I forget that internet edict calls caps yelling which was not my intention, nor was it my intention to disagree with what you were saying..I was just posting info for everyone to read and it was not directed specifically at your postSorry if it appeared that way I like the x2 and I do think its going to shape up quite well once the driver issues get nailed. It just hit the streets so in terms of dev time its a newbornone thing that -can- be said... it has less issues than a Nvidia core/design which has had well over a YEAR to develop, thats for -darn sure-:)

Hi Nick,Thanks for the clarification. Face to face would be so much easier. :-)I appreciate your postings today especially those saying trust no-one. The problem is unless people make brave decisions and buy the newest cards how will we know what performs best? Sod's Law means you will always buy the wrong card. :-(I do have a lot of faith in the 3870X2. When I have all the sliders for AutoGen and Scenery Complexity to max and can still get fps in the teens with a unclocked E8400 I'm totally happy with my decisions.The Catalyst 8.3 drivers resulted in a score of 8191 in 3DMark06 compared to 14307 with the 8.2 ones. Clearly there is still work to be done on them and hopefully ATI will sort out the problems soon.In terms of image quality I'm totally satisfied. The images are stunning. Kudos to ATI for that.Anyway, keep chipping in with your views. You clearly know what you're talking about which benefits everyone. :-)Cheers,

Without a doubt, IQ is ATI's turfI will probably making some shifts in my primary tower toward ATi come this summer. I want to get back to the ATi IQ I have missed for the last year+. I dont use vista for regular flying and do use ATi for the Vista tower.. for a reason LOL

That simply won't work Ray.The rest of the system has so much influence on the outcome it would be impossible to correllate the results to any single source being different.Also, you'd almost certainly get skewed results because (as is common in forums) the majority of respondents will be those with an axe to grind or otherwise disgruntled or disappointed with the results. You are even likely to get results that are deliberately doctored to get lower than actual numbers.Those will not be offset by far by the few who post results simply to show off their new rigs.So your results will trend heavily down from what they should be if the tests were performed under controlled conditions.

it would be nice if we could come up with a database of results however in order for the information to have true value the tests would need to be set up with software that would read the items I mentioned above to ensure somewhat reasonable outcomesand then people who read the results need to understand the other factors such as memory sub-timing, CPU STRAP, etc, etc, etcWith that type of database and the correct understanding of the influence with all the data in the mix the results could then be used to make an reasonable decisionbut then again, as I said FSX is DYNAMIC and as such what one test run shows may not have the exact same influence as test 2 on the same hardware. Heck I have even seen time of day make a difference because there is texture and shader changes taking place.There are so many factors around FSX, benchmarks are very difficult to draw conclusion from except when comparing ones own system to itself and a change.

Nick,I understand your position on benchmarks, but right now the best relative hardware comparison I've seen are the benchmarks that Gary ( Reset MCP ) did on this forum as few months ago.You've written before about the importance of correctly setting your Cas, trd, fsb/strap, timing, sub timing etc.... on your system to obtain some benefit from the x38/48 motherboards and DDR3 memeory. But I haven't seen you quantify these benefits. I apologigize if you have and I missed it.I'm running a P4 3.0 right now and I'm pretty confident from reading posts here and the benchmark work that Gary did that upgrading to a Q6600, P35 motherboard, and 4 gb of ddr2 memory will double my FSX performance for about $600. Getting an x38 will double my mother board costs and getting an x38 and the ddr3 memory will probably double my total upgrade costs. I have not read about any real benefits in FSX from this x38/ddr3 hardware but you seem to an advocate of it.Can you tell me what percentage of performance improvement I should expect to get from:a) correctly configuring a P35/ddr2 system as you mention above vs. running everything in the bios on auto?,:( running an x38/ddr3 system on auto bios vs. a p35/ddr2 auto bios system?,c) running an x38/ddr3 system set correctly vs. on auto bios?Assume a Q6600 CPU. If you feel a different CPU is required to see a benefit, please state what that would be.I appreciate your knowledge and contributions here, and don't mean to put you on the spot. I've read that you have done a lot of hardware testing so I am hoping that you can provide some insight to my questions above. The only reason that I've seen so far to get an x38/ddr3 is to allow you to upgrade to future hardware.One other question, is there a document on the net that you can point us to that describes how to correctly set up your timings/system as you've mentioned above?Again, appreciate your time and contributions here,Ted

TedI don

I already did this for someone here: http://forums.avsim.net/dcboard.php?az=sho..._id=36035&page=DDR2 can use the same methods.. the advantage DDR3 has s above DDR3 1800 which is where DDR2 starts getting left in the dust IF the memory has the right CAS rating. Running DDR3 1333 @ CAS 8 or even 7 is a total joke.. your better off on DDR2

Okay chaps, I accept it probably isn't going to work for the reasons you give. In times when PCs were simpler (no overclocking, no RAID, no gits who would send false reports) we did have a system that provided useful info on the best hardare to get.That was also in a relatively closed shop. I'm not suggesting most folk here would not supply valid data - I'm certain they would.So, we can't trust many of the reviews and we can't setup a meaningful test scenario ourselves. Looks like we're stuffed!:-lolEnough typing, I need to fly!

Thanks for the response and links Nick. I just got back from Easter dinner and have got to pack for a flight tomorrow. I will read the links on my trip this week.It appears all of the this optimizing of the memory setting is to increase the dataflow, mb/sec in and out of the memory.I guess the question I was really asking above is when running FSX is the bottleneck in the CPU, GPU or memory I/O with the currently available CPUs. Or maybe I should be asking at what CPU clock speed does ddr2-1066 become a bottleneck. Sorry if my questions are frustrating for you, I'm a turbine engineer and think in terms of flow path. I'm imagining all these little electrons flowing down these circuits and I'm looking for the restrictions and points of choked flow. :) When I use FSX I move the sliders as far right as I can while still maintaining smooth flight and 20 FPS locked. What I meant by performance increase above was the ability to move the sliders further to the right without degrading the flight smoothness or image quality.When you stated above"Between default and properly clocked/set, AND, assuming the memory was indeed purchased CORRECTLY to allow the right result, I would estimate a good 25% over the stock values the board provides. I could say if was me setting the system up, possibly 30-40 depending on the stock BIOS and its maturity"were you referring to FSX performance or memory benchmarks?Ted

Hey Nick,I tried reading the info in the links that you provided, but it hurts my head too much (even being a ginger beer myself). Before I get too excited with memory settings, is any of what you have posted above applicable to 965P motherboards or is it only for newer motherboard chipsets? Also, on what basis do you conclude that FSX performance will be "very noticably increased" by adjusting the memory timings as you suggest? I am sure that memory benchmarks show the gain, but in my experience with FS benchmarking, playing with memory timings has little to neglible effect on FS performance. Have you verified your findings on either a benchmark of your own or the FSXMark07 dynamic benchmark I posted to the AVSIM library late last year?Gary

>>>Knowing Toms Hardware as well as I do, and for a very long>time>>I would say the Nvidia card flew airports such as Jackson>Hole, Wyoming and the ATI card flew LAX and NYCLOL!!!!! You mean the same Toms Hardware that said the Cyrix 6x86 was going to put Intel out of business.

>Thanks for the response and links Nick. I just got back from Easter dinner and have got to pack for >a flight tomorrow. I will read the links on my trip this week.>It appears all of the this optimizing of the memory setting is to increase the dataflow, mb/sec in and >out of the memory.>I guess the question I was really asking above is when running FSX is the bottleneck in the CPU, >GPU or memory I/O with the currently available CPUs. Or maybe I should be asking at what CPU >clock speed does ddr2-1066 become a bottleneck. Sorry if my questions are frustrating for you, >I'm a turbine engineer and think in terms of flow path. I'm imagining all these little electrons >flowing down these circuits and I'm looking for the restrictions and points of choked flow. You are on the right track Ted and being a engineer I highly recommend you check out the links I posted because they will provide you with the information which supports the method of opening the primary restrictions. The two critical areas are CAS and tRD however they must be in sync and work in harmony with the FSB/STRAP (divider). The resulting memory speed is not the primary issue, it

I have in fact run tests however Gary, there is no such thing as a benchmark that will give me a readout on image quality, number of tiles rendered sharp, average distance those tiles rendered in correct LOD, number of autogen items rendered successfully, background cycles skipped due to requests, and, information about how smooth the flight was in its entirety.FSX is dynamic and there is no such thing as a benchmark that will tell me what I need to know... my eyes tell me that and I can use the average frame rate as a guideline however I dont care if i have 20FPS or 40FPS... if they were both smooth and both rendered equally in terms of clarity, no blurs and scenery items remained in the flight... the result is still equal with the exception of the frame number. In real world use, what I stated is all that matters. In fantasy and hardware drag racing, the frame rate presents a value, but that is as far as it goes.Speaking strictly on analysis of FPS, because of the dynamic nature of the software the user can not change views during a test or a FPS test, or comparable results from previous runs, are no longer linear in nature.There are too many factors in FSX which are not recorded for analysis and the user has no way of knowing what background activity is being skipped between tests. All they have to go on is a frame number, which in fact and in the 'big picture' means very little.Yes, what I posted will work for 965 chipsets.With DDR2 as a rule of thumb your target FSB should be 400 (thats just a start but most will use that value), your target CAS 3, 4 or 5, MAX with a DDR2 speed of either 1000 or 1200 (prime values if possible). The initial part of setting that up is finding the CALWI or 'CAS Access Latency Window of Interest' which is where the user needs information about their memory, its sub timings, maximum safe Vdimm ability, etc http://www.thetechrepository.com/showthread.php?t=195Findiing the CALWI and max stable CPU @ CALWI is only the beginning and the CAS or other values may change somewhat in the 2nd part of that setup. The last part is calibrating for tRD which is based on what CAS and FSB you are trying to achieve... bottom line, in MEMSET the PERFORMANCE LEVEL should read the calculated value for tRD and at that point if the tRAS formula is followed which for DDR2 is tRCD + tRP + 2, and, the proper sub-timings are established you should see a marked improvement.Its not as simple as 1-2-3. It does require the user find out from the memory manufacture what the memory can sustain in voltage and it usually requires the user do some testing to find the lowest CAS values possible @ DDR, stable, however it is not about jacking up the memory speed as fast as possible, that will net you nothing if the same 'wait' states are in effect.That is the same as an intersection/road being widened for more traffic but the traffic light remains on the same timing cycle. If the road is widened and the 'red' light cycle decreased significantly and the 'green' light signal increased significantly, the flow of traffic is then optimised... for a computer that means FSX can and will use it.It works on the same principle as my Windows tune-up list. The list removes redundant CPU cycles and the wait states being forced on the CPU at the same time.

even if everyone who responds provides valid data, that doesn't make the total of that data any useful to draw conclusions.It's highly unlikely that that data is collected from a collection of systems that come close to following the bell curve of systems in existence, thus the results are skewed.