Picking The Right Cpu For Fsx

[Sharrow 0]

Not to mention that with a quad one also has the possibility of setting affinity to specific cores (core 4 for example) for addons which run alongside FSX, thereby keeping the addon well away from the business end of things as it were. If you run windowed mode on multi-monitor set ups you can achieve the same effect with other apps you may be running alongside FSX. All sweet goodness, no?Sorry, will shut up now and let you all wake up over there and have a chance to respond... :(

[TechguyMaxC 516]

By stressing the core we are talking 95%+ utilization, more or less?

Depends on the application. Most "stress-tests" run the CPU as close to 100% as possible so I'd say that's really stressing the core. Most applications are designed to take as many clock cycles as they can, if they are computationally-heavy. In this case though it appears FSX is basically just constantly issuing memory requests on the extra cores, so while they are kept busy, they're not contributing significantly to performance because they're not performing any computational work.

I am not saying or implying that loading textures (and all the rest Nick mentioned) is going to stress any given core, what I am saying is that all this offloading to core 3 & 4 simply provides core 1 & 2 with more available cycles to get on with the really important bits. Hence the "rule of thumb" 500MHz+ extra you need on a dual core to make up for the loss of 2 cores...

It's true that offloading work to spare cores helps free up cycles on the busier cores, hence the extra clockspeed (i.e. extra cycles) needed to make up for this fact.

I have read on many, many occasions of high end FSX users stressing their dual cores on a regular basis and then moving to quad and experiencing significantly lower CPU utlization across their 4 cores. This speaks for itself, does it not? Am I still missing the forest for the trees here?

Well if you double the number of cores and run the same workload, of course your CPU utilization will decrease as a percent. This doesn't imply any increase in performance though. In fact, an increase in performance would likely be noted under similar utilization numbers after doubling the cores as this would mean 2 cores was not enough to handle the workload in the same time as 4 cores. Bottom-line: the extra cores are useful, but not for computational work, meaning they will not significantly increase frame rates (again, except during load times).

Not to mention that with a quad one also has the possibility of setting affinity to specific cores (core 4 for example) for addons which run alongside FSX, thereby keeping the addon well away from the business end of things as it were. If you run windowed mode on multi-monitor set ups you can achieve the same effect with other apps you may be running alongside FSX. All sweet goodness, no?Sorry, will shut up now and let you all wake up over there and have a chance to respond... :(

That's the point though - you have to find extra work for your extra cores, FSX won't do it for you automatically.

[Guest Nick_N]

frame rates after a smooth 24-30 is not a concern for me. Scenery, clarity and smooth perf, is

[TechguyMaxC 516]

frame rates after a smooth 24-30 is not a concern for me. Scenery, clarity and smooth perf, is

Certainly. Still have to get there though. 24-30 fps is easier said than done particularly in urban environments with heavy traffic and autogen.In other words: there is a use for "extra" CPU horsepower which delivers "extra" frames, and that is in scenarios where those "extra" frames disappear, such as that I just described.

[Guest Nick_N]

Certainly. Still have to get there though. 24-30 fps is easier said than done particularly in urban environments with heavy traffic and autogen.In other words: there is a use for "extra" CPU horsepower which delivers "extra" frames, and that is in scenarios where those "extra" frames disappear, such as that I just described.

I dont seem to have that FPS drop problem hereFrames can not be used to judge perf on any linear scale,... that approach don't work with FSXtoo many variables in play in free flightOne can use frame rates as 'part' of an indicator if, and only if, the flight is 100% controlled with a flight file which can be repeated and a compare can be gauged against a hardware change/driver change/settings changehowever the FR is only part of the result in that quantified examinationotherwise FR are just about out of the loop for use in proper examination of a result in FSX

[ryanbatc 12,836]

Sure but a E8600 will get you there faster than a quad will...Say FPS don't matter, ok, but there will be a level at which the user sets autogen/scenery/etc options at. I guess it comes down to, would you rather have a low framerate (15-20 FPS) in large cities but no stutters, or 24-30FPS in the same area with a few microstutters? I'd take the 24-30, especially once you add in payware planes like the PMDG/Level D stuff. And btw, 3.33ghz -2.83 ghz = 500 Mhz

[TechguyMaxC 516]

I dont seem to have that FPS drop problem hereFrames can not be used to judge perf on any linear scale,... that approach don't work with FSXtoo many variables in play in free flightOne can use frame rates as 'part' of an indicator if, and only if, the flight is 100% controlled with a flight file which can be repeated and a compare can be gauged against a hardware change/driver change/settings changehowever the FR is only part of the result in that quantified examinationotherwise FR are just about out of the loop for use in proper examination of a result in FSX

MSFS' FPS counter isn't real-time. Your entire argument ignores this crucial fact. The cause for these frames not being displayed is irrelevant, the result is a lack of perceived motion because frames aren't being rendered.