It seems that a fast i7 CPU doesn't need faster RAM over and above DDR3-1066 as the real world performance 'increase' is negligible. From TOM'S HARDWARE,....read on...MOST INTERESTING! I had gone with DDR3-1800, but have since spec'ed back to DDR3-1066. No wonder Intel doesn't 'officially' support anything past DDR3-1066, or that Dell on their high end has stated that they ship with DDR3-1066 as their own labs and engineers have seen little gain in performance over and above DDR3-1066. The following is from Tom's Hardware, and you can catch the whole report there, if you wish....--------------------------------------------------------------------------------------------------------------The memory industry, which typically focuses on offering faster (and more expensive) memory products, must find these results rather annoying

It seems that a fast i7 CPU doesn't need faster RAM over and above DDR3-1066 as the real world performance 'increase' is negligible. From TOM'S HARDWARE,....read on...MOST INTERESTING! I had gone with DDR3-1800, but have since spec'ed back to DDR3-1066. No wonder Intel doesn't 'officially' support anything past DDR3-1066, or that Dell on their high end has stated that they ship with DDR3-1066 as their own labs and engineers have seen little gain in performance over and above DDR3-1066. The following is from Tom's Hardware, and you can catch the whole report there, if you wish....--------------------------------------------------------------------------------------------------------------

Good luck to you and your FSX fun if you believe that Grade-A USDA Choice Horse ManureKeep looking for reasons to not spend the money on the right parts....then people buy a AMD system with a ATI card and tell everyone how much better it is than i7 correctly supportedThere is a sucker born every minute -PT BarnumDDR3-1600 6-6-5-18 MushinLearn itLive itRun it

I have to say, looking at the results of this test, all run on one motherboard, I have serious doubts as to whether the settings made in the BIOS for memory speed and timing for these tests are actually being implemented by the chipset. It just does not compute. I would have to see the memory column address strobe waveform on a o-scope with my own eyes to even begin to believe what these guys are reporting. My gut feel is that they made the changes in the BIOS, but the chipset is instead using the SPD settings from the DIMMs, or some similar scenario. I've heard of this sort of behavior with at least one X58 mobo and a flaky BIOS version.RegardsBob ScottColonel, USAF (ret)ATP IMEL Gulfstream II-III-IV-VColorado Springs, CO

Good luck to you and your FSX fun if you believe that Grade-A USDA Choice Horse ManureKeep looking for reasons to not spend the money on the right parts....then people buy a AMD system with a ATI card and tell everyone how much better it is than i7 correctly supportedThere is a sucker born every minute -PT BarnumDDR3-1600 6-6-5-18 MushinLearn itLive itRun it

------------------------------------------------------------------------Hi Nick,I'm sure that as strongly as you feel...if you were to debate with their findings from their (Tom's Hardware Site) lab, it would be a very interesting read! Perhaps you should, and then point out that link to us. I sure for one would follow it! It wasn't about money at all...it was about, do I need then to buy this? They are saying that with a fast i7-975, there is no payout that can be seen. If you have actual FPS increased performance or faster rendering of scenery over DDR3-1066 RAM or other information to the contrary of what they published....I personally am certainly open minded to read that. Why not use their report as your base line and then show that having DDR3 RAM past 1066 MHz will make the purchase/performance a sound investment when using an i7-975 CPU in your system. I follow your threads, as I follow others, and enjoy all opinions.

I have to say, looking at the results of this test, all run on one motherboard, I have serious doubts as to whether the settings made in the BIOS for memory speed and timing for these tests are actually being implemented by the chipset. It just does not compute. I would have to see the memory column address strobe waveform on a o-scope with my own eyes to even begin to believe what these guys are reporting. My gut feel is that they made the changes in the BIOS, but the chipset is instead using the SPD settings from the DIMMs, or some similar scenario. I've heard of this sort of behavior with at least one X58 mobo and a flaky BIOS version.RegardsBob ScottColonel, USAF (ret)ATP IMEL Gulfstream II-III-IV-VColorado Springs, CO

----------------------------------------------------------------------Hi Bob,Bob, I am in no way a 'Champion' for Tom's Hardware, but I do visit their site and read their articles and reviews coming out of their labs. I have also not seen too many manufacturers of equipment take them on for having created false, erroneous, or exaggerated findings. I can think of only perhaps one Company that thought to openly challenge their lab findings on their product..and after the opening salvo, did not offer any further facts or findings, to debunk Tom's Hardware lab's findings. Nothing...I waited every day for it. Nothing. They only took umbrage with their negative report on a part's perceived performance enhancements. Again...I have followed that site for as many years as I have been sim'ing...and have not seen them fumble the ball, if that is even a correct term,except on one occasion when they had reported on a Celeron CPU and it turned out to be a bad part. Intel replaced that part and testing again...came out with better and anticipated results. In this review, if I take it you went over there and read the whole thing for yourself, they tested multiple samples and came out with the same findings. I also believe that they have a level of lab expertise that rivals others that do this for a living. This IS their bread and butter too! Until I see someone come up to the plate with lab findings that puts this article into doubt, I shall rather believe what I have read and have viewed in their masks. This does not make me a Barnum-Sucker. This makes me an astute reader that follows due-diligence with his purchases. If there is a truly increased performance gain by having faster RAM than what even Intel Spec's...then by all means...out will come the wallet. I saw the performance gain for having an i7-975 based system...with a GTX-285...and the wallet came out just last week! :) It's not about spending money...it's about wasting money. I did spec my system with DDR3 1800 based on what people were saying...but I must admit, I did not base it upon what a reputable lab was reporting. I decided to base my purchase upon a lab's findings. If another lab reputes Tom's...I will certainly change my M.O. for memory in my i7 system. It only takes opening the wallet again...(smile).Cheers!Mitch

------------------------------------------------------------------------Hi Nick,I'm sure that as strongly as you feel

its not about how I FEELComing from a highly certified background in Aerospace with respect to electronic engineering its about what I KNOWand Bob is highly qualified to KNOW it as wellTo Bob and I, a few simple formulas applied and we KNOW what's BS and what it true without questionLets forget everything Tom

You have been informed.. the choice is now yours.. make it a good one! :(---------------------------------------------------------------------------------That is always the mission, Nick!The road...is one of constant enlightenment, until the last breath....or...the power goes out.... :)Mitch

You have been informed.. the choice is now yours.. make it a good one! :(---------------------------------------------------------------------------------That is always the mission, Nick!The road...is one of constant enlightenment, until the last breath....or...the power goes out.... :)Mitch

Go back up and re-read my post.. I added some tech to it so you understand where Bob and I are coming from

Go back up and re-read my post.. I added some tech to it so you understand where Bob and I are coming from

----------------------------------I was just sitting down with a coffee and to fire up FSX and take a night flight...BUT....I will do that first!Regards,MitchP.S.Ok Nick...I just came back from reading your added thoughts.After reading your added content, was most interresting, with thanks.

sesquashtoo: hmm remember reading a similar article for the i7 launch, you might want to dig in there I think they have actual Intel Benchmarks. From what I recall they Intel assumed a QPI channel of 400mhz and the tests they ran showed that from 3X400mhz they were getting about 801MhZ... just enough to call it three, but twice as much as they started out with. The numbers do get confusing but a regular pentium would be effective only one QPI channel or 400mhz :) But what I see on the market is ultra low latency 1066/1333 chips priced better than the equivalent 1600mhz. Personally I hate all these brands.. 100 middlemen between me and Samsung all trying to make a buck on specs I don't need.

sesquashtoo: hmm remember reading a similar article for the i7 launch, you might want to dig in there I think they have actual Intel Benchmarks. From what I recall they Intel assumed a QPI channel of 400mhz and the tests they ran showed that from 3X400mhz they were getting about 801MhZ... just enough to call it three, but twice as much as they started out with. The numbers do get confusing but a regular pentium would be effective only one QPI channel or 400mhz :) But what I see on the market is ultra low latency 1066/1333 chips priced better than the equivalent 1600mhz. Personally I hate all these brands.. 100 middlemen between me and Samsung all trying to make a buck on specs I don't need.

------------------------------------------------------------------It seems that we all are only separated by what do they say...6 degrees? LOLOLOLOL! I have JUST BEEN seeing that myself over the last hour and a half....that there are some pretty fast 1066 timings out there....and with Nick showing that timings REALLY influence the final 'performance/speed' rather than than mere xxxx Mhz ratings...it has become a 'free-for-all'...LOLFunny that you just posted this....as I had just come back from this discovery myself, LOL!I guess that Nick was suggesting that DDR3-1600 8-8-8-24 is a sweet spot. I had seen 1066's with timings of 5-5-5-12!What to do...what to do....oh dear...ROTFLMBO!Cheers!

Words are meaningless without benchmarks to back them up, and so far no one has benched FSX with varying RAM speeds and timings. Anandtech did a similar piece not long ago and came to the same conclusions.

------------------------------------------------------------------I guess that Nick was suggesting that DDR3-1600 8-8-8-24 is a sweet spot.

8-8-8? NO! Thats SLOW!6-6-5 YES!..and look at wherer CAS 5 is on the chart Mitch,.. in relation to 1066. For CAS 5 you want to be at DDR3 1333CAS 8 @ DDR3 2100 is EQUAL to CAS 6 DDR3 1600 in memory latency. There is something to be said for bandwidth too however it is the the memory timing, or latency, which drives the advantage to FSX in the hardware. Phil Taylor said quite specifically that one of the major issues with the FSX software is its poorly thought out design with respect to memory and CPU optimizing http://en.wikipedia.org/wiki/Memory_disambiguation The application litterally attacks the physical memory to CPU cache with a scrambled mess those two must unscramble.. the faster the communication path betweem the memory and the CPU cache, and, the larger and more advanced that CPU cache, the better the FSX result Intels Smart Memory Access and Smart Cache design are two very specific reasons WHY the Core2/Quad far exceeded AMD in FSX performance when they were released back in 2006/07. http://www.intel.com/technology/product/demos/cache/demo.htm. i7 has taken that to the next level. This comes down to knowing how to use the hardware you have purchased, and knowing how to set up the application with respect to that hardware.. Mitch.. do you remember what happened after you followed my directions for setting up the OS and defragging correctly? This is the same!you DO get what you pay for IF you know how to use it@ MAX.. tell ya what son... you waste your time money and let us know instead. When you get finished post your results and I will show you the mistakes in assumption that were made in those results that even the so called 'experts' at hardware sites would not know or understand they made as well. <sarcasm_on>Dont forget to buy an ATi card too because we all know they run circles around Nvidia in with heavy weather in FSX <sarcasm_off>

"There is something to be said for bandwidth too however it is the the memory timing, or latency, which drives the advantage to FSX in the hardware. "Thanks for posting that chart again Nick. Am I understanding this correctly in that one should purchase memory that puts you the lowest on the left hand scale?I have an old Q6600 that I am looking for some dual channel memory for. On Newegg, the fastest memory available for Cas 4 is DDR2-900mhz, Cas 5 is DDR2 1200mhz, Cas 6 is DDR3 1333, and Cas 7 is DDR3 1600. According to your chart the lowest CAS latency is obtained with the 1200 mhz ddr2 memory, so would this be the best memory choice for FSX for a Q6600 on a P45 chipset motherboard?Ted

"There is something to be said for bandwidth too however it is the the memory timing, or latency, which drives the advantage to FSX in the hardware. "Thanks for posting that chart again Nick. Am I understanding this correctly in that one should purchase memory that puts you the lowest on the left hand scale?I have an old Q6600 that I am looking for some dual channel memory for. On Newegg, the fastest memory available for Cas 4 is DDR2-900mhz, Cas 5 is DDR2 1200mhz, Cas 6 is DDR3 1333, and Cas 7 is DDR3 1600. According to your chart the lowest CAS latency is obtained with the 1200 mhz ddr2 memory, so would this be the best memory choice for FSX for a Q6600 on a P45 chipset motherboard?Ted

NoIn the case of 1st generation DDR3 which has the memory controller in the Northbridge the goal would be DDR3 1600 7-7-7 @ tRD (Performance Level 6) obtained with a CPU STRAP of 333 or 400 (depending on the proc/motherboard in use in use), with a FSB of 400 (or greater with better procs)DDR2 and 1st generation DDR3 is limited in timing ability Also, with DDR3 memory further perf can be gained with the setting DRAM Static Read Control: ENABLED which can also cause instability if the memory is not high quality but will increase performance in DDR3 1st gen if that memory is of good qualityhttp://www.newegg.com/Product/Product.aspx...N82E16820146777 thats what you wantWith 1st Gen DDR3 you are dealing with the northbridge. The chart above IS in play no matter what platform however the NB which was eliminated with i7 is a factor tham must be correctly set up to gain the best advantage possible with the memory speed/timing. Furthermore, 1st gen DDR3 will only deliver a MAXIMUM of 60% efficiency based on bandwidth in use, THEREFORE the higher the SPEED with 1st gen DDR3 on the lowest latency with the RIGHT northbridge settings, the better the result. Memory speed IS a factor based on the northbridge restrictions on bandwidth efficiency! i7 does not deal with that.That is done with memory that can and will run tRD 6 (AKA Common Performance Level or Transaction Booster) and will also run 1T CMD (AKA Command Rate) and a motherboard that will allow CPU STRAP 333 or 400 with a minimum FSB of 400A Q6600 would need to run 3.6GHz on the right HSF to do what I posted correctlyA Q6600 is really old news so upgrading that proc with the above memory and settings would be a boost for you.. A Q9650 would go a long way with that