Last week we sent out our first build of the Vulkan/Metal ready X-Plane (which will be version 11.50) to an external sight, and last night we sent out the first private beta.
Our plan is to move to an open public beta soon once we have some feedback on how well the build is working. When that is will depend on the bug reports; I’m crossing my fingers.
How It Works
In X-Plane 11.50, there is a check-box in the rendering settings that enables Vulkan or Metal; you’ll have to restart for the change to take effect. If the sim crashes really early (e.g. before the main menu) it will revert to OpenGL to keep you from getting “locked out” but random crashes while flying won’t do this. There are also command line overrides to control the driver for people running performance testing scenarios.
I should also mention that we’ve changed what the reflections slider does – it’s now like X-Plane 10 and does not attempt to update the environment cubemaps in real-time. (If you really like flying at 5 fps, you can re-enable this in settings.txt – the underlying tech has not been removed.)
What we have found over the last few years is that users will drag every slider all the way to the right, and then be surprised that the sim has poor performance. While I have in the past dismissively claimed that they are doing it wrong (“don’t put every topping on your pizza”) the inevitable truth is that we are violating an industry UI norm by having the highest possible settings run beyond the scope of the fastest possible computers our users can get their hands on.
Now that we have Vulkan and Metal running, we can make some judgments about what will and won’t be real-time. Full scenery shadows are still available and are helped by the Vulkan and Metal driver; a decent machine should be able to run full autogen + scenery shadows at 30 fps.
The Vulkan and Metal back-ends change their approach to running out of VRAM. With the OpenGL driver, when VRAM is exhausted, textures are moved to system memory, and the application stutters while the driver makes this mad last-minute shuffle.
When running with Metal and Vulkan, X-Plane will reduce the resolution of your textures while you fly, in the background, without stutters, preferring to reduce resolution on textures that are not being used right now. The result should be smoother and generally less disruptive. You can still lower the overall texture resolution to better fit your card’s budget.
This new “on the fly paging” is…well, it’s completely new, and I suspect it will need a lot of debugging during beta. It’s an area where we just need to collect use-case data to better tune the algorithm.
X-Plane 11.50 no longer allows for uncompressed textures – if a DDS is present on disk, we will use it. I have to recommend texture compression in the strongest possible terms; while the artifacts are annoying, when we turn it off, everything looks worse due to the VRAM pressure – uncompressed just isn’t a good way to use our VRAM budget. Pre-compress your textures for best results!
A Detailed View
One of the great things about running with the Vulkan/Metal back-end is that there isn’t any dark matter in the rendering path – everything that takes time takes that time when we say so. This means that when we have a stutter we can take the frame apart and figure out exactly why that happens.
One of the last fire-drills before sending out a build happened last week – with the Vulkan back-end and memory paging working correctly, we would still see these violent stutters flying over downtown LA. As it turned out, the code to delete autogen as you flew away from it was too slow and was causing hitching every time a 3 km AG square had to be torn down.
That code is now rewritten and the results are much smoother. Here’s what winning looks like:
This is my old iMac running at 60 fps – it can’t do that with everything maxed out, but when it does hit 60 fps you’ll note the dead even orange bar at the top – that’s consistent framerate. The bottom of the screen shows a block diagram of some of the tasks that went into the current frame render.
With “weird driver behavior” removed, when something stutters, so far we’ve been able to identify it, e.g. the autogen stutters from last week. Here’s what stutters look like:
The orange spikes you see are the stutters – I induced these by command-tabbing to other apps while flying; when the window manager swaps apps it’s pretty disruptive. That big blank space in the middle of the screen is time spent doing something (probably in the window manager) that we do not yet track.
One last show-and-tell pic: for some reason when running at 45 fps, the sim is actually running at a mix of 60 and 30 fps every other frame, which is not good. We can easily see this with our tools:
Here we can see two frames, a long one on the left and a short one on the right. The difference: “acquire surface” is taking a ton of time. The long frame is stuck waiting for the window manager to give us some VRAM to draw into. With this kind of detail, we can at least understand the bug and try to fix it.
Next Steps
At this point the only thing left to do is fix bugs, some of which are still open even as the private beta goes out. We will move to public beta once we have something where we think the initial experience isn’t a blue screen and tears.
X-Plane 11.40 is now final! You will be prompted to update to X-Plane 11.40 when you start X-Plane; Steam users will receive the update automatically via Steam. Here’s what’s next in the pipeline:
We are working on a bug-fix update (11.41) to catch one or two bugs that didn’t make the RC, as well as the inevitable bug that will be reported after go final. I expect to cut an 11.41 release candidate some time next week, and it should be a pretty quick release.
In the meantime, we are pushing hard to get Vulkan/Metal ready so that we can do an X-Plane 11.50 beta. We may start private testing of Vulkan and Metal before 11.41 is done, depending on what gets fixed first.
The limiting factor on getting to a public Vulkan/Metal beta will be bug fixing – it doesn’t make sense to go public beta with known bugs that will be reported over and over and make the beta unusable. There are also two remaining areas of development we need to close up:
Load Times
Vulkan and Metal fight stutters by pre-loading everything they need to render the frame – this means whatever we need to draw, it’s ready. But this also means X-Plane needs to prepare every shader it might need to render a given aircraft and scenery pack.
Because X-Plane’s shaders are so flexible, this can mean tens of thousands of shaders, and that hurts load time. We have new code that tries to analyze exactly what we’ll need for a given frame. For example, if an object comes from a DSF, we never need an “interior” shader for it because the scenery object can never be inside an aircraft. By carefully applying this kind of logic, we can cut down the number of preloaded shaders and speed up load times.
Sharing Memory
With Vulkan and Metal, X-Plane manages its own use of VRAM – this is a totally new capability that we have had to code from scratch for the Vulkan/Metal port. We watch our total memory usage and reduce texture resolution dynamically (hopefully on less important and less used textures) to stay within budget.
While this code is working, it hits a wrinkle on Windows:
Third party add-ons that chew up VRAM via their OpenGL VRAM usage can use up part of our budget. For example, if you change to an aircraft with a plugin that allocates its own textures to render to, we can’t use that VRAM anymore.
The Vulkan version of X-Plane allocates VRAM in big blocks for efficiency. If we reduce our texture res to make room for the plugin, we might end up with a bunch of big pre-allocated blocks (that are mostly unused) inside Vulkan – the memory still isn’t usable by the OpenGL plugin.
Sidney is working on compacting our textures so that some of the big blocks can be given back to OpenGL – yet more new memory management code for Vulkan.
There Be Dragons
Once we have the number of shaders cut down and memory compacting working, we’ll begin heavier testing of the Vulkan/Metal build – I can’t say how close to public beta we are because we haven’t gotten the initial wave of bugs from this testing.
This post has been on my todo list for a while – long enough that X-Plane 11.40 came out before I had time to write up a post saying “X-Plane 11.40” is coming. But just to put 11.40 into context, here’s what our patch roadmap looks like for X-Plane 11 this year.
All Physics All the Time
X-Plane 11.40 is a physics release. Almost all of the changes in X-Plane 11.40 come from Austin’s work on the physics engine over the last six months. This is a new approach for us. In the past, when we’ve updated the physics or systems, it would be in a giant “omnibus” release, where everybody’s latest code went out at once (e.g. X-Plane 11.10).
The problem with the omnibus releases is that they would take forever to get debugged. With so many people changing so many things, we never knew what had gone wrong when a bug report came in. And with all of the code changed, we had to investigate every single bug report carefully (no matter how unlikely or vague the report) because anything could have been broken.
So far, at risk of jinxing the beta, it appears that the physics-only approach is working a lot better. It has been quicker to find bugs when they are reported, and the overall level of crazy is a lot lower than in past releases.
NaNNaNNaNNaNNaNaNNaN Batman!
There aren’t many open bugs left in the 11.40 beta, but one particular bug has caused the beta count to run up: we were seeing crashes due to NaNs in the flight model.
NaN stands for Not A Number, and it’s what you get when you have divide-by-zeros run amok in the physics. To catch them, we’ve turned on a lot of auditing code and we’ve been collecting automatic crash reports. At risk of jinxing it, I think Austin has fixed one of the two root causes in beta 8. We are going to keep chasing them until the other one is fixed, then turn down the checks once we’re done. So we may make it to beta nine or ten and we may have another week with two betas; the high tempo is just to get more checks in fast.
Experimental Physics
There have been a number of questions in the comments on the state of the experimental flight model, so I want to clarify how it works and what is happening in 11.40.
Normally, new X-Plane features get beta tested during the beta of an X-Plane patch. This means we have somewhere between two and eight weeks to debug the feature and get it ready to ship. Once it ships, if we change the feature, we have to consider how this would affect authors using the feature and whether it would screw up their add-ons.
That’s not a lot of time to debug! In particular, it’s really not enough time for the flight model, where people need months just to develop the aircraft and measure the performance to get us feedback.
The experimental flight model is basically a giant year-long open beta of a future revision of the flight model that hasn’t shipped yet. By checking the “experimental FM” box, you’re getting to beta test the flight model of the future, now. By keeping the experimental flight model as an experiment for so long, this frees Austin up to simply fix bugs and improve it, as opposed to worrying how the X-Plane 11.40 experimental FM changes will affect X-Plane 11.30 users.
To be clear: there is no attempt at backward compatibility between the experimental flight model from one sim version to another! The goal is to have the experimental flight model not interfere with the “normal” flight model at all.
Physics changes in 11.40 fall into two broad categories:
Simulation changes that can change how an aircraft flies, improving the accuracy of X-Plane’s predictions about the airframe. An example of this is the delay in wash propagation from the prop to the tail of the aircraft. This makes the aircraft more stable, but may change how it flies; an author might have added artificial stability or reduced control surface efficacy to work around the lack of this feature in the past, and these work-arounds would be inappropriate with the new, more accurate physics.
Simulations that change how the aircraft flies in unusual circumstances that you can’t tune your aircraft to. Examples of this include stalls and wake turbulence. There isn’t going to be a book value for the effect, so all we can do is try to produce the most sane results given an aircraft that simulates properly in regular flight.
Features in the first category require the experimental flight model to be enabled, while the second category of features is always on.
At some point in the future, the experimental flight model will become the flight model for X-Plane, but we are not there yet, and we are not planning to do this as part of making X-Plane 11.40 final.
Third Party Aircraft
If you develop a third party aircraft, you need to test it now against both the experimental and non-experimental flight mode. If the non-experimental flight model doesn’t fly the same as 11.36, please file a bug, and please provide flight testing details. For the experimental flight model, you may see book numbers change a little bit; the real question is whether the overall physics response is better or worse.
Vulkan and Metal
X-Plane 11.50 will be the next major patch once X-Plane 11.40 is out of beta, and it will feature Vulkan and Metal support.
The marketing guys showed the Vulkan build of X-Plane live at Cosford last week; that build did not have any support for texture paging. Since then, Sidney has a basic texture paging implementation running, so hopefully we’re in good shape to get this into developer’s hands after 11.40.
Our expectation for add-on compatibility is:
Add-ons doing supported things, like 2-d panel drawing and UI should just work in Vulkan and Metal – we’ll take bug reports to fix compatibility issues.
Add-ons doing unsupported things won’t work in Vulkan and Metal at all. Your 3-d drawing callback won’t be called, or your attempt to grab internal GL resources will just fail (because none of our resources are GL).
X-Plane running under OpenGL should “just work” for pretty much every add-on, including ones doing sketchy things, and should be faster than 11.40 but not as fast as Vulkan or Metal.
I expect the Vulkan beta to be a relatively long one. We want it to start this year, but it probably won’t end this year, and my guess is that initially Vulkan will be fantastic for some users and will crash for others. During the beta we’ll gain useful information about how well Vulkan works “in the field” for different cards and drivers.
One reason I am looking forward to the Vulkan beta: we now have tremendous visibility into what the rendering engine and driver are doing. With OpenGL, the driver was often a black box. We still get reports of “the 3-d mouse in VR make my machine really slow” and frankly, we may never know why this happens to just some users and not others with the same hardware, drivers, and version of X-Plane.
With Vulkan and Metal it is going to be different. A lot more of the graphics work happens inside X-Plane, and the work that happens inside the driver is much more predictable, bounded, and can be viewed via modern profiling tools.
So while we will have a lot of debugging to do based on user feedback, it should be straightforward to get the information we need to really make the Vulkan renderer scream.
After Vulkan
It’s a little too soon to discuss what comes after Vulkan, but I can say this: for almost two years now, Sidney and I have been rewriting the rendering engine with a rather strange goal: performance and predictability, but with the same visual output. So anything that looked ugly on the screen is supposed to keep looking ugly. Vulkan was a change of the how but not the what of our rendering engine.
Once Vulkan is out the door, that all changes. We have a number of fundamental changes we want to make to how we deal with light, with the atmosphere, with color, and with organizing our frame. Once we have Vulkan, we get to use it as our foundation for what comes next.
This weekend we’ll be in Cosford for Flight Sim 2019. Well, not me personally, but Philipp, Marty and Thomson are there.
This summer when Marty asked if we could run the Vulkan version of X-Plane on our machines at the show I laughed a bit and then said “hell no” – at the time we were running scenery packs in the sim with default aircraft, but Vulkan was still a work in progress.
Vulkan is still a work in progress, but there has been a lot of progress, so when the marketing guys asked again I said “sure, yeah, what could go wrong” and Sidney cut them a build. We also told them not to throw out the shipping version, but if the app is well behaved people at the show may get to fly the Vulkan build for the first time.
Betas, Betas, Everywhere
At this point I’d judge us to be in the middle of the X-Plane 11.40 beta run. Beta 6 is reasonably stable and has fixed a bunch of problems that broke third party aircraft, and it’s now available to steam users. We’re still tracking down a bug where the physics code produces a NaN value. Since it just happens while flying and we haven’t gotten any useful reproduction steps, we’re pursing it by adding more checks and looking at automatic crash reports.
Please, always auto-report your crash. We don’t need you to fill out the description text field – just the auto-report and email is enough. We are looking for as many reports as possible, and we will probably stick with two betas a week until we chase this one down.
Third parties: if you haven’t yet, please test your add-on with the experimental flight model off against beta 6 – we don’t have any known bugs with the legacy flight model and third party aircraft at this point.
We are also killing off the last beta bugs in the new version of X-Plane mobile. There’s some really cool tech going into the lighting on mobile that I’ll write up soon in a blog post.
Observant Steam pilots flying VFR noticed that the Cessna 172’s windows were completely solid gray in 11.35 release candidate 1 – not a huge problem for IFR, but not great for site seeing.
This should now be fixed – if you let Steam update the app you’ll get the clear windows back.
This bug was totally bizarre and astonishing. The interior glass texture for the Cessna was missing from the installation, but not from the master files we build the sim from or from the Laminar version. As best as I can tell, Steam’s tool for building the sim just lost the file randomly. I rebuilt the Steam install this morning and the file came back.
Here’s a feature that went into X-Plane 11.35 that will be really exciting to a very small number of scenery authors: in X-Plane 11.35, line files (.lin) can have custom painted end-caps on each end, and the texture repeats can be aligned to a grid.
Scenery authors who have gotten deep into X-Plane’s scenery system will already know from that first paragraph why this is useful, but for normal people who have seen the sun in the last 30 days, .lin files are the art files that define the look of thin linear features. X-Plane uses them for the painted yellow and white lines in the airport environment, but a developer can use them for anything linear.
Line files repeat, tile, and can follow the ground in a bezier curved path, which makes them great for curved taxiway lines. Their achilles heal up to now has been that when the line ends, the line just … stops. This makes them inappropriate for really thick uses, where that hard “cut” at the end of the line would be really noticeable and ugly.
In X-Plane 11.35 you can provide a start and end cap for each line definition. Like the line itself, it can be mulit-layered if desired. So, for example, you could use it to make dirt paths between buildings in a rural airport – where the path ends you can have a “soft” ending to the path and not have to worry about tucking the line under another scenery element to hide the cut.
We’re not using these in the default art yet, but the code is done, and I have updated the .lin file format specification with the new syntax.
X-Plane 11.35 beta 4 is now live – run our updater and check “get betas”. Release notes here.
Steam users: we’ll release this tomorrow if there isn’t a sign of a huge fire overnight – it’s already uploaded to the servers.
Add-on Developers: if you haven’t run your add-on in 11.35, please put it through its paces now. Everything that might be risky or weird is already in and we’re just killing off bugs now. If your add-on has a compatibility problem, we need to know now if we’re going to fix the problem during beta. While we’ve tried to make everything “just work” with old add-ons, there’s always the chance that your add-on does something special we didn’t think of. Please test now!
Here’s our keynote talk from FlightSimExpo 2019 last week!
FSExpo 2019
Evans video team really did a fantastic job with the feed this year – everything was live-streamed in great quality, and the talk has a direct feed to the slides and good quality audio. Totally what we were hoping for!
As you may have heard, we’ll be at FlightSimExpo 2019 in Orlando this weekend! If you are attending, please stop by our booth and drop in on our talk Saturday afternoon. We’ll be qualifying people for a VR landing contest too.
For developers: Austin, Philipp, Chris and I will all be there – find us to talk about developing add-ons for X-Plane!
Since my last Vulkan update, we now have the full sim running natively with Vulkan and Metal! There is still a pretty big list of random things turned off or bypassed to make this happen, but we can fly in the cockpit and use the sim.
Here’s some stuff that is now working:
Using regular and HDR rendering, and SSAO on metal.
Flying with Vulkan on AMD, NVidia, and Intel drivers.
Hardware stereo rendering on OS X – this never existed before Metal because the Mac GL drivers didn’t support it. Hardware stereo rendering is necessary for VR support.
MSAA on Metal – with the restructuring of our code, you’ll be able to change MSAA settings without restarting. I don’t know if this code works on Vulkan right now; it might.
Here’s some stuff that does not work yet:
Plugins are bypassed right now. We have not yet written the plugin-OpenGL-interop layer.
VR only works when using OpenGL as the driver; we need to write some new VR code to pass Metal and Vulkan frames directly to the OVR and Rift APIs.
Screenshots/Movie capture are a work in progress – that’s what I’ve been working on this week.
We still have a number of visual bugs, so screenshots are an important feature so we can run our automated test system. The test system takes hundreds of screenshots of the sim in many configurations and compares them to 11.30 to catch bugs introduced by the new Vulkan and Metal back-ends. Clearly I’ll have to fix my color problems first.
One fun aspect of this port: Metal and Vulkan copy Direct3D’s convention where the viewport Y axis points down and not up. This resulted in a whole series of weird “it’s upside-down again” bugs, most of which have been fixed.
The world is drawn correctly but the atmosphere around it is upside down.
(That airplane might look silly, but at least all of the image is consistently upside down. That’s because the only remaining upside down code is the movie/screenshot capture code itself.)
We should have some performance numbers to post next week; right now our primary focus is fixing bugs, particularly bugs that bring the whole machine down.
Here are a few more pics of things that have gone wrong during development.
