Arun,Yes you can replace your 8500 with your 7950 and still have DX9 work. But before you do, please realise that you can run the 8500 in DX9 mode and claw back that 30% FPS loss and that your system may infact may behave differntly than mine and not cause that 30% loss in the first place.To answer all these questions, just run the benchmark on your system in each configuration (eg. DX9 mode, DX10 mode, with 8500, with 7950) and the results will speak for themselves.Gary

Gary,Thanks! Don't worry about the delayed reply, instead of waiting until Feb-March, I have settled with a Q6600, hopefully I will still get good performance (without SP2 for the time being).

Sure thing... Yes I was very grateful that the ACES team did the post-RTM work on SMP. It breathed a little more useful life into this system.I didn't post the results before because I thought no one would have any interest in a system built in the summer of '06. Fwiw am running Win XP x64 on a Tyan S2895/4 GB DDR 400 ECC/ATI X1800/2 WD Caviars 160s w/mb RAID 0/1680x1050x32 full screen AA-4X & AF-8X.Gary (RESET MCP ALT) had to have put alot of thought and effort into this; so Gary, thanks for the work.Re: his post on 08Dec07, his conclusions are very interesting if not intriguing (e.g. setting the cores to other than 4 in a 4 cores system) and, fwiw, I see the recommendations as being "spot on". Seems there have been many posts on "what's best?"... would be nice to see his post pinned. :-)Rob O.

Hello again Gary,I had a few questions. First off, does this mean that FSX will run faster on just 2 cores rather than all 4? So would a Core 2 Duo just have sufficed?Thanks for your time!BoeingGuy

BoeingGuy,I don't think it's that simple, because when FSX is limited to only use 2 out of 4 cores the other 2 cores are being used by the OS and other apps. A standard dual core setup does not have this luxury, so those 2 cores are supporting FSX, the OS and other apps which must mean lower performance than the quad core setup I just mentioned. What that actual difference is I have not measured yet (nor can I for a while as my desktop system is currently in a shipping container bound for Sydney Australia), but I'd say it's in the 2-5% region so a dual core setup is still pretty potent.Gary

yeah it was pinned, but I don't know why it was unpinned...Dear Mod, please pin, Thanks!

just a quick question re performance ... i'm currently considering doing an upgrade of my AGP based machine to PCIe and have received a quote from my local dealer for some bits that'll kick performance issues into touch once and for all: Asus 8500GT video card/Intel dual-core cpu/appropriate motherboard, RAM and PSU ... i notice occasional mention in this thread of TH2Go system which i use ... is there any point in doing my upgrade if the most that can be delivered to my monitors is the native 3840X1024X32 resolution the foregoing delivers? or to put it another way could upgrading my cpu alone move me from the marginal territory i'm in at the moment as regards performance? ... AlanCurrent setup: Intel P4 3.4 with HyperThreading Asustec Oxford Mobo geforce 7600GS AGP 2 Gig RAM

OK...what's really scarey about this post is that it is very misleading.How are you defining "performance" ?You state at the beginning - in a footnote - that a quad core does reduce loading times for textures.Does this mean that "stuttering" is reduced...and possibly "blurries" ?Of course many people only fly from the cockpit and this would not be an issue...but some fly from the outside and do a lot of panning around; in which case, stuttering is very much a performance issue.For me, reduced stuttering - aka: faster texture loading - is akin to doubling the RAM from 1GB to 2GB (something I did some time ago)and it made a MASSIVE difference. Of course measuring "blurries" and "smoothness" is very subjective - unlike frame rates - but overall (and this is important)an article about performance, particularly when people are making (possibly erroneous) conclusions such as: "it is not worth upgrading my hardware" etc., etc., should really define its terms.I'd like to bet - and I'll find out for myself shortly - that those extra cores are - when it matters - very useful to overall "performance", certainly when you define performance as a multi-faceted concept. To reduce stuttering and/or reduce blurries is, arguably, more important than an extra 4 or 5 frames per second.FPS can easily be improved by reducing sliders a little here and there; unfortunately, inadequate loading times leading to blurries and stutters is not something that is easy to cure, unless "once a minute, the cores all hit 100% as textures are loaded" and that "loading time is inversely proportionate to the number of cores"...to paraphrase your own words.So your "footnote" ironically, holds the key, and suggests to me that the conclusion should be that a quad core will help massively with performance, WHEN you consider that there is a lot more to performance than flippin' frame rates. Unfortunately most people appear to be drawing a different conclusion !I will report back as soon as I do my own entirely subjective tests on a quad-core setup...but I am extremely sensitive to smoothness - or rather lack of it - my eye picks up every little jitter and any improvement will be noticeable. If textures are loading more quickly it will be clear and to me, worth the entry price of a quad core regardless of the effect on frame rates.Regards.

Hi I have a Intel Quad Core 6700 with gb ram and VistaI don't what you are upgrading from, be in single or duel core or duo.But I went from Duo core cpu to a quad last April, didn't help at lot until I downloaded a hotfix for Vista and also the SP1 for FSX also help as part of that SP was to help FSX use mutiple cpu'sAnd the hotfix for Vista also help Vista utlerlize mutiple core cpu's overall, not just for FSX.I also like to pan around and view from outside, look at AI traffic etc and I can say from my experience I do not get any stutters and I have settings pretty much maxed out.Of course since SP2 came on the scene, the decison to run the DXT10 preview came about, if you have a DXT10 capable card, use the preview, as there is a noticable difference in performance.CheersPhil M

Did you know that we can turn cores on and off "on the fly." Try it. Set Get a flight stabilized over known terrain. Try a jet with the A/P engaged at 200knots and 5K'. That will provide a stable platform that can stress FSX's terrain loading capability. Turn AI and autogen totally off. We want to look only at terrain loading capability. Get the frame counter going. Now pop up task manger and watch how your cores are responding. If you have a single monitor, get into window's mode so you can watch both FS and TM (or zoom down to BB and grab another monitor. It's a great excuse!) Go outside and AlSo get a feel for how well the terrain resolution is keeping up with the airplane. You will likely notice terrain blocks loading through several resolution increments in some proximity to the airplane. No need for a statistical analysis. Just note FPS and the character of how the terrain is loading. If you have a quad core, shut one core off, then 2, then 3. Task Manager > Processes > Rt Click fsx.exe > Set Affinity > Check and un-check cores as desired.Now back into the game. How is the terrain loading responding? My experience is that my 4 cores (Q6600 @ 3.6 / 4g-O-ram / 8800GT@1920 x 1080 res) can just about (but not quite) keep up with max settings at 200knts and 5K'. Core 0 is always maxed out and the rest of the gang are always rocking in and out of 100%. As I shut down cores my terrain turns into a blurry mess. . . . . but FPS remains unchanged. As I re-enable cores, the terrain resolution catches back up . . . and still, FPS are unaffected. Try the same drill with autogen and AI both totally off, then full blast. The extra 3 cores of my quad made no difference. The FPS did not change between a single core v quad core with full autogen and AI enabled, nor did the autogen visuals change.It really appears the reason those 2 v 4 core benchmarks are similar is because only OnE core is delivering FPS performance. I expect the same test suite would display similar FPS performance with a single core.The quad most certainly provides an richer game experience, it's just NoT related to FPS. It appears that FSX's Frame per Second performance is based on a Single core. Even a dual core won't help, much. It's no wonder that a quad provided little benefit by use of this measurement metric. (imHo) A singular FPS metric provides an incomplete analysis.

Thanks for a very well written post and for sharing this interesting information!

Yes I agree, very useful infoI was aware and have done it myself, that you can set a line in your FSX.cfg affinity mask the number of cores.First read this tweak I think from Aces at one of their blogs.CheersPhil M

One thing that clearly has changed from FS9 -> FSX is the importance of FPS where you see in FSX that the actual FPS value isn't that important anymore for the all-around feeling of performance in the simulator and I have read in other places lots of people have noticed the very same thing.

Does anyone want me to do some benchmarks when i get my new 9800GTX?

WOW...that's one impressive rig you got there!! What kind of FPS do you get with that one and I presume you're more than able to have all settings to the far right ;) ?