Thanks guys!SD

These findings do not make any sense as the devs have all been saying that dual core is utilised.What's going on?

I've tried it on both single and dual core and found that on single it runs 95-100% CPU utilisation and on dual around 55-60% of both cores combined (which equals 110-120% in single core equivalence), so dual core does seems to make a difference.To further qualify this, I set up a recording in FSX where I take off from TNCM and do a quick circuit to land, encompassing as much visual load as I can, then ran FRAPS for a 2 min average FPS count on both a single and dual core A64 in exactly the same system. The single core (running at 2.35GHz) pulled 20 FPS average whereas the dual (running at 2.4GHz) pulled 26 FPS. Scaling the cores back to the same 2.35GHz speed, thats 20 and 25.5 FPS respectively, giving a 27.5% performance advantage to the dual core, which roughly equates to the CPU utilisation figures I saw above.Gary

Gary -- Thanks for the work. This kind of hard data is what we see too little of. I wish there was more information like this for flight simulation and hardware-related information out there.RH

Gary,Thats great news.....well done that man!Some hard facts and figures at last.

Thanks this is excellent research. We needed a person with both dual and single core CPUs to come up with this kind of data. So the bottom line is that dual core does perform better but will not use 100% CPU on both cores.Rgds,JCMK

BUMP,As there seems to be a lot of questions on the 'issues' thread regarding dual core utilizationGlenn

As is said above, the primary issue is the type of application, with so much of the render loop in this case being single threaded, because it has to be.But I like to think of dual core as this: it's not what it adds to a single application, it's the added capacity to run concurrent threads in mostly.Benefits from multiple core on largely single threaded applications come in with things not related directly to the application: example, Nvidia indicates their SLI drivers are multi-threaded and benefit from multi-core processors, add-on programs like ActiveSky can run on the same machine as FS using the other core, and of course, a number of operating system background services at the OS level such as network, etc...This ability to run more than one thread in parallel will not have the amazing doubling of performance that can be expected, but it probably has a perceptible impact by allowing more processor time to the application itself.EM

>I tested (finally ;-)). in my system : >>Dual xeon at 3.0 and confirm all responses ... :-(.>>Only one process (fsx) has launched by simulator... no>advantatge on systems with multiple processors (or dual cores)>. That only one process is launched makes sense (fsx.exe), the really important thing is how many threads that process utilizes, those will make use of additional processors. And, above all, how well those threads balance the work among themselves. I would assume that FSX is not so good at the balancing part from what we read up to nowHappy flying!Cristian.

From what I read the dual-core threading is mainly handled by the OS supported by 3rd party driver, so maybe Vista will show some improvements (or newer 3rd party drivers from Intel/AMD). Like any new technology, the details seem to be a bit fuzzy. Oblivion, for example, uses almost 90% of both of my cores on a AMD4800X2 and it's just a single executable as well.Pat

[...]>There's nothing magical about multi-core that will>ensure all your cores are maxed out all the time, especially>an application that has a hard constraint on a foreground>render loop. Only so much stuff can be scheduled and offloaded>to background threads before they have to yield to allow time>for the main process. Flight Sim is not a web server or>database.Sounds good assuming that the foreground render loop cannot be split into independent tasks. But even then you could do something like "task one, you render now, task 2 you go on and start working on the next frame" and tickle the magic out of the hardware :-)Anyhow, the old wisdom that software lags many generations behind hardware is there for a reason, so I guess we'll have to wait for XI or even XII...