we are into the world ofload balancing. This is why FS9 will swap between cores since theprogram is likely to have been written in modules with many threadsstarting and stopping. Where any one thread ends up is completelyundefined and depends the instantaneous load balance and how the OSpartitions these thread. How you set your sliders for quality, FSB speed,memory latency, F1/F2 bus speed/size and all sorts of thingseffect the speed up of multi core; then there is the OS and howwell it is written - has aggressive branch detection in the predictivecode bla- bla. So, I think that multi-core are useful for simulation (and all complex programs). How useful depends on many things you cannotcontrol, and some things you can control. In FS9 and FSX youcan play with the quality of view etc, sometimes you change somethingand you get a lot of bangs for you buck, sometimes it grinds to a halt and you get a slide show. It all comes down to tweaking yourFS9, setting the parameters in fs9.cfg, and seeing what works. Ihave tweaked my to about 180% over a single CPU. No more from me on the subject, take it or leave it, and yesI do know what I am talking about, honest :DRegardstom

[virtualstuff 750]

>run FS9 at an outstanding level with full add-ons and high>frame rates, long before buying a new machine just to run FSX>on its low end settings?>In my case it does all settings to the right traffic 100%and all other programs such as ASV on the second cpu :-)FS9 looked never so great as now :-) Andr

[Bob Scott 7,714]

A lot of correct details in here, Tom, but not at all organized to support the idea that a multicore CPU directly enhances FS9's performance.Yes, things like FSB speed, memory latency, F1/F2 bus speed/size, and OS design affect multi-core speed, but they affect speed on all other architectures as well. Not sure what the point was.FS9 can and does swap between CPUs...not just because the OS is allocating the various threads being opened and closed by the primary FS thread, but because WinXP can migrate the primary thread itself between CPUs. According to MS, "In Windows 2000 and later, a thread in a process can migrate from processor to processor, with each migration reloading the processor cache. Under heavy system loads, specifying which processor should run a specific thread can improve performance by reducing the number of times the processor cache is reloaded."(See http://msdn2.microsoft.com/en-us/library/s...oraffinity.aspx)I haven't yet found a combination of settings in FS itself that will increase utilization of dual CPUs above around the 50-55% mark. At the heart of FS is a 18Hz master process thread that can only run on one CPU at a time, and everything else certainly appears to be tied to that.RegardsBob ScottATP IMEL Gulfstream II-III-IV-VSantiago de Chile

[Bob Scott 7,714]

>As for using multiple computers - there is no problem with>this>as long as what you want on the "secondary" computer does not>need>to send to much data across the network - of you will end up>with>a wait for data. Ie the data transfer rate through a network>is really slow compared to anything within a single computer.WideFS keeps a buffer of the last data received from FS9, so there is always local data available without waiting for the network...management of latency becomes the issue there. Most of the data passed via WideFS is state date and really is low bandwidth.Use of multiple computers can transfer a significant amount of workload off the primary PC...an example being moving the ATC audio in Radar Contact to the slave PC. But the unused excess CPU capacity on a dual-core machine running FS9, which does not, in my experience use more than ~50% of the available CPU cycles (see discussion earlier in this discussion thread), now can take care of lots (in my case all) of that workload without the complexity of a second linked PC. I now find myself using my second PC to show approach plates, watch the ATC environment when using VATSIM, etc.RegardsBob ScottATP IMEL Gulfstream II-III-IV-VSantiago de Chile

[wrenchca 0]

Yeah, I just write advanced panels without knowing anything about operating system behavior and performance. I'll go throw my university diploma in Computer Science away as soon as I finish posting this... Too funny.Glad you are amused. Being just another dumb simmer, I am not privy to programming knowledge such as yourself. I can only look at what I can see with my own eyes and what I can actually prove out. If my logic is flawed...my bad.For what it is worth FS9 runs better on both cores than just one on my machine. And that, I can see, feel and experience with my own senses...I do not need a degree in programming.

[Guest over and out]

So now I guess I