FSX SP2 - Blurries - Any Real Benefits of Limiting Frame Rate

[FlyingsCool 3]

Hi Phil,I find it amazing that you still spend time to read in detail through these forums and respond to people's comments, even though it's basically different people repeating the same things that have been said before (and often "wive tales" that I don't bother to pay attention to anymore (not a comment on anything said here, I haven't read it in detail nor put much thought into it)).Thank you Phil,Thomas[a href=http://www.flyingscool.com] http://www.flyingscool.com/images/Signature.jpg [/a]I like using VC's :-)N15802 KASH '73 Piper Cherokee Challenger 180

[PhilTaylor 45]

There is a lot of misinformed and cross-linked data here.Reducing global texture resolution, mesh texture resolution, mesh resolution, scenery density, and autogen density all help reduce bus traffic. As does turning off bloom, shadows, and water 2.x since that removes extra passes across the scene. When either the graphics memory is saturated ( causing texture swapping ) or the bus is saturated ( causing amazingly bad bus behavior in multicore processors ) the blurries can often result as can stutters. As far as the issue wrt bus traffic, here is a telling quote from an Intel document:"The utilization of the front side bus is a performance concern because latencies of the memory and address FSB will vary with the amount of traffic on the bus. Graphing the FSB utilization vs. memory latencies shows that memory latencies increase at a rapid rate after ~60% FSB utilization. Latencies then go infinite as the FSB bandwidth passes 70%."This is specific to multicore FSB performance, and it certainly isnt the performance curve I would expect from the hardware. This is certainly contributing to the problems and reducing settings to get bus traffic below 60% is the only cure.Perhaps the 1600Mhz FSB will help us, as it will move the line by about 25%.Note there are plenty of threads about new hardware removing the blurries, as in:http://forums.avsim.net/dcboard.php?az=sho...ing_type=searchhttp://forums.avsim.net/dcboard.php?az=sho...ing_type=searchI am sure there are more as I did about 30sec of browsing through the search results.So how can you claim updating the hardware does not help? I am puzzled by such statements in the face of other community posts to the opposite. That does not mean faster hardware makes it impossible to get blurries, as overloading any hardware is going to result in blurries, just that updating the hardware is really helping people.And there are quite a few people posting images claiming they are blurry where many of us do not see it. So there is a lot of misunderstanding of what blurries are, how one gets them, etcFurther, SolarEagles' thread about texture loading herehttp://forums.avsim.net/dcboard.php?az=sho...ing_type=searchindeed talks about flying at high speed without blurries. The sim does not allow velocities of 20,000knts so I assume that is a typo, but SolarEagle shows 1300knts in the FA-18 so that is close to the scenario you pose. And he shows success. If what you claim is true, then what SolarEagle claims should never happen. How to reconcile these two 180-degree-opposed claims?So on the one hand we have success stories; and the number of them seems to be increasing. And on the other hand we have some failure stories; which seem to be constant based on older hardware.Based on these data points, it is hard to see how a credible analysis would show this as anything but a local problem since:a) some people never had blurries:( some people who had blurries and updated their hardware then do not get them. c) some people who do not update their hardware still have the blurriesNote my response in the SolarEagle thread talks in a bit of detail about exactly what SP1 did and why. So there is a solid reason behind the behavior change from RTM to SP1 and we have been clear about that since SP1 shipped. Further, in SolarEagles' thread I stated I have talked to one of the leading photoscenery vendors and they do not see a global blurry problem with the sim rendering engine. If every single user of FSX had the blurries that would indicate a global problem. But we do not have that result as shown on this forum. If new hardware did not make any difference in the blurries that would indicate a global problem. But we do not have that result as shown on this forum. And we have leading scenery vendors showcasing products ( VOZ, Horizon VFR ) without global blurry issues. The theory must take into account all data points when attempting to make a curve fit the data points. These data points all lead towards conclusions different than you are espousing.As far as Chicago or any other large city, these areas have their own issues with FPS that is separate from texture loading and the combination is undoubtedly causing an issue. Reducing settings and reducing the workload is all you can do.In closing, I see plenty of innovation in 3rd party products for FSX. And I thought FEX was FSX specific and offers, to quote:"The program adds revolutionary control of your environment, all of it in stunning high-definition detail. This version is also very different compared to the original release for FS9."So I dont see that comparison between FS9 and FSX as fair. However, if you are more happy with FS9 please do stay with that.

[Bert Pieke 7,584]

Phil, I've got an incomplete understanding of how FSX uses multiple cores and I have no evidence to show that AffinityMask=14 performs any better than no AffinityMask on a Quad processor, but -With no Affinity Mask, the Task Manager shows Core 0 at 100% most of the time, and the three remaining cores swinging between low activity and high activity.With the mask at 14, Core 0 shows very little activity, Core 1 is at 100% most of the time, and the remaining two cores swing between low and high activity.So, it would appear that the main load has moved from Core 0 to Core 1.Is the "fiber system" in your note not a heavy user of the processor?

[PhilTaylor 45]

The fiber system is significantly off-loaded since SP1 but does still perform work.

[Bert Pieke 7,584]

OK, thanks !