I just installed Vista Home Ultimate 64 on my system that I recently upgraded to quad core and am currently installing FSX. As I mentioned in a post a week or so ago, I was planning on running my dynamic FSX benchmark on FSX RTM, SP1, SP2 DX9 and SP2 DX10 and posting the results here, but before I do I was wondering what config you guys think will be best to run against each FSX install state.First up are what FSX settings to use. In the past, I have used global high settings, unlimited FPS + ingame AA/AF which, apart from the AA/AF setting, is what FSX defaulted to when I was running my E6400 CPU I previously had installed. As clinical as this is, I don't actually use FSX with this configuration, because I turn water down to 2.x low, ground shadows on and AI airliners up to at least 50%. I'm sure each of you have your own specific settings too, so it makes it difficult to make valid comparisons with any data I collect.Although I would normally run the benchmark in full screen mode, SP2 has changed all that since I joined the SP2 menu lockup in full screen club last week. Supposedly the 169.09 drivers fix that, but it turns out that FSX RTM doesn't like these drivers full screen, giving alternating flashing and black textures. What to do?At this stage, I am thinking for the RTM run, I will just do the middle of the road global high + ingame AA/AF, and for SP1, SP2 DX9/DX10 doing this same setting again + a max slider scenario (which will give bad FPS I know, but will help answer the many "is hardware fast enough to run all sliders maxed" question that often pops up here) and a scenario with the settings that I usually use. The other aspect to this benchmarking activity is what hardware config to run with. Of course, I will do runs at my stock CPU clock, but I am also thinking of doing a max overclock (I can squeeze out 3.6GHz for testing) as it represents what should be available for mainstream purchase in about 12 months time. Also, I run Matrox TH2G, which if run at max res of 3840x1024 is not really representative of most systems out there, so I was thinking of using 1280x1024 as the base setting.Any other suggestions are welcome, but please be aware that it takes about 10 mins total per run (including setup and data collection time), so this is going to take a lot of time already.Gary

I worked out the flashing textures issue. Vista 64 doesn't like dual monitor enabled when in full screen :-(Gary

Hmmm, a bit of a one way conversation here, but I'll soldier on anyway. Results from the RTM versus SP1 runs are summarised as follows, with only one item varying on each run from the Q6600 SP1 baseline run:http://forums.avsim.net/user_files/180794.jpgSeems SP1 did wonders for a quad core installation, and CPU overclocking has a marked increase in overall FSX performance.More to follow after my Accel / SP2 installation.Gary

Alrighty, the silence is deafening on this post to date, so I'll put this effort on ice for now. Time to crank up Crysis that I just bought :-)Gary

Thanks for this data Gary! I like the various columns, its interesting to see TH2G vs Wide and the Overclock.It might be the "game time factor" is why you aren't hearing more replies right now.

Thanks Phil. As an Australian living in the US, I have never really understood the US thanksgiving week thing, except that we do the same thing in Oz although it is at Xmas which, even though a religious time, is not necessarily religious but family oriented for most Australians. OK, I have just confused myself enough to sit back and relax for the evening. You should too Phil! :-beerchug Gary

Interesting numbers. 1. What does the Q6600 SP1 Opt stand for ?2. Confirm that Q6600 SP1 avg fps is exactly similar in as far as game settings are concerned with Q6600 SP1 3.6GHz ?(Nice work by the way)Pierre

>1. What does the Q6600 SP1 Opt stand for ?Those are my changes to the global high default settings, which compromise aircraft self shadows off, water 2.x low, ground shadows on and AI airline traffic at 80%.>2. Confirm that Q6600 SP1 avg fps is exactly similar in as far>as game settings are concerned with Q6600 SP1 3.6GHz ?YesGary

Hi Gary,I don't know about the rest of the country, but in Denver, most people were outside all day enjoying probably one of the last warm days of Fall. I heard "hard winter" is hitting next week. 70 degrees today, a high of 23 next Wednesday.Thanks for taking the time to run these tests and sharing the data with us. Everyone is going to want to see different display settings, so I say set them as you prefer and let others duplicate them with their systems to compare. I am also considering a Q6600 vs a dual core so I will just list questions that I have on the two setups. If you have the time to run some tests that address them I thank you.1. In addition to the frame rate differences between RTM, SP1, and SP2, it would be interesting to know if you are seeing any change in image quality.2. Non-overclocked performance vs. overclocked performance3. There were two TH2GO systems at FANCON, both of them were supposed to be high end systems, but both were running FS9 because FSX, I was told, lowered the frame rates to unacceptable levels. I am also considering a TH2GO, so I would be interested in seeing your tests results with the TH2GO in RTM, SP1, SP2.4. The general consensus on the forums is that the 3rd and 4rth cores don't get you much with FSX and that you are better off with a faster dual core. I'd be interested in any testing that addresses this. Also is it possible to run 3rd party software like ASX, radar contact etc. exclusively on the 3rd or 4rth cores thereby adding some value to having a quad by not taking a performance hit with the 3rd party software.I'd better stop now...Thanks again, looking forward to seeing the results of all your testing.Ted

Hi Gary,Like Ted, I was out enjoying the last "summer-in-fall" weekend in the Denver area (Boulder). All I can say it's a good thing that I don't ski :)I'm a New Zealander living in the US (now a naturalized citizen, but still retain my NZ citizenship as well). Maybe I've been here longer, but I really enjoy the T/G weekend and the turkey etc., it's fun.On the benchmarks: It's been some time since I have seen a standard set of benchmarks for FS, someone did some when AGP first came out. I think that rather than cater for everyone's settings, just choose a set that represents a workout for your system- so obviously settings to the right-ish side and unlimited fps to capture any results. The key, as I understand it, is getting FS to do something frame-intensive that is easily replicable on any computer, implying stock aircraft and scenery. Meigs Field used to be good for a C172 take-off and a pan to the left to capture all those big buildings. KLGA could be a good bet with the Manhatten skyline that always seems to be the biggest fps hog in any FS version :)I too am buying (not yet installed) a Quad Core Q6600, so would be very interested in any comments you may have- I'm going from an AMD 4800+ X2 on an A8N32-SLI Delux. What chipset (P35 or X38) did you choose, and what mobo vendor/model? I'm still out on how much better the X38 might be, other than possibly DDR3 support.I like your table above, I'm going to copy that for reference. You might consider editing your fsx.cfg to limit the number of cores, from 3 down to 1, which might indicate how well the quad is performing compared to a dual or single core. Just a thought.Thanks- Bruce.

Nice work, Gary.I'm wondering if the fps drop in RTM ties into many people feeling the textures become softer in a shorter radius with SP1?Any chance you could post screenshots or comment on the appearance as far as ground texture sharpness range in RTM vs SP1.

Ted,>1. In addition to the frame rate differences between RTM, SP1,>and SP2, it would be interesting to know if you are seeing any>change in image quality.Other than autogen popping instead of fading in/out in SP1, there is no noticable visual difference between the two releases.>2. Non-overclocked performance vs. overclocked performancePer the data I gave above, there is a 34% avg FPS boost for a 50% CPU overclock.>3. There were two TH2GO systems at FANCON, both of them were>supposed to be high end systems, but both were running FS9>because FSX, I was told, lowered the frame rates to>unacceptable levels. I am also considering a TH2GO, so I would>be interested in seeing your tests results with the TH2GO in>RTM, SP1, SP2.I only did the one run with RTM at the baseline SP1 settings, as it was too slow to seriously consider using in that state and the TH2G was only going to make it slower. Nonetheless, I think you will find the performance drop by going TH2G pretty consistent across all three FSX variants.As you can see by the SP1 runs, there is a drop of 29% for a 200% increase in pixel output. Additionally, I enabled WideViewAspect=true for that TH2G run, which you can see from the separate run I did of just setting WideViewAspect=true that it was already 9% lower as a result, so the TH2G reduction is about 19% itself. This is pretty good IMO, as if FSX were GPU bound it could potentially be a 66% reduction.>4. The general consensus on the forums is that the 3rd and>4rth cores don't get you much with FSX and that you are better>off with a faster dual core. I'd be interested in any testing>that addresses this. Also is it possible to run 3rd party>software like ASX, radar contact etc. exclusively on the 3rd>or 4rth cores thereby adding some value to having a quad by>not taking a performance hit with the 3rd party software.You can see by the second last run that with two cores turned off there is a 3% performance reduction, so that addage the that as long as you have at least two cores to throw at FSX, actual core speed is more important than more cores still holds true. This is for just FSX alone, as I found that with the dual core I used to have that when running FSX with addons like ASX, FSInn, TrackIR, FSCommander, AISmooth and multiple web browsers that both cores would be running at 100% at times, suggesting extra cores would be utilised. I haven't had the change yet to run a full online FSX session, but will report on quad core usage when I do.Gary

Bruce,The benchmark definitely produces consistent and repeatable results, which is why I like to use it. Sometimes I get a stutter during a run that is obviously caused by a running windows process, so I just keep an eye out for such dips and repeat the run if it happens.I managed to get my Q6600 working in my year old Gigabyte 965P-DS3 Rev 1.0 motherboard and it even overclocks up to 3.6GHz, so I am pretty happy with the outcome. I was expecting it wouldn't overclock very well beforehand and was all ready to get a P35 motherboard to unleash it, but it works too well to bother now.One of the runs I did do was with 2 cores only, but I haven't run one with just a single core (RTM effectively only uses one core anyway), but I might do a run with one core.I have been using the 163.75 WHQL driver up until now and unfortunately when I did the first SP2 DX10 run with global high settings, there was a drop of over 30% compared to the same run in DX10, so I am still considering where to go from here. If I change drivers, I want to do at least the baseline runs again to make sure the core benchmarks haven't been notably changed, but all this takes time of course!Gary

Excellent work with this... looking forward to seeing your further results, especially with quad core, as to whether it helps with various add-ons to see if they can be assigned to the extra cores so the main core can run the FSX rendering while the 2nd core handles the data I/O with less distraction.I think this kind of testing will be very important to helping FS users decide whether they need to spend the extra money on a quad versus a dual core if they are a GA "low and slow - VFR only" pilot and will not be flying heavies with other running special modules such as you mention in your post above.Also, it would prove helpful in purchase decision making for people to know if the Matrox TH2G (Triple Head To Go) triple monitor attachment device will benefit from having the 3rd or 4th core available in a quad core CPU versus a dual core CPU system with FSX, so that the main FSX rendering and data I/O processes can be unburdened while working on their own CPU cores.I don't know much about the TH2G, but I gather from your info above that it exacts a toll on system performance, therefore uses the CPU to do its grunt work rather than inside its own external hardware?Can its software related requirements be assigned to the 3rd or 4th core to avoid competing with the main FSX activity on cores 1 and 2?I also wasn't sure what you meant by the statement: "the TH2G reduction is about 19% itself. This is pretty good IMO, as if FSX were GPU bound it could potentially be a 66% reduction"... could you explain?A lot of folks are trying to plan a more FSX capable computer upgrade purchase before or during the upcoming holidays, and this could help them decide on whether it is within their budget if they can know if their type of flying will really need a quad core expenditure right now (before FS11), or if for their personal FSX flying style they can better (and sooner) afford a system ASAP with an even more highly overclockable dual core architecture.Thanks very much again for your work on this, and I look forward to your next report! :-) GaryGB

Thanks Gary. To echo the others here, this is excellent work, well done. Looking for when you post your next batch of data.Bruce.