Sorry !!!, yet another upgrade question.

[Tabs 2,790]

Yeah Ray is definitely right - I've had 3D accelerators since the very beginning with the 3dfx Voodoo Graphics chip and ALWAYS you've needed a fast CPU capable of pushing tons of triangles to the video card to get the most out of it. I could take my old Pentium II system, slap my GF3 in it and it's gonna run like pure trash, but stick it in a brand new P4 or Athlon XP system and it'll max the card out. The system is only as good as it's slowest component.This is why I'm planing to upgrade my whole system core arond Feb or March next year - I'll get the fastest CPU, new RAM (probably DDR400) and the new GeForce FX card - that should scream and hopefully FS will run extrememly smooth for me...

[Guest SD Sim]

Hey Paul,I haven't seen you in this neck of the woods for quite some time. I've always enjoyed your posts - good information supported by facts. Much of what I've gained in my personal flight sim optimization can be directly attributed to your fine work. I hope you'll be able to come around more often.:-wave

[Guest SD Sim]

Just don't forget that about one year from now, the next version of MSFS should hit the shelves. Then the viscious cycle of trying to build the optimal system for FS will probably repeat all over again. Of course you can bypass this by sticking with FS2K2. I have to admit that I've been pretty happy with it. But I just know that MS will eventually lure me over to the dark side :-lol

[Randall 82]

Don't forget the old test--lower your resolution and see what effect it has on framerates. If the effect is substantial, then your graphics card is the bottleneck. If not, the CPU/memory subsystem is.

[Ray Proudfoot 8,944]

Hi chaps,I just wanted to thank you all for confirming what I was saying last night was basically correct. I really had to control my temper at one stage.I hope all the comments help you Paul to reach the best decision for you.Regards,

[KERNEL32 500]

" QUOTE- Don't forget the old test--lower your resolution and see what effect it has on framerates. If the effect is substantial, then your graphics card is the bottleneck. If not, the CPU/memory subsystem is."That is very interesting and works just like you said. One comment I have and I would like to verify the validity of it-I have been having lockup problems not related to heat, drivers, O/S, sound, etc. (maddening really!)anyhow, I tried replacing my GF3 with a GF4 (not the fastest one really inexpensive one- not sure of the model) and it was SLOWER than the GF3! I couldn't run 1280x960 but at like 9 FPS, whereas I'm getting about 14 -18 FPS with the GF3! But when I switched the res on the GF4 to 1024x768 it had decent frame rates. Amazing. Brought the GF4 back as it still locked up.I fear it may be the processor itself, as I have tried everything in the book. But my question here is about the comparison of the GF3 and GF4 can that really be right?!

[Guest]

Tom - If the GF4 had an "MX" designation in the model number I'm not surprised. Those are basically reworked GF2's and are often (always?) slower than the GF3's.Trip

[KERNEL32 500]

Trip-I think it did as a matter of fact. [emabrrassed](-: Thanks

[Guest JonP01]

Trip,I guess that means then that I have to disagree with both of you. Either that or perhaps there is a misunderstanding as to what exactly I am objecting to with regard to Ray's comments.My objection was based upon these initial comments from Ray:"The change of processor represents a 50% increase in processing power so you will see a 50% improvement in frame rates. 10 becomes 15, 20 becomes 30 but 6 only becomes 9".As you can see, Ray has assumed a direct linear relationship between CPU clock speed and frame rates. This is specifically what I am disagreeing with. I do agree that the CPU is of principal importance to frame rates. But what I do NOT agree with is Ray's premise that increasing the CPU clock speed by 50% also increases their frame rates by 50%.Trip, you have refered me to some benchmark studies that prima facie confirm Ray is correct. Firstly with regard to Paul Leetzaw's findings. I am not sure why you have even refered me to them, as they only go to prove my point. Paul's findings are encapsulated in the statement (quoted directly from this thread also):"FS2k2 Performance is pretty close to 60/25/15%=cpu/memory-sub-system/video card dependant. (fully tested)"Paul does NOT say that FS2K2 is 100/0/0 cpu/memory-sub-system/video card dependant. In order for Ray's statement to be correct, the net effect of the memory-sub system and video card would have to be zero. Secondly, you refer me to FSBench. Again, this also confirms my point. And to save people looking it up, I've even summarised the results into an Excel table that I have attached below. In column one, I have classified the various CPU speeds, begining with 500 Mhz as the baseline, all the way up to 2 Ghz. In column two, I have indicated the percentage amount of improvement that would be required for the faster CPU to scale in a linear fashion - that is, to be consistent with Ray's original statement - e.g 50% more mhz = 50% more frame rates. For example, an 800 Mhz processor is a 60% improvement in clock speed compared to a 500 Mhz processor. Then, in column three, I have tabulated the top 100 results from the latest FsBench that I could find at Avsim:http://avsim.com/fsbench/2k1024.htmI have summarised the data and averaged out the percentage frame rate improvement from increasing clock speeds.As you can see from these results above the 500 mhz baseline, there is not one single result where the FS frame rate scales at the same rate as the CPU clock speed. So as far as FsBench is concerned, the results shown there are also inconsistent with Ray's original statement.So in summary, what I am saying is that, yes, the CPU is the most important ingredient, but the frame rate in FS is still influenced measurably by other sub-systems within the PC. That is why I contend that Ray's statement is incorrect, because he drew a distinct linear relationship between Mhz and frame rate. There are many factors that influence frame rate apart from the CPU - operating system, mainboard, RAM quantity, RAM type, RAM timings, drivers, AGP bandwidth, video card architecture, video memory, video bandwidth, bus speeds, HD access times, background tasks, FS settings, the particularly aircraft you fly in FS, FS add-ons - just to name a few of many. So to make a comment that Mhz is directly proportional to frame rate seems quite ridiculous to me.