As with any FSX user, tuning and tweaking is an ongoing process. this is simply because, its rare to be able to lock down an FSX instance and never have to touch it. there is always a new aircraft to install, a new airport scenery to apply, the next revolution of weather engine that simply can’t be passed. Given this, while it is worthy to think of locking down a well working instance of FSX, it is not practical.

There is a long slew of FSX tweaks that have been discussed, debated, argued, accepted and rejected. One of them is BufferPools.

This is a partly understood tweak and even though I have tried working with this for many years I am yet not convinced that I have it set optimally. Any discussion regarding FSX tweaks and results MUST BE in the context of the configuration that the simulator is built on. Here is a little about my configuration – i7 3770, 8GB, nvidia 660 GPU/2GB. Triple projection rendering 3840×1280 pixels.

Firstly BufferPools is best set as the section header in fsx.cfg file. UsePools is the setting. This is typically applied along with the RejectThreshold parameter.

I have tried this with Nvidia 580 and Nvidia 660 cards.
My best mileage has been achieved with UsePools=0. For some reason, Nvidia cards don’t seem to like the BufferPools model.

I see a average frame rate drop of at least 4-6 frames/second each time I try to use the parameter – with or without the RejectThreshold. And before the question comes up as how i measure FPS, I want to clarify that whichever method is used the impact is relatively the same.

On the other hand, I don’t subscribe to the view that we should be increasing the water shader parameter to HIGH merely to force FSX to leverage the BufferPools model.

So if you are using an nvidia card! especially the 500 or 600 series, I would say leave BufferPools at 0.

As always, YMMV.

CPJ

This blog is primarily about the simulation hobby. However, from time to time I am tempted to post some of my travel experiences. Here is one to that list…

Long-haul travel can make one weary, it also provides for very unique experiences.

My most recent trip took me to the Eastern Hemisphere. On the way back to the Midwest, travel routing called for a flight over the Orient.

Leaving Singapore in the wee hours of the morning, it was a treat to watch the sun come up on the horizon as the 747 dash 400 rolled down runway 02 Center. The Pratt and Whitney engines are a marvel in themselves. They make lifting 875,000 pounds look easy. The dash 400 is another story – when the ‘Queen of the Skies’ rotates for takeoff it makes one wonder if the airplane is flying fast enough to even take flight.

Making a straight-out departure north-east, we left the Straits behind. Flying over the South China sea, we flew past Kuala Lumpur and then Ho Chi Minh city to arrive in Hong Kong 4 hours 20 mins later. Change of planes.

Less than hour later, we were rolling down Hong Kong’s Runway 28R for the long flight to Chicago. 7779 miles/~13000 kilometers – planned flight time 14 hours.

The flight plan would take us past Shanghai, China and then over the North Pacific past Osaka and Tokyo, Japan. Past that it was blue ocean beneath all the way to the 63rd parallel. Turning east here we made landfall on the Alaskan shoreline, past Juneau into Northwest Canada. With Saskatchewan past us, it was direct Edmonton, over North Dakota and Minnesota we flew into Chicago.

We had been about 8 hours into our departure from Singapore at sunrise and it was beginning to get dark. The light had begun to fade. What a strange feeling.

The longitudes had been incrementing all this time. We had gone past the 170 East longitude and soon crossed the international date line. The longitudes will now begin to decrement, The time zone clock had been advancing all along. We were UTC+12. Now in a flash, we were UTC-12!

In about 5 hours from that point the sky began to turn to early morning blue. Soon the orange glow appeared and within minutes a bright sun filled the sky.

Within 10 hours of seeing the sun come up, I had seen it turn dark outside and then witnessed the sunrise again! Speaking of modern air travel shrinking the world, it even shrinks solar cycles.

These experiences make one realize the vastness of the planet we live in and laws of nature that rule.

It is humbling in many ways and makes us understand what an infinitesimal part of this ecosystem we really are….

CPJ

Continuing from my last post… having been unable to do away with the Matrox appliance because I couldnt get 3 outputs to work from the 660 with the XGA projectors – the Matrox Th2Go is still in place.

Now for some results…

The first step on the physical install was that the system asked for the install of the driver that came with it – that driver is WHQL 305.27. This is important, Please note it because of a point that comes later in the post. The driver uninstalled the 304.yy driver and installed the 305.yy driver successfully.

Rebooted the PC.

The nvidia auto-updater asked if I wanted to update it to the latest version – WHQL 310.yy and performed the update.

Started up NThusim and FSX. Everything works well- till now!

Here is where the problem began – didnt feel like everything was like before. Images were not sharp. Blur scenery even on ground. Aircraft textures were bad. Felt like my old 800×600 monitor and PC XT in the 80s.

The graphics card – one of the best in its class. The driver – the most recent.
Then what could be wrong??

I didn’t want to adjust anything on the projector. However I tried a few things just to be sure..
No improvement.

Restored the 304.xx driver.

Came back to basics of FSX tuning – started with the FSX Config File.
Started by recovering an old baseline FSX.CFG file. Went through all of the routine mods that had to be placed.

Experimented with texture size parameter – went back to 2048 rebooted FSX and came back repeated that with 4096 later.

Tried reworking the LOD parameter. No improvement

Reduced the TBM number back to default and then increased it upwards. From what I have read and understood, TBM acts a throttle of how much is being pumped into the GPU.

And finally I think the right match between TBM and FIber Frame Fraction did the trick. The blurs went away – in heavy airports and in the sky.

I also noted that this combination also helped me achieve something that had eluded me for quite sometime – the max load in the GPU. I had never seen that load number go beyond 20% – even in heavy airport or significant WXR. With this match I now see my GPU being used a whole lot more – spikes reaching 87-90 percent and an average load of 60-70 percent.

The average use of VRAM went up a little bit – with a lot of traffic and heavy airports it uses about 800-900MB up from the standard 500-550MB it would consistently use.

My FPS (Nvidia FPS gauge) went up a little in very heavy airports like Aerosoft EDDF or FSDT Zurich (even with Jetways operational). Used to average 22fps, now get around 27-28 fps on the ground in heavy airports.

While the other tweaks may have helped get rid of the blurs, my conclusion is that the 310 driver caused the blurs.

If you choose to use it, please backup your config files and the old driver installer before installing the 310 driver.

CPJ

After trying the GTX 560Ti, (thanks to the folks at Jetline Systems), I was in a position to try the GTX 660.

The 600 series cards have gained a lot of coverage due to their ability to provide video surround without the need for a Matrox TH2Go appliance. The 660 for example likes a space shuttle and and its cooling fan has the appearance of the turbine blades on a 737 engine! I think this makes it even more apt for the flight sim hobby.

The 660 is a GDDR5 device and comes with 4 ports. The one I had came with 2GB of video RAM. Three video devices can (in my case projectors) can directly plug into the card without needing the Matrox device. In any case, this was the plan.

However, upon opening the box, I discovered that the 4 ports were all of a different type. There was one of each of the following – DVI-D, DVI-I, HDMI and DisplayPort.

Here is where I hit the first snag – each of these, with the exception of the DVI-I port, I needed adapters that would convert to DVI-I or directly to VGA – because VGA connectors are the only ones that the XGA projectors have.

Amazon to the rescue, ordered those connectors. They took thier own time to show up. One arrived but didn’t work, another came with the wrong part being sent and the third came all right but wouldnt was too wide in its moulding that it wouldnt fit.

Anyway, a week later, had all the right parts. Plug them all in.
two work, the other two wouldnt. I only need three so I could manage with DVI-I to VGA, DisplayPort to VGA. I needed one more. The one that never came to be.

I tried different adapters. The projector wouldnt light up even though the 660 detected the third device. Back to Google for more research and reading. While I dont have a conclusive answer as to why, I am now sure that while there are adapters to convert DVI-D to DVI-I or VGA, HDMI to DVI-I – they wont work. So my limited knowledge on the subject, it is clear that there is no passive method to converting an Analog RGB video signal to a Digital RGB signal. It requires an active converter. So please watch for this. If you dont have a video output devices with digital ports on them, the 660 wont work for you.

With the native ports not going to do it for me, I was back to plugging the Matrox Th2GO device into the DVI-D port on the 660. That worked okay however it nullified the primary need for the 660 – i.e. to eliminate the need for a Matrox device in the ecosystem.

Will describe the results in my next post…

Here we go again… yes… for reasons beyond the scope of this topic, I went down the path of replacing the GTX 560/2MB with GTX 560Ti/1MB.

Here are the basic specs on the two cards. Essentially the distinctions are small between them. Albeit, the 560Ti is an improved and optimized version.

I had been running the 304.91 WHQL driver suite all along since June 2012. No issues. That was not the reason for the replacement anyway.

As mentioned before, its important to describe the setup, because way too much is written in forums and blogs about results without first providing context. The context matters – all results are only relevant within a context.

Mine is a triple projection setup, running on a ultra-high-end Jetline Systems rig.
A i3770 – 3.4Ghz processor, OC’d to 4.4GHz with 8GB of RAM.
There are 3 undocked windows on FSX and a fourth to provide the main, making it 4 windows for FSX to draw. I run 3072×768 @ 60Hz because the native on my XGA projectors.

Here are the results.
The 560Ti uses the same driver that the 560 used. I didn’t expect any major improvements.

For running a pseudo benchmark, I use a standard flight that I have setup when I first started with fsx. It points at a main terminal in a heavy airport with FSDT scenery. I usually run it for 10 mins on the airport, without traffic, then with 50% traffic and 100% traffic.

With everything running the same on the MS FSX setup (I mean in the fsx.cfg file) and in the nvidia inspector, the most improvement I got were in the average fps count – not by a lot, a 2-3 frame improvement.

I did notice though that in flight at 5000 feet (i.e. approx 2500AGL) my textures were a little more sharper than in the 560. The GPU did run a little warmer than my 560. The average GPU use hovered around 45% with occasional spikes to 70%. The memory usage were level around 475-517MB.

Not much to speak about with the 560Ti. On the other hand, I didnt know if the 1GB less memory that the 560Ti had would make any difference to fsx use because I have never seen memory use go beyond 520-550MB in most of my sim use.

I also have a nVidia 660 in my hands now and will be testing that in the coming week.

CPJ

Happy New Year to everyone.
Wishing every one the very best that 2013 has to offer.

Simulation Research

The recent finding around Amelia Earhart’s disappearance has placed renewed emphasis on the topic of LOP. In pursuing the topic further, i came across many articles, points and counterpoints regarding LOP, the use of octants and more.

Amongst everything i read, the article referenced in the link below caught my interest. It brought to focus the various facets of traditional celestial navigation.

For anyone interested in aviation in any form, navigation is an integral part of such interest. Thought I would share…

https://sites.google.com/site/fredienoonan/discussions/why-it-was-not-possible-to-follow-lop-to-nikumaroro

CPJ

 

A lot of my recent simulation research has been focused on improving visuals or haptics of larger simulations. Shifting focus a little, i thought its time to devote some effort to flying GA. When it comes to General Aviation, my favorite is the Piper Cherokee family. The Archer or the Warrior would be my pick. Hence i set out to find a realistic Warrior or Archer. I first downloaded a freeware Warrior download for FS9. The aircraft turned out to be reasonably true in looks and appearance. Soon i found that flight model was not true to the real aircraft. The first discrepancy was the P-factor. It made taxiing very difficult and unrealistic. Even at slow speeds, smooth steady taxi was impossible. In flight, the plane had a serious left yaw. I tried centering the controls. It was still there. Was unable to find out whether the rudder had a center issue or the ailerons were not centered. In the end, i figured it was still engine center-lining that was causing the yaw.

The search had to be extended. A quick search showed a few products. Carenado, Australian Simulations, JustFlight were a few.

Aus Sim product is known to be a very good product and has numerous reviews. However, the company closed out operations and hence the product wont innovate any further.

Carenado is a very active with products. New products, incremental releases and rich variety are its hallmark. Their products are packaged independently for FS9 and FSX which is a little bit of an issue for those of us who use the product for either simulator. Carenado has an Archer product for FS9 and FSX.

JustFlight has a Warrior product packaged for FS9 and FSX combined. The product was on sale and it was my choice.

My larger interest in simulators is more attuned towards accuracy in procedures and flight characteristics. Given this, I am a little less concerned about the livery or about the door opening correctly.

The install was flawless and simple. The product came up on FS9 immediately without any more effort. I setup for an initial flight. John Wayne airport, the SOCAL area, runway 19L. Turned off all WXR and ensured winds were set to zero.

Taxi was very accurate. P-factor representation is well done. Rotation speed matched real-life behavior. A mild tug of the yoke and the plane began to climb. Climb was around 550 fpm. Prep for level flight – pitch, power and trim. Level flight was easy to maintain. Very life-like.

Performed a cruise climb at 87 knots. Leveled out at 2500 feet. Cleared the area with clearing turns. Performed slow flight, power off stalls, power on stalls, and then steep turns. The flight model performed i would say within 10% of the characteristics of the real PA28-161. Not bad for a simulation product flight model.

Returned back to John Wayne. Performed cruise descent, approach, GUMPS check and dropped engine down to 1700 rpm abeam the numbers with one notch of flaps. At the 45 to the numbers began my base turn; another notch of flaps and plane slowed accurately to 70 knots. Called my final and dropped the third notch of flaps half way into the threshold, 65 knots all the way down to the numbers. Just over the numbers, leveled out, the nose came up nicely and i held it up just to the point where i could transition my eyes to the end of the runway. Continued to hold it up till the stall horn came on and the wheels touched. It was nicer when i could hold the nose up even after touchdown till the plane slowed further and the nose wheel came down and touched.

Will come back another day to check flight characteristics under density altitude and various wind conditions.

Taxied back to parking…. Glad to have found a realistic model of my favorite GA plane.

CPJ

Aug 30 – Edit & Update – The app works seamlessly. Just enable the device on the X-Plane settings page and then turn ForeFlight on your iOS device. The location occurs instantly. Very straightforward.
———————-
The ForeFlight iPad app is now compatible with X-Plane 10. The integration is interesting because it brings a simple way to interface with a hardware device. More on this after I try using it when i fly.

SimulationResearch

Flight Level 370, 0640Z, Queen of the skies.
Flying over Dusseldorf, Germany en-route to Frankfurt.

The sun is just about to rise….