FSX-MS How Addons Perform Relative to C172 w/Graph

[ryanbatc 12,888]

This is my take on a user who did a graph a few years back.  He called it the Petraeus Index.  Basically this graph shows the performance loss against the default C172 VC.  All test were done with fresh reboots, FSX settings no AI traffic, no weather, daytime at noon, vsync disabled, unlimited FPS.  Once the aircraft is loaded I spun around the cockpit once and made sure all the avionics were on, and also spun around the external view once.  Mouse cursor moved off the screen as well.

 

Tests are all done at KEDW Ry 4, sitting on the runway.

 

It doesn't show FPS, but it could be used to determine what sort of FPS loss one might get if you paired up your addon vs your default C172 VC.  You can easily test yourself on your own PC.  Start with default 172 in the VC, spin around inside and out, make sure avionics are on.  Note your FPS.  That is your baseline or 100%.  Load another aircraft, wait for the FPS to stabilize or take the average of the different bumps as the FPS vary.  Divide your fps compared to whatever your stock 172 gets... that's your PI 

 

To read the chart you just look at the percent, for example: Baytower RV7, it's about 66% - so I lost approx 34% of my FPS compared to stock 172.

 

note: Some of the info was compiled by other users

 

[Bjoern 1,157]

I remember the Petraeus Index. The Super 80 was actually the best compromise between systems modeling and framerates. IIRC, it performed even better than the default aircraft that served as the comparison baseline.

 

It would probably be better if add-on aircraft would be compared against their default counterparts. Airliners against airliners, multi-engine GA against multi-engine GA, etc...

[ryanbatc 12,888]

Maybe but everyone has the 172 so they can compare against that.. And it makes everything more uniform

 

I don't even have a lot of older addons installed... Maybe I can scrounge up the data on them and add it

[Luis Hernandez 575]

Thanks for refloating this, Ryan. Did you use windowed or fullscreen mode? I wonder what would be the Petraeus Index of the iFly B744 v2...

[fullmetal02 23]

I think this is brilliant. But is the trike a better base line?

[ryanbatc 12,888]

Windowed mode

 

Any default plane would work as long as the comparisons are done on users systems

[fakeflyer737 992]

The Mustang is way way up there, I always thought that one was pretty hard on the FPS.

[gizmosellsbunnys 106]

Good idea, and any plane getting less than 65% of fps default c172 is likely no buy to some users. Why pay $50.00 for plane get low fps.

[ryanbatc 12,888]

The Mustang is way way up there, I always thought that one was pretty hard on the FPS.

With a PI of 47 I wouldn't call that way way up there... Meaning it loses 53% of the Fps compared to default C172 - were you looking at the C206 instead?

 

Two other things this Index doesn't cover:

Framerate Variance (stutters you'll likely see)

VAS Consumption

 

With my older data tests I did run VAS measurements maybe I can post that - should be more helpful. A lot of users can run all those planes listed but throw in complex scenery etc and you make get an OOM

[fakeflyer737 992]

With a PI of 47 I wouldn't call that way way up there... Meaning it loses 53% of the Fps compared to default C172 - were you looking at the C206 instead?

 

Two other things this Index doesn't cover:

Framerate Variance (stutters you'll likely see)

VAS Consumption

 

With my older data tests I did run VAS measurements maybe I can post that - should be more helpful. A lot of users can run all those planes listed but throw in complex scenery etc and you make get an OOM

 

Well on my system I get better FPS in the NGX then I do in the Mustang, the Mustang kills my system. 

[ryanbatc 12,888]

There's always going to be a bit of that.

 

Ideally a bunch of users would send me their tests and we'd average them out .

[Bjoern 1,157]

Maybe but everyone has the 172 so they can compare against that.. And it makes everything more uniform

 

C172: 53/79 drawcalls (ext/int)

A321/B738: 164/~60 drawcalls

CRJ7: 129/55 drawcalls

C208: 101/125 drawcalls

B744: 270/68 drawcalls

F-18: 255/375 drawcalls

LJ45: 103/30 drawcalls

Trike: 17/12 drawcalls

DC-3: 121/73 drawcalls

EH101: 99/261 drawcalls

 

Drawcalls aren't the key to model performance, but they indicate the complexity and performance impact of a particular aircraft rendition pretty well. Theoretically, you'd also need to account for glass cockpits as the default G1000 comes with a slight performance hit even in a featherweight like the C172.

 

So by expanding the comparison to the nearest default equivalent, you'd cut the 3rd party addon developers some slack. No one can be expected to make a complex 747 perform as well as a simple C172.

 

 

That said, feel free to continue your chart though. Just make clear that the default 172 really is only present to provide the 100% mark and that the true comparison is one 3rd party model against another.

 

Now look at that bloody EH101 and F-18...they're absolute hogs! And default! D:

[Jimmy RFR 406]

I've done something like this before, but for slightly different reasons.

 

I established a baseine with a few aircraft ranging from the default 172 to the F1 KA, and used it to gauge system performance of payware airports.  That way I kind of knew which airports I might not want to venture to with certain aircraft.  ^_^

 

The one thing I found, when testing a variety of aircraft, is that you almost need a heavy scenery load to actually force differences to show.  At a default airport in the middle of nowhere, for example, I can't tell much of a difference between the default 172 and the Lancair Legacy.  But you toss in a heavy payware airport, and things become very noticeable.

 

Basically, this chart has more or less meaning depending on your system capabilities and where you like to fly.

 

Edit:  I should also qualify that opinion by saying that for a long time now, I've only ever run fsx locked at 30 inside inspector (I think? It's been forever since I've felt the need to tweak).  As a result, things boil down to a single question: Can I maintain 30fps, with this plane, in this scenery?  Very, very seldom these days do I ever have to answer 'no' to that question.  And when I do, my solution is typically to adjust the scenery load to fit. Therefore I end up ignoring the differences in performance between one addon aircraft to another, because they are no longer relevant. To me.

 

Which in no way makes it less important or relevant for anyone else!!

[Petraeus 147]

I remember the Petraeus Index.

 

 

So do I! Sadly, the results got trashed when the AVSIM site was hacked.

 

Thanks, Ryan, for reviving this approach. I've been tempted to restart the project, but using looking at VAS rather than FPS. With better hardware we can buy better FPS, but sadly not better VAS.

[ryanbatc 12,888]

Still here huh?

 

I had a chart I did with vas but it's been a while.

 

Rob did one over at p3d forum