Fsx needs clock speeds.

Fsx needs clock speeds.

That's correct as long as you're talking about the processor.Regarding memory see above.

That's correct as long as you're talking about the processor.Regarding memory see above.

Yes I do agree.

The reason behind the disconnect between RAM clock speed, memory chip timings and gaming performance is the CPU memory cache (L1, L2, L3). Suppose the cache hit ratio is 90% (a not unrealistic figure because most programs spend 90% of their time spinning in tight loops, the phenomenon known as "locality of execution" behind memory cache performance), then speeding-up RAM by 2x will only increase your gaming performance by 2 x (1 - 90%) = 2 x 10% = 20%, so 2x gets whitlled down to 0.2x. Now memory timings are a different story because they affect the time the CPU idles after a cache fail and waits for the memory chips to make the data available to the CPU. Cheers, - jahman.

Am I reading the results wrong? Because from what I can see the difference in performance between all of the latency/memory frequency settings are negligible, but they do not show an advantage to low timings as you say GSalden, if anything its a tiny tiny performance increase when using higher memory Mhz. This is both in Dazz findings and yours.

I just tested a couple different configs:2133 CL8Frames, Time (ms), Min, Max, Avg10670, 300000, 23, 50, 35.5672133 CL9Frames, Time (ms), Min, Max, Avg10669, 300000, 23, 49, 35.5631600 CL7Frames, Time (ms), Min, Max, Avg10428, 300000, 23, 49, 34.7601600 CL9Frames, Time (ms), Min, Max, Avg10099, 300000, 22, 47, 33.6631333 CL7

The highest avg FPS can be seen when using the highest Mhz and low CL. When comparing 2133 @ CL9 with 1333 @ CL7, the higher memory frequency gives a slight performance increase on all sides. Same is true with 1600Mhz CL9 vs 1333 Mhz CL7.

Exactly. Look at the difference between 1333 CL7 and 2133 CL8. 2 FPS more average.

Am I reading the results wrong? Because from what I can see the difference in performance between all of the latency/memory frequency settings are negligible, but they do not show an advantage to low timings as you say GSalden, if anything its a tiny tiny performance increase when using higher memory Mhz. This is both in Dazz findings and yours. The highest avg FPS can be seen when using the highest Mhz and low CL. When comparing 2133 @ CL9 with 1333 @ CL7, the higher memory frequency gives a slight performance increase on all sides. Same is true with 1600Mhz CL9 vs 1333 Mhz CL7.

You acknowlegde what I am saying the whole time.Fast memory can run at slower speeds , BUT very short timings , making them faster than higher clockspeeds with looser timings. A 1600mhz memory with default timings of 8-8-8 will never run at 1500 mhz 6-6-6, but a 1800+ memory can.Therefore you lowest, average and highest framerate will be higher too. There is the advantage.

There has been a tonne of tests done clock vs. bandwidth. General consensus is go for clock first bandwidth second. I heard it stated that FSX responds to clock almost linear i.e. 5% increase clock~ 5% increase FPS. However FSX is one of the few programs around that remains sensitive to memory speed. If it was me I would go for the highest clock and the fastest timings at the greatest bandwidth, in that order. While this thread http://forum.avsim.net/topic/329356-i5-2500k-vs-ram-frequency-and-latency/ does not address the timing issue as it pertains to FSX specifically, you may find it interesting to read.

So then, (as it looked before), CL11-2133 could potentially run at CL5-1600 because it is faster memory. The initial CAS rating were slower because of the fast Mhz speed. So as you dial down the Mhz, the CAS speed gets (potentially) better?Mark

4.2 and 1600 anyday.

Agreed.

So then, (as it looked before), CL11-2133 could potentially run at CL5-1600 because it is faster memory. The initial CAS rating were slower because of the fast Mhz speed. So as you dial down the Mhz, the CAS speed gets (potentially) better? Mark

That is correct.FSX is an old game programmed in 2004/2005 and 2006 and does not make use of all the bandwith a modern memory with fast clockspeeds has. It also is important if you have an Sandy Bridge processor or one before that ( like I have). Quote from Overclock.net"In the newer Sandy Bridge chips fast memory speeds and tighter timings don't do squat.While tighter timings and higher frequencies matter on older builds like 775/1156/1366 it doesn't matter as much on Sandy Bridge because it doesn't care what your ram speed is.You overclock by the multiplier not the by the FSB and CPU multiplier anymore."