Well, the correlation is so evident between all the relevant variables that I don't think it's necessary to calculate a coeficient

Correlation would meaningfully reduce multiple numbers to a single number, similar to what std dev does, and is a mathematical way of obtaining the slope of a best-fit line on a chart (so you don't have to plot the chart and guess the best fit line).

How could we determine the memory throughput for a given test?

PassMark MemTest?

2) RAM frequency and latency don't even affect Stuttering, in any obvious/visible/tangible/objective and non-idiosyncratic, worth mentioning way.which translates also to --> Don't buy Faster (or/and "tighter") Ram just to Improve FS' motion fluidity (stuttering). You won't.

I do not agree with that statement at all, here scroll down 3/4 of the page when Nick is explaining the up's and down's of tighter timing when we were playing around together trying to squeeze every ounces of perf out of our rigs... http://www.simforums...4241_page4.htmlHere is a quote from him on the next page when I asked him if he saw some improvement in FSX...now keep in mind he's at 4.5GHz 1900MHz 7-7-7-24 and I'm at 4.67GHz at 2000MHz 7-7-6-20 .....Originally posted by alainneedle1Any improvement in FSX at all?Quote" uh huh fixed the really fine micro-stutters that come up for about 5-8 seconds over a AI jammed hub in one of my saved tests. since it was only one spot I saw it, I didnt put a lot into it. also, I found that it was better to remove my bufferpool entry completely. I was running 90MB, now none BP=0, all that other nonsense ran worse everything else is the same, TBM 120, wide screen enabled, disable preload, hide infotext, highmem but I remove that one if I am not using a big payware plane. The rest is all FSX default. I run 80AL/80GA UT2 Traffic with the jetways disabled and also make sure those pesky cars are disabled in the UTX interface. Otherwise 100% SC/AGEnd of quote.Maybe the result can't be the same on the 2600K because of the 100 BCLK limit on this CPU....I don't know but I got my stutterings at almost nothing with a BCLK at 201 and tweaking my memory to 2010MHz 7-7-6-20 iT trfc at 95.....from 2000MHz 7-7-7-21 2T trfc 112 including other memory tweaks.Also just like Nick, I do not have to use any tweak in my fsx.CFG folder, matter fact I have the TBM at 400 and the load_radius at 8.5000.....now before you all get on me about the TBM math numbers not adding up (yes I've also read that part from Nick)......well....I do not see any negative impact on FSX resulting from that TBM number...I do not have the 2600K set up so I can't speak about it but I can see you guys getting some very good results from it so if my high BCLK with a well balanced rig do not have any impact on FSX I do not know what else will make it run like a dream, is it perfect...not at all but I can enjoy FSX with with almost no stutterings.

Very intereresting thread indeed.This is my fourth PC since FSX release (...). I even don't remember them all. But there is one thing I know, stutters or micro-pauses don't depend on CPU power nor on GPU power.Maybe are you on something interesting here with RAM latency.Regarding stutters I've noticed the following:- Launch FSX without fps limiter.- Set your fps to 20 and select your test flight. In external view try to observe the general feeling while flying over lot of autogen buildings and trees. Obviously it won't be that smooth but almost stutters free.- Set your fps to 24 and observe the overall improved fluidity but also the beginning of micro pauses.- Set your fps to 30. The stutters will slightly increase.The best balance for me so far is 22-23.Does this makes sense to you guys?

I do not agree with that statement at all, here scroll down 3/4 of the page when Nick is explaining the up's and down's of tighter timing when we were playing around together trying to squeeze every ounces of perf out of our rigs... http://www.simforums...4241_page4.htmlHere is a quote from him on the next page when I asked him if he saw some improvement in FSX...now keep in mind he's at 4.5GHz 1900MHz 7-7-7-24 and I'm at 4.67GHz at 2000MHz 7-7-6-20 .....Originally posted by alainneedle1Any improvement in FSX at all?Quote" uh huh fixed the really fine micro-stutters that come up for about 5-8 seconds over a AI jammed hub in one of my saved tests. since it was only one spot I saw it, I didnt put a lot into it. also, I found that it was better to remove my bufferpool entry completely. I was running 90MB, now none BP=0, all that other nonsense ran worse everything else is the same, TBM 120, wide screen enabled, disable preload, hide infotext, highmem but I remove that one if I am not using a big payware plane. The rest is all FSX default. I run 80AL/80GA UT2 Traffic with the jetways disabled and also make sure those pesky cars are disabled in the UTX interface. Otherwise 100% SC/AGEnd of quote.Maybe the result can't be the same on the 2600K because of the 100 BCLK limit on this CPU....I don't know but I got my stutterings at almost nothing with a BCLK at 201 and tweaking my memory to 2010MHz 7-7-6-20 iT trfc at 95.....from 2000MHz 7-7-7-21 2T trfc 112 including other memory tweaks.Also just like Nick, I do not have to use any tweak in my fsx.CFG folder, matter fact I have the TBM at 400 and the load_radius at 8.5000.....now before you all get on me about the TBM math numbers not adding up (yes I've also read that part from Nick)......well....I do not see any negative impact on FSX resulting from that TBM number...I do not have the 2600K set up so I can't speak about it but I can see you guys getting some very good results from it so if my high BCLK with a well balanced rig do not have any impact on FSX I do not know what else will make it run like a dream, is it perfect...not at all but I can enjoy FSX with with almost no stutterings.

Alain,First things first,Thanks for providing this highly educative link from Simforums. Can't go through it right now but it's bookmarked for further reading.Now, I feel happy that you managed to enjoy stutter-free (or near stutter-free) flights since we must not forget that this should normally be the goal: to simulate flight and most of its real-life aspects such as motion fluidity, which in terms of serious simulation must be considered fundamental.I still though argue that RAM upgrades (in terms of frequency and latency) can't provide us with reproducible proofs of performance gain (either fps or stutters or both). Almost all reviews I've stumbled upon say so, Dazz's and Corey's research says so, even if they tend to imply that there must be something going on with the frequency (one every 45sec in their example) of stutter appearance. Even so, Dazz only gained 3fps moving from 1333 to 2000 while he reports there was no improvement in stuttering whatsoever.

• 4Gb of the cheapest Kingston DD3 1333Mhz (2x2gb) cost about 40 euros in my place.
• 4Gb of the most expensive SuperTalent DD3 2000Mhz (2x2gb) CL8 cost 165 euros.

Say you want 8Gb, figure that tends to be the next gaming norm. Prices climb to 80e and 330e respectively. Normally, one need to justify these 250 extra euros from a "slow" to a "fast" memory. Roughly speaking, if one chooses to spend 250e extra for faster RAM, he/she finally will end up having paid about 80e per RAM-gained frame..... Or not? Now imagine the same fellow simmer spending 250e to a faster CPU or a better Mobo or more amount (in MB) of RAM. Wouldn't it be better?Under these terms, I consider any expenses towards ultra-high fast memory a waste of money for most (or for the average, serious though) fsimmers BUT I can't stress enough how pleased (and jealous maybe!!) I am that you eliminated most of your fs-stutters.You know, stuttering in FS is like arguing with your wife. One can always be extra cautious and fine-tuned, having all air and hydro-coolers set and running, experienced and such, but, at the end you never know where, when and how it all started... . Usually, they "overclock" badly...!

Hey George, I'm not disagreeing with you on what you just said, I'm not talking about $$ vs perf., is my PCI-e SSD a good investment perf. wise compare to the price I paid just for FSX....ABSOLUTLY not....but I'm glad I did it.There is something going on with the rams for the SB, I don't know if this is because the BCLK is locked at 100 on the the SB but each result that Dazz provided are almost the same regardless of the timing 2133 CL 7 vs 2133 CL 9.....and it should not, high MHz with a lower latency will always result in higher perf.Now, I'm not an expert about rams but I do see a difference stutterings wise from 2000MHz 7-7-7-21 vs 2000MHz 9-10-9-27...BIG time having tested it myself and the lost of stutterings is also noticable from 2000MHz 7-7-7-21 2T, trfc (112) vs 2010MHz 7-7-6-20 1T, trfc (95).Will the triple chanel (X58) compare to the double chanels (SB) be better because you can overclock them or is it because I run a high BCLK compare to the SB or is it all the component together ?? I don't know.....

Very interesting thread! I have a question or two that has has me thinking for some time, without coming up with a definitive answer:What possible routine happens in FSX programming/scheduling every 45 seconds to a minute or so that produces a spike in workload, or a large transfer between disk/CPU/GPU/RAM, or a recalibration, or exceptional rendering or stacked combination, or something similiar? Is it an adjustment in time/light/shadow/shade/sun/moon/moon/stars/horizon/deflection/angle/update/seek? If so what? Does anyone know?If there is something, that might be the primary cause of the predictable sequential stutter.Kinds regards,

Well guys, let's not jump into conclusions too early. This is just a single test on a single machine and maybe my interpretation for the results is not right. It doesn't really prove a thing.There is a performance boost with faster RAM. It's small, but it's there, so it up to each one of us as individuals to decide if it's worth our money or not. Personally when I give someone advice on these things I try to make sure I don't forget we all have different needs. It's not the same if you already have the RAM or not. Not everyone needs the fastest hardware, as not everyone needs all the eyecandy etc.. Budget also is important.It's probably better to tell someone what to expect from an upgrade and let him decide than simply post a link to what you would get... that's assuming we know what to expect and, like I said, I don't think the stutter tests are even close to being conclusive

Correlation would meaningfully reduce multiple numbers to a single number, similar to what std dev does, and is a mathematical way of obtaining the slope of a best-fit line on a chart (so you don't have to plot the chart and guess the best fit line). PassMark MemTest?

Ok, I thought you meant throughput in FSX. Sounds good. I could try and correlate theoretical RAM speed vs avg, max, min, etc.... so for example, if we take 1333 CL7 as the reference speed level at say 100, 1600 CL7 would be = (1600/1333 * 100) = 120, then 1600 CL9 would be 7/9 * 120 = 93.33, and so on, then using real throughput / bandwidth / latency instead of theoretical speed.

I wouldn't conclude that there is no performance gain. If you're truly looking for every possible FPS, 2133 CL7 appears to provide about 1.5 fps above 1600 CL7. It's obviously a small advantage, but hey, it's something. Just don't expect it to improve stutters whatsoever. Another point worth noting is that the stutters appear to happen at a regular time interval - just look at the dips in the plots below. Each dot represents a frame. I see four prominent stutters occuring at 58s, 109s, 152s, and 195s - so approximately every 45 seconds. Perhaps we should investigate the cause of that.*pic**pic*

Interesting graph close up Corey, thanks again! Do you guys recall that great thread from the guy who broke down FSX performance in a post a couple of months ago? it was his first post and very interensting stuff. What was that he described that increased CPU usage periodically? (twice a minute? can't remember) The fibers or some texture related process maybe?

Every 45 Seconds... I once had an issue with FSUIPC doing an 'auto save' on the PMDG 747 every 60 seconds and realised it would cause a very slight pause but as the complex aircraft needed to be saved with all the FMC info etc it kind of made sense...I am not sure about everybody else but do you find that when one goes through a new build, CPU, Motherboard, New Memory/Video Card that often when you install the OS and a vanilla FSX you have a period of no stutters and over excitement of higher FPS etc and then as you start to add all your scenery + airplanes etc get all your assignments set etc etc etc one day you boot up take a plane into the sky turn left and stutter, stutter, stutter. Unfortunately its never so obvious that you immediately say something is totally wrong but you know its there, as if something has changed....I have just done a new 2500K build, 570GTX + Corsair 1866Mhz 9,10,9,27 and I had FSX running amazingly till last weekend and then the dreaded stutters started to appear. I think it is usually down to being a hardware situation as I always check by putting the scenery sliders down to a minimum and taking a flight but the stutters remain... so it cannot be an over strained scenery situation it could be a software issue but its just so hard to track down...I am about to replace my Asus MB as I need the new B3 Stepping so I will be reseating my CPU and will do a vanilla install and this time I will check to see if I am stutter free and if that is the case I will be re installing all my software very slowly and methodically just in case its software related...This is a great thread and any chance to rid us of these annoying stutters would be most welcome!!Hamish

Very intereresting thread indeed.This is my fourth PC since FSX release (...). I even don't remember them all. But there is one thing I know, stutters or micro-pauses don't depend on CPU power nor on GPU power.Maybe are you on something interesting here with RAM latency.Regarding stutters I've noticed the following:- Launch FSX without fps limiter.- Set your fps to 20 and select your test flight. In external view try to observe the general feeling while flying over lot of autogen buildings and trees. Obviously it won't be that smooth but almost stutters free.- Set your fps to 24 and observe the overall improved fluidity but also the beginning of micro pauses.- Set your fps to 30. The stutters will slightly increase.The best balance for me so far is 22-23.Does this makes sense to you guys?

Is that with or without a BP setting. I've found no indication that RAM speed reduce stuttering or increase fps, One would assume that more ram = better performance, but not when using FSX.My source of stuttering has always been attributed to default BP setting and unlocked fps in FSX. The limiter does nothing to reduce blurries. Only FSX limited (locked at 40) does the trick for me. Affinity Mask at 14 or 15 has helped as well.

Hey George, I'm not disagreeing with you on what you just said, I'm not talking about $ vs perf., is my PCI-e SSD a good investment perf. wise compare to the price I paid just for FSX....ABSOLUTLY not....but I'm glad I did it.There is something going on with the rams for the SB, I don't know if this is because the BCLK is locked at 100 on the the SB but each result that Dazz provided are almost the same regardless of the timing 2133 CL 7 vs 2133 CL 9.....and it should not, high MHz with a lower latency will always result in higher perf.Now, I'm not an expert about rams but I do see a difference stutterings wise from 2000MHz 7-7-7-21 vs 2000MHz 9-10-9-27...BIG time having tested it myself and the lost of stutterings is also noticable from 2000MHz 7-7-7-21 2T, trfc (112) vs 2010MHz 7-7-6-20 1T, trfc (95).Will the triple chanel (X58) compare to the double chanels (SB) be better because you can overclock them or is it because I run a high BCLK compare to the SB or is it all the component together ?? I don't know.....

Alain. Can you please save a flight, and run fraps on that with two different RAM configs but everything else the same? then send me a zip with the files thenBTW, would be interesting to test GPU's with this. Too bad I don't have any of my old cards anymore

The best balance for me so far is 22-23.Does this makes sense to you guys?

I always used to run my frames unlimited until I started to use 8xSQ FSAA. Now I have them locked at 40 with the external limiter, but just because of antialiasing induced jerkiness in dense clouds with certain aircrafts.The in-game limiter is too hard on frames and I never ever use it

Alain. Can you please save a flight, and run fraps on that with two different RAM configs but everything else the same? then send me a zip with the files then

I can, like I said before the FPS improvement is not the point, what I'm getting at is what I can see as far the stutterings on the screen.Any flight?

I wouldn't conclude that there is no performance gain. If you're truly looking for every possible FPS, 2133 CL7 appears to provide about 1.5 fps above 1600 CL7. It's obviously a small advantage, but hey, it's something. Just don't expect it to improve stutters whatsoever. Another point worth noting is that the stutters appear to happen at a regular time interval - just look at the dips in the plots below. Each dot represents a frame. I see four prominent stutters occuring at 58s, 109s, 152s, and 195s - so approximately every 45 seconds. Perhaps we should investigate the cause of that.

Nice charts, cmeeks!Suggestion: Why not define a stutter as FPS <= 20, then have Excel count how many of those you have in your test run, normalize that number to a "per minute" basis, and add that stat (together with standard deviation) to the benchmark result set?Alternatively, to avoid counting all stutters equally and recognize FPS dips according to severity, you could weigh each dip, where DipWeight = (20 - DipFPS), then sum all your weights and normalize to a 1 minute run. The result will be proportional to how many frames were "missed" (not rendered) by stutters.Cheers,- jahman.

I can, like I said before the FPS improvement is not the point, what I'm getting at is what I can see as far the stutterings on the screen.Any flight?

Just the flight you tested where you experienced more stutters with slower RAM. I just want to try my spreadsheets in an different scenario

Does FSX have an internal clock, if so what is it?Oh, could someone please point me to the FSX RedBull Reno Air race benchmarks?

Just the flight you tested where you experienced more stutters with slower RAM. I just want to try my spreadsheets in an different scenario

I see what you mean now but I don't have the 2000MHz 9-10-9-27 since they were my friends rams, he came over one day for me to test his rams with memtest cause he was building a new rig and he did want to make sure they were good.They did pass memtest so I tested them with FSX also, that's where I saw a difference.