I have been using multiple processors (Xeons) for many years. They deliver nice, responsive performance particularly for machines running multiple applications because you can switch between them snappily. You also get better performance from certain applications: some of the tasks in Photoshop, for example, benefit greatly from multiple cores. Just one extra core gives much more headroom for background services to run without interrupting your applications. A system with 2 cores therefore makes good sense. 4 cores is still a luxury.Thus, extra cores let your computer handle more work at the same time without getting bogged down. This is the sense in which extra cores deliver better "performance". But generally, extra cores do not increase the overall "speed" at which any particular application executes. The brute reality is that faster performance for consumer applications like FSX comes from increasing the clock speed. It is true that in theory, you can programme an application to work faster by breaking it down into different tasks and giving each a separate thread, to be handled by different cores. But in practice, implementing this technique effectively is very labour intensive for the programmers and thus expensive for the developers. Very few applications (even Photoshop) make full use of it. At best, they tend just to dip into multi-threading for particular chores which can easily be farmed out to a separate thread. FSX is certainly guilty of this tendency: it only uses multi-threading when loading textures. This undoubtedly makes an appreciable difference, but it does not produce a constant increase in framerates. To take advantage of multiple cores in a way which constantly delivers significant extra framerates for FSX, it would be necessary to programme separate threads for (for example) ATC, weather, updating the instruments, receiving user input, modelling the aerodynamics, plotting distant graphics, plotting near graphics, etc. The output of each thread must then be synchronised. The complexities of such a task are so enormous that we will probably never see MS pull it off.I am sorry to say that in my opinion the multiplication of cores within processors is really just a marketing gimmick, which has (successfuly) distracted consumers from the disappointing clock speeds of Intel's processors. Intel found that the heat generated by its Pentiums prevented it from increasing clock speeds beyond about 3.6/3.8GHz; the Core 2 technology has injected a new lease of life into Intel's products by, essentially, squeezing out better performance per clock cycle; but the bottom line is that Intel simply ran into a brick wall with clock speeds. This meant it had to come up with some other way of persuading us to keep on buying new processors. So it started bundling multiple cores onto a single chip, presenting these as if they were a new way of gaining performance. For most consumers, extra "performance" simply means extra "speed", which as I have tried to explain is not something multicore processors can deliver without expensive re-programming of existing applications. The industry as a whole has done little to correct this false impression, as the misunderstandings evident in this thread (and other threads here and on other forums) continually reveal. I cringe when I read of some of the money wasted by people who bought machines apparently thinking that 4 cores instead of 2 would double the speed of particular applications.In my view, Intel has really just been playing for time while working out how to start increasing clock speeds again. With its manufacturing process now turning out more heat-friendly processors, we are at last (I hope) beginning to see a resumption of the journey towards higher clock speeds. It will be interesting to see whether anyone regards it as worthwhile to re-write complex software to use multi-threading effectively. I doubt it, at least for the foreseeable future: I think it's more likely that the forthcoming generations of processors will work at fast enough clock speeds to relieve programmers from the burden of trying to split their applications into multiple threads - at least for the foreseeable future.That's a long-winded way of saying that if you want a computer with 8 cores, make sure it isn't because you expect that FSX will perform significantly faster than on a computer with 2 cores of the same generation and clockspeed: you will get slightly faster overall performance, but it will not be anywhere near 4 times as fast. Tim

>Inbound to LAX on (any) STAR at 210 knots in the PMDG 744,>Megascenery has my 4 cores ConStantLy into their 100% stop. If>I want to show off my scenery to a passenger (at megascenery's>max resolution), I still have to pause to let it catch up. I>NeeD more cores. >>I expect if you fed it, it would eat.What is your system configuration?

>Do you have a link?>All I could find on YouTube was Phil Taylor the Dart Champion>:)No...I did not to think to bookmark it. This is certainly not to start a MS bash but I was amazed that a game developer knew FSX required that kind of hardware...stuff that is not available.....and market a product. FS11....if there is one....will be on the market way before those kind of hardware specs ever come out. I use FSX on a P4 and am reasonably happy with it. In the winter I am flying 3 to 8 hours a day. I know I need to have much better hardware to run it but those specs are way over the top.

I found the link.http://www.flightsimx.co.uk/flightsimulato...ulator-x-video/

All I can say is I have a quad 9650 running @ 4ghz. On my system FSX deffinately uses all 4 cores and they all peak to 100% approx once a minute. (I think when FSX relights the scenery)I wish people would stop saying FSX doesn't benefit from a quad, because it's not true.You might not see higher PEAK frame rates over a dual, and if you do it will problably be small. (because FSX's main thread still only runs on the first core). BUT, the quad will provide higher average and minimum framerates and faster texture loading. And you should be able to run a bit more ai. But if you go hog wild with your settings you will still bog down the first core and cause FSX to stutter or run slowly.

Thanks!But why does FSX need to load ALL 64 world's terrain tiles, as PT says in that video (~2:30)? What is the benefit of loading all global terrain tiles into memory? I thought multi-core is about detail, not the coverage -- I'm puzzled. At the same time he says, AI and physics are not yet scalable, which hopefully is going to be addressed in the vNext. But at least, if the main core is freed from the terrain load, this should lead to a big boost in performance.

I hope that MS will be able to "pull it off" (you mention they may never be able to farm out other processes such as AI to other cores).I am confident that they can, and will. The platform strategy (ESP) seems to be a logical move in that direction...RhettAMD 3700+ (@2585 mhz), eVGA 7800GT 256 (Guru3D 93.71), ASUS A8N-E, PC Power 510 SLI, 2gb Corsair XMS 3-3-3-8 (1T), WD 150 gig 10000rpm Raptor, WD 250gig 7200rpm SATA2, Seagate 120gb 5400 rpm external HD, CoolerMaster Praetorian

Ref "What is your system configuration?"P5K-e/[email protected]/4Gram/8800GT@700Mhz core/42"@1920x1080/3ea Seagate .11. Raid 0'd w/280MB/s sustained xfer & 6ms access. . . . and it wants more.

Tim,Basically what you are saying is: there's a physical deadlock in CPU clock speeds per core. I think it should be an EXCELLENT reason for MS to catch up in the area of high-performance distributed computing, i.e., to make products like FSX truly scalable by rewriting them in MPI mode. Any scientific modeling nowadays is being done with MPI architecture in mind, because in HPC area they achieve higher speeds by increasing the number of CPUs, not the clock speeds. True, not that many interactive applications were successfully implemented for the cluster use but it is very much possible. And, as you were saying, the clock speeds won't get faster so MPI is the only way out to progress with higher performance simulation. Cheers,=S.V.=eMachines T5026/P4/3.07GHz/1Gb RAM/160Gb S-ATA HDD/Windows XP Home SP2/ATI RADEON 9250 PCI 256Mb/ViewSonic VX910 19' 1280x1024/Microsoft SideWinder Force Feedback 2

Nice depth on these screenies!Cheers,=S.V.=

>And, as you were saying, the clock speeds won't get faster so>MPI is the only way out to progress with higher performance>simulation. >You all are making good points. Maybe this is the same general line of discussion going on right now at ACES?RhettAMD 3700+ (@2585 mhz), eVGA 7800GT 256 (Guru3D 93.71), ASUS A8N-E, PC Power 510 SLI, 2gb Corsair XMS 3-3-3-8 (1T), WD 150 gig 10000rpm Raptor, WD 250gig 7200rpm SATA2, Seagate 120gb 5400 rpm external HD, CoolerMaster Praetorian

What I was saying, rather long-windedly, is that there HAS been a deadlock in clock speeds; but that - after a very long time (can it really be nearly 2 years since the 3GHz 5160 Xeon?) - Intel is now beginning to break through the deadlock; and that this may be a disincentive for developers to invest in doing multi-threading properly, at least for now. Let's hope I'm wrong, because if done properly I believe multi-threading could deliver a very nice performance boost to FSX/FS11/FS12.Also, of course, it remains to be seen how much further Intel can push clock speeds. I think Bill Gates (how would he know?) expects them to plateau at 6GHz some years from now. Perhaps MS will take the view that it might as well start building multi-threading into its applications now, so that when/if a "final" deadlock on clockspeeds is reached, its applications will be ready to maximise speed by using multiple cores instead of relying exclusively on clockspeed.Tim