Nick,Just for giggles, I installed an updated BIOS for my motherboard and am now finally able to boot and run at 3.6GHz stable (prime95x4 for 4hrs), ie, 400x9.I configured my RAM as follows and received the following Sandra and FSXMark07 (global high default, AA/AF on 1280x1024 windowed) results:4-4-4-10 - Sandra memory 6673/6697, FSXMark07 15 min, 41 max, 26.703 avg5-5-5-18 - Sandra memory 6443/6471, FSXMark07 14 min, 40 max, 25.827 avgWith the FSXMark07 runs, I watched both like a hawk to see if there was any visual difference that I could detect, of which there was none. I even did a couple of stop/go runs where I paused the sim every minute or so to see any of the scenery needed to catch up, which it did not. So from my perspective, the two runs were about FSX performing well enough to do everything asked of the scheduler and therefore that any FPS difference should come down to the one thing I changed between runs - memory speed. These results translate to a 3.5% boost in Sandra and a 3.4% boost in FSXMark07 at the higher memory speed. This shows to me that FSMMark07 is indeed picking up the memory boost, and is therefore a useful indicator of measuring the effect of changing RAM speed in FSX. What was a little disappointing (although expected from the synthetic benchmarks out on the net already), is that there is little return on investment for buying faster RAM as it will cost you hundreds extra for a couple of percent gain at most. At least FSX comes to the party by realising most of the synthetic benchmark gains.Gary

I can tell you one thing I will not be doing and that is running the benchmark with AA and AF controlled by the sim. I will be running FSX as I would normally run it allowing my video card to handle AA and AF. I am actually not too keen on using the config files at all this time. I want a base of exactly what I know is correct for settings on the test system to remove as much of the unknown factor as possible.I know what the system will do when its set correctly and that is the baseline I intend to maintainIf I have time I will run through tests from DDR 1200 through DDR 1440 using various setups (ie; default BIOS, CPU only, full clock) so the result can be looked at and graphed.. as I said time is an issue right now but I do wish to provide a good data set.OK, so your up to 3.6GHz 400FSB and 4-4-4... @ what mem-speed and divider was that.. if it was not DDR2 1000, (500MHz) your results will be limited. In order to force tRD where it belongs that memory and divider must be in sync and as mentioned the 965 is a bit tricky about that. If you use the software MEMSET it will give you a readout of the system and the performance level (tRD) it is currently sitting at.You should be at PL 5, 6 at most to see a defined changeI can not place more emphasis on the frame rate being used to completely judge value in a clock or a change in memory timing, as nearly useless. The overhead being given to the application with all my add-ons and settings running during a test like that is what tells me if its worth it or not.

Nick,It was at 1:1 (ie 800) STRAP. That's as far as my memory will go. In fact, I had to boost VDimm by +0.3V to get it to work there long enough to run FSXMark, so it is not a stable memory config for me. As such, I will not be able to see the fruits of much better memory timings on my system at least.Re: "I can can not place more emphasis on the frame rate being used to completely judge value in a clock or a change in memory timing, as nearly useless."I mean, I have shown that FSXMark07, when operating within the capability of the FSX scheduler to do everything that is asked of it, can in fact mostly replicate what the synthetic benchmarks are showing as improvements in memory timing. And it is repeatable. Whilst I wouldn't use the term "completely judge" as you so absolutely define, I do say that it is a pretty darn good indication of the effect of such changes.The alternative is to peer at the screen, looking for what you perceive to be differences from the last run, all the while fighting the placebo effect that humans are renowned for falling victim to and the variances in our individual interpretations of what we are actually seeing fly across the screen. I'm sorry, but I do not see how this can be better.So I would say in return: I can not place more emphasis on a subjective look and feel being used to completely judge value in a clock or a change in memory timing, as nearly useless.And at this point, we once again should agree to disagree. :-)Gary

I went ahead and ran preliminary testsI don

Nick,You changed the method of measurment between runs, so of course the results are not comparable! Obviously FSX measures FPS differently than FRAPS does, but that's not the point. The point is to compare performance between runs performed using the same method. It's the relative performance we are looking for, not the absolute. For future runs please follow the instructions verbatim, not bend and modify them to suit your argument ;-)And so what if my config was not acceptable to your eyes? It was picked because it is super easy for anyone trying to set the test up without much knowledge of config files and because it is a default out of the box setting from MS. It is also a baseline that allows results to have some common meaning when comparing results with others. I am pleased that you have created a better CFG file and in fact would be interested in seeing what you have done to make such an improvement.If you want to debunk the science of my efforts any further then can you please be more scientific in your debunking methods. And also, can you please espouse how human perception, as the alternative, can provide more consistent and reliable results.Gary

I think I am finished GaryThere is no need to go any further with this... It was not my intent to debunk your benchmark. My intent was to show you it is impossible to use frame rates (internal or external) from MSFS as a means of measuring true and accurate performance with respect to real world use/visual quality, and, carry that into a number which places a linear perf curve into an increase or a decrease. The results are not linear.I use your benchmark when I set towers up however I use it another way when I do for a reason. Your work is not useless, I never said that and I also never implied the benchmark did not have value, only that the user needed a clear understanding of what it is they were observing on the screen to validate the result.The difference between 21.1 and 36% is quite a change. The 2nd result should have produced close to if not an equal increase if FPS can be used in FSX or any MSFS title to establish true perf. I saw a 26% increase with test 2, not 36.Even if I gave your methodology a 5% margin of error, it still doesn

Nick - what do you see on the horizon for new technology based cards coming up in mid year?

Now, although subjective since we have seen the result, one way to use your benchmark to see DDR2 and DDR3 (MEMSPEED) in action since we are both running the same CPU speed is for you to use my config file, set up the drivers as I did and run the FRAPS test again. That result would then display the difference between DDR2 800 @ CAS4 and DDR3 1440 @ CAS6Even that old P35 test tower running the cheap Kingson CAS 6 memory is not a good DDR3 example, but that test would show both the difference and the potential at the very least. Since it uses what you consider the correct frame counter, the result should have value and be correct.The Kingston memory running CAS6 and the Corsair running CAS4 is quite equal in the charts when it comes to where they should be for optimal performance. We are both equally off in that respect due to memory limits. Your DDR2 800 should be at CAS3 and my DDR3 1440 at CAS5 however the only difference is I have set up based on tRD/MEMSPEED/CPU/STRAP and you have not.

Nick,You are still comparing the difference between runs that used a different measurement method, which I have already stated is a foul from a scientific perspective. You then go on to compare the difference in performance between two different CFG settings on the basis of those different measurement methods, which is a double foul because you are making a conclusion based on initial flawed logic.Granted, I did not set my benchmark up to use the FSX frame rate counter. According to you that is a good thing, as you do not consider it to be accurate. FRAPS has been, and still is, a defacto standard FPS measurement tool over the years, used by many reputable websites with their hardware reviews. It is independent of the application being tested and can therefore not be accused of cooking the FPS figures to suit a particular program's marketing machine. Using the FRAPS method, I get repeatability of testing of much less than 5 percent across the entire 5 minute run (ie. not just the average FPS figure), which is a reasonable margin of error IMO. In short, just use FRAPS as directed ;-)Gary

OK, GaryI am going to drop itYou can cry foul all you want and claim science is not being followed. I offered a confirmation recheck using your own method and FRAPS with a control in the test and unfortunately you insist on making this personal and call it a T --I --T for tat in another thread, your own words, not mine.I am sorry I bothered you and attempted to have a conversation in which some discovery and learning for everyone may have been possibleHave a great day

Nick,Well I was hoping to get you past the start line of my benchmark and actually look at it for what I advertised it to be. If you were truly trying to be objective, you would have got past the red herring of FSX FPS measurement versus FRAPS, and tried the benchmark for the very thing (I thought) we were trying to discuss - do RAM speed changes make measurable performance changes in FSX or are they better assessed subjectively by look and feel? I am sorry that we could not see this discussion through, as I really would like to have seen whether your claim of improved FSX function by balanced memory timings is actually measurable. I still think it is, but I don't have the right RAM to verify my claim. You think it isn't but won't run the test as is to dispute my claim. Stalemate perhaps, but it is far easier for you to just run the test than it is for me to buy and install new RAM. I am also sorry that you did not get to espouse how you think the alternative, ie. look and feel, is a better measure, particularly given the shortfalls I have identified in human perception in general. Anyway, despite this particular discussion going off the rails, I want to acknowledge the great work you have done in getting many in these forums of late back onto their FSX horse. At the end of the day, that's what really counts in these forums - users helping other users! Gary

double post.. deleted

Gary I am going to make this as simple as I canI offered to run additional tests giving you my config file which would have allowed an examination of several results with a control, one of which was the big question; Does DDR2/timing and DDR3/timing truly make a difference in a real world flight environment?Go back up and read, as I never asked you to run CAS3 or purchase anything. I stated quite clearly our systems as they are right now would serve as a good base of that examination.If you had read that with an open mind and with the understanding it is not my goal to play a game with your benchmark or work, perhaps you would have a different attitude.I have been around for a long time Gary. I have been involved with the design of electronic guidance and control systems since before Apollo, including involvement with the creation of the lunar module systems with the folks from MIT and have enough background in computer science and aircraft/weapons guidance systems to know when something is not right or accurate.Regardless of the apology I do not intend to remain in this part of the topic and be insulted any further.

It is sad that you have to end this with a show of credentials (without knowing mine BTW) to make yourself seem higher and mightier than I in this discussion. As you have moved into personal insult territory rather than addressing points that I have repeatedly asked you to comment upon, particularly the "how is your method better?" question, then yes it is best we end this now.Gary

I addressed every point and requested participation to continue and confirm despite your rude and ignorant comments. I know you can read and if you are highly, or even moderately, skilled by computer technical standards you can in fact easily comprehend what I said in that request. Instead, you dodged it with something strange for a skilled technician or engineer, about not having the ability to participate due to components?. My request required no ability and only your system as it stands.If you did not read what I said and passed it over or browsed it in haste because of an emotional reaction, I can understand your subsequent comments. Otherwise they made no sense.You on the other hand have displayed juvenile personal issues in this and other threads in a need to, and I quote, "Make a point" instead of listening and participating to discover and learn with me.As for credentials, I listed nothing official about mine as I have no need.If I had you would not be smarting off as you still are.Good night, sir and yes, this has ended