02 July 2015

Game Performance: Data-Oriented Programming

Posted by Shanee Nishry, Game Developer Advocate

To improve game performance, we’d like to highlight a programming paradigm that will help you maximize your CPU potential, make your game more efficient, and code smarter.

Before we get into detail of data-oriented programming, let’s explain the problems it solves and common pitfalls for programmers.

Memory

The first thing a programmer must understand is that memory is slow and the way you code affects how efficiently it is utilized. Inefficient memory layout and order of operations forces the CPU idle waiting for memory so it can proceed doing work.

The easiest way to demonstrate is by using an example. Take this simple code for instance:

char data[1000000]; // One Million bytes
unsigned int sum = 0;

for ( int i = 0; i < 1000000; ++i )
{
  sum += data[ i ];
}

An array of one million bytes is declared and iterated on one byte at a time. Now let's change things a little to illustrate the underlying hardware. Changes marked in bold:

char data[16000000]; // Sixteen Million bytes
unsigned int sum = 0;

for ( int i = 0; i < 16000000; i += 16 )
{
  sum += data[ i ];
}

The array is changed to contain sixteen million bytes and we iterate over one million of them, skipping 16 at a time.

A quick look suggests there shouldn't be any effect on performance as the code is translated to the same number of instructions and runs the same number of times, however that is not the case. Here is the difference graph. Note that this is on a logarithmic scale--if the scale were linear, the performance difference would be too large to display on any reasonably-sized graph!


Graph in logarithmic scale

The simple change making the loop skip 16 bytes at a time makes the program run 5 times slower!

The average difference in performance is 5x and is consistent when iterating 1,000 bytes up to a million bytes, sometimes increasing up to 7x. This is a serious change in performance.

Note: The benchmark was run on multiple hardware configurations including a desktop with Intel 5930K 3.50GHz CPU, a Macbook Pro Retina laptop with 2.6 GHz Intel i7 CPU and Android Nexus 5 and Nexus 6 devices. The results were pretty consistent.

If you wish to replicate the test, you might have to ensure the memory is out of the cache before running the loop because some compilers will cache the array on declaration. Read below to understand more on how it works.

Explanation

What happens in the example is quite simply explained when you understand how the CPU accesses data. The CPU can’t access data in RAM; the data must be copied to the cache, a smaller but extremely fast memory line which resides near the CPU chip.

When the program starts, the CPU is set to run an instruction on part of the array but that data is still not in the cache, therefore causing a cache miss and forcing the CPU to wait for the data to be copied into the cache.

For simplicity sake, assume a cache size of 16 bytes for the L1 cache line, this means 16 bytes will be copied starting from the requested address for the instruction.

In the first code example, the program next tries to operate on the following byte, which is already copied into the cache following the initial cache miss, therefore continuing smoothly. This is also true for the next 14 bytes. After 16 bytes, since the first cache miss the loop, will encounter another cache miss and the CPU will again wait for data to operate on, copying the next 16 bytes into the cache.

In the second code sample, the loop skips 16 bytes at a time but hardware continues to operate the same. The cache copies the 16 subsequent bytes each time it encounters a cache miss which means the loop will trigger a cache miss with each iteration and cause the CPU to wait idle for data each time!

Note: Modern hardware implements cache prefetch algorithms to prevent incurring a cache miss per frame, but even with prefetching, more bandwidth is used and performance is lower in our example test.

In reality the cache lines tend to be larger than 16 bytes, the program would run much slower if it were to wait for data at every iteration. A Krait-400 found in the Nexus 5 has a L0 data cache of 4 KB with 64 Bytes per line.

If you are wondering why cache lines are so small, the main reason is that making fast memory is expensive.

Data-Oriented Design

The way to solve such performance issues is by designing your data to fit into the cache and have the program to operate on the entire data continuously.

This can be done by organizing your game objects inside Structures of Arrays (SoA) instead of Arrays of Structures (AoS) and pre-allocating enough memory to contain the expected data.

For example, a simple physics object in an AoS layout might look like this:

struct PhysicsObject
{
  Vec3 mPosition;
  Vec3 mVelocity;

  float mMass;
  float mDrag;
  Vec3 mCenterOfMass;

  Vec3 mRotation;
  Vec3 mAngularVelocity;

  float mAngularDrag;
};

This is a common way way to present an object in C++.

On the other hand, using SoA layout looks more like this:

class PhysicsSystem
{
private:
  size_t mNumObjects;
  std::vector< Vec3 > mPositions;
  std::vector< Vec3 > mVelocities;
  std::vector< float > mMasses;
  std::vector< float > mDrags;

  // ...
};

Let’s compare how a simple function to update object positions by their velocity would operate.

For the AoS layout, a function would look like this:

void UpdatePositions( PhysicsObject* objects, const size_t num_objects, const float delta_time )
{
  for ( int i = 0; i < num_objects; ++i )
  {
    objects[i].mPosition += objects[i].mVelocity * delta_time;
  }
}

The PhysicsObject is loaded into the cache but only the first 2 variables are used. Being 12 bytes each amounts to 24 bytes of the cache line being utilised per iteration and causing a cache miss with every object on a 64 bytes cache line of a Nexus 5.

Now let’s look at the SoA way. This is our iteration code:

void PhysicsSystem::SimulateObjects( const float delta_time )
{
  for ( int i = 0; i < mNumObjects; ++i )
  {
    mPositions[ i ] += mVelocities[i] * delta_time;
  }
}

With this code, we immediately cause 2 cache misses, but we are then able to run smoothly for about 5.3 iterations before causing the next 2 cache misses resulting in a significant performance increase!

The way data is sent to the hardware matters. Be aware of data-oriented design and look for places it will perform better than object-oriented code.

We have barely scratched the surface. There is still more to data-oriented programming than structuring your objects. For example, the cache is used for storing instructions and function memory so optimizing your functions and local variables affects cache misses and hits. We also did not mention the L2 cache and how data-oriented design makes your application easier to multithread.

Make sure to profile your code to find out where you might want to implement data-oriented design. You can use different profilers for different architecture, including the NVIDIA Tegra System Profiler, ARM Streamline Performance Analyzer, Intel and PowerVR PVRMonitor.

If you want to learn more on how to optimize for your cache, read on cache prefetching for various CPU architectures.

26 June 2015

An update on Eclipse Android Developer Tools

Posted by Jamal Eason, Product Manager, Android

Over the past few years, our team has focused on improving the development experience for building Android apps with Android Studio. Since the launch of Android Studio, we have been impressed with the excitement and positive feedback. As the official Android IDE, Android Studio gives you access to a powerful and comprehensive suite of tools to evolve your app across Android platforms, whether it's on the phone, wrist, car or TV.

To that end and to focus all of our efforts on making Android Studio better and faster, we are ending development and official support for the Android Developer Tools (ADT) in Eclipse at the end of the year. This specifically includes the Eclipse ADT plugin and Android Ant build system.

Time to Migrate

If you have not had the chance to migrate your projects to Android Studio, now is the time. To get started, download Android Studio. For many developers, migration is as simple as importing your existing Eclipse ADT projects in Android Studio with File → New→ Import Project as shown below:

For more details on the migration process, check out the migration guide. Also, to learn more about Android Studio and the underlying build system, check out this overview page.

Next Steps

Over the next few months, we are migrating the rest of the standalone performance tools (e.g. DDMS, Trace Viewer) and building in additional support for the Android NDK into Android Studio.

We are focused on Android Studio so that our team can deliver a great experience on a unified development environment. Android tools inside Eclipse will continue to live on in the open source community via the Eclipse Foundation. Check out the latest Eclipse Andmore project if you are interested in contributing or learning more.

For those of you that are new to Android Studio, we are excited for you to integrate Android Studio into your development workflow. Also, if you want to contribute to Android Studio, you can also check out the project source code. To follow all the updates on Android Studio, join our Google+ community.

24 June 2015

Android Developer Story: Shifty Jelly drives double-digit growth with material design and expansion to the car and wearables

Posted by Lily Sheringham, Google Play team

Pocket Casts is a leading podcasting app on Google Play built by Australian-based mobile development company Shifty Jelly. The company recently achieved $1 million in sales for the first time, reaching more than 500K users.

According to the co-founder Russell Ivanovic, the adoption of material design played a significant role in driving user engagement for Pocket Casts by streamlining the user experience. Moreover, users are now able to access the app beyond the smartphone -- in the car with Android Auto, on a watch with Android Wear or on the TV with Google Cast. The rapid innovation of Android features helped Pocket Casts increase sales by 30 percent.

We chatted with co-founders and Android developers Russell and Philip Simpson to learn more about how they are growing their business with Android.

Here are some of the features Pocket Casts used:

  • Material Design: Learn more about material design and how it helps you create beautiful, engaging apps.
  • Android Wear: Extend your app to Android Wear devices with enhanced notifications or a standalone wearable app.
  • Android Auto: Extend your app to an interface that’s optimized for driving with Android Auto.
  • Google Cast: let your users cast your app’s content to Google Cast devices like Chromecast, Android TV, and speakers with Google Cast built-in.

And check out the Pocket Casts app on Google Play!

23 June 2015

Fitness Apps on Android Wear

Posted by Joshua Gordon, Developer Advocate

Go for a run, improve your game, and explore the great outdoors with Android Wear! Developers are creating a diverse array of fitness apps that provide everything from pace and heart rate while running, to golf tips on your favorite course, to trail maps for hiking. Let’s take a look features of the open and flexible Wear platform they use to create great user experiences.

Always-on stats

If your app supports always-on, you’ll never have to touch or twist your watch to activate the display. Running and want to see your pace? Glance at your wrist and it’s there! Runtastic, Endomondo, and MapMyRun use always-on to keep your stats visible, even in ambient mode. When it’s time for golf, I use Golfshot. Likewise, Golfshot uses always-on to continuously show yardage to the hole, so I never have to drop my club. Check out the doc, DevByte, and code sample to learn more.

Runtastic automatically transitions to ambient mode to conserve battery. There, it reduces the frequency at which stats are updated to about once per 10 seconds.

Maps, routes, and markers

It's encouraging to see how much ground I’ve covered when I go for a run or ride! Using the Maps API, you can show users their route, position, and place markers on the map they can tap to see more info you provide. All of this functionality is available to you using the same Maps API you’ve already worked with on Android. Check out the doc, DevByte, code sample, and blog post to learn more.

Endomondo tracks your route while your run. You can pan and zoom the map.

Google Fit

Google Fit is an open platform designed to make it easier to write fitness apps. It provides APIs to help with many common tasks. For example, you can use the Recording API to estimate how many steps the user has taken and how many calories they've burned. You can make that data to your app via the History API, and even access it over the web via REST, without having to write your own backend. Now, Google Fit can store data from a wide variety of exercises, from running to weightlifting. Check out the DevByte and code samples to learn more.

Bluetooth Low Energy: pair with your watch

With the latest release of Android Wear, developers can now pair BLE devices directly with the Wearable. This is a great opportunity for all fitness apps -- and especially for running -- where carrying both a phone and the Wearable can be problematic. Imagine if your users could pair their heart rate straps or bicycle cadence sensors directly to their Wear device, and leave their phones at home. BLE is now supported by all Wear devices, and is supported by Google Fit. To learn more about it, check out this guide and DevByte.

Pack light with onboard GPS

When I’m running, carrying both a phone and a wearable can be a bit much. If you’re using an Android Wear device that supports onboard GPS, you can leave your phone at home! Since not all Wear devices have an onboard GPS sensor, you can use the FusedLocationProviderApi to seamlessly retrieve GPS coordinates from the phone if not available on the wearable. Check out this handy guide for more about detecting location on Wear.

RunKeeper supports onboard GPS if it’s available on your Wearable.

Sync data transparently

When I’m back home and ready for more details on my activity, I can see them by opening the app on my phone. My favorite fitness apps transparently sync data between my Wearable and phone. To learn more about syncing data between devices, watch this DevByte on the DataLayer API.

Next Steps

Android Wear gives you the tools and training you need to create exceptional fitness apps. To get started on yours, visit developer.android.com/wear and join the discussion at g.co/androidweardev.

18 June 2015

Growing Android TV engagement with search and recommendations

Posted by Maru Ahues, Media Developer Advocate

When it comes to TV, content is king. But to enjoy great content, you first need to find it. We created Android TV with that in mind: a truly smart TV should deliver interesting content to users. Today, EPIX® joins a growing list of apps that use the Android TV platform to make it easy to enjoy movies, TV shows, sports highlights, music videos and more.

Making TV Apps Searchable

Think of your favorite movie. Now try to locate it in one of your streaming apps. If you have a few apps to choose from, it might take some hunting before you can watch that movie. With Android TV, we want to make it easier to be entertained. Finding ‘Teenage Mutant Ninja Turtles’ should be as easy as picking up the remote, saying ‘Teenage Mutant Ninja Turtles’ and letting the TV find it.

Searching for ‘Teenage Mutant Ninja Turtles’ shows results from Google Play and EPIX

You can drive users directly to content within your app by making it searchable from the Android TV search interface. Join app developers like EPIX, Sky News, YouTube, and Hulu Plus who are already making content discovery a breeze.

Recommending TV Content

When users want suggestions for content, the recommendations row on Android TV helps them quickly access relevant content right from the home screen. Recommendations are based on the user’s recent and frequent usage behaviors, as well as content preferences.

Recommendations from installed apps, like EPIX, appear in the Android TV home screen

Android TV allows developers to create recommendations for movies, TV shows, music and other types of content. Your app can provide recommendations to users to help get your content noticed. As an example, EPIX shows hollywood movies. NBA Game Time serves up basketball highlights. Washington Post offers video summaries of world events, and YouTube suggests videos based on your subscriptions and viewing history.

With less than one year since the consumer launch of Android TV, we’re already building upon a simpler, smarter and more personalized TV experience, and we can’t wait to see what you create.

16 June 2015

More Material Design with Topeka for Android

Posted by Ben Weiss, Developer Programs Engineer

Material design is a new system for visual, interaction and motion design. We originally launched the Topeka web app as an Open Source example of material design on the web.

Today, we’re publishing a new material design example: The Android version of Topeka. It demonstrates that the same branding and material design principles can be used to create a consistent experience across platforms. Grab the code today on GitHub.

The juicy bits

While the project demonstrates a lot of different aspects of material design, let’s take a quick look at some of the most interesting bits.

Transitions

Topeka for Android features several possibilities for transition implementation. For starters the Transitions API within ActivityOptions provides an easy, yet effective way to make great transitions between Activities.

To achieve this, we register the shared string in a resources file like this:

<resources>
    <string name="transition_avatar">AvatarTransition</string>
</resources>

Then we use it within the source’s and target’s view as transitionName

<ImageView
    android:id="@+id/avatar"
    android:layout_width="@dimen/avatar_size"
    android:layout_height="@dimen/avatar_size"
    android:layout_marginEnd="@dimen/keyline_16"
    android:transitionName="@string/transition_avatar"/>

And then make the actual transition happen within SignInFragment.

private void performSignInWithTransition(View v) {
    Activity activity = getActivity();
    ActivityOptions activityOptions = ActivityOptions
            .makeSceneTransitionAnimation(activity, v,
                    activity.getString(R.string.transition_avatar));
    CategorySelectionActivity.start(activity, mPlayer, activityOptions);
    activity.finishAfterTransition();
}

For multiple transition participants with ActivityOptions you can take a look at the CategorySelectionFragment.

Animations

When it comes to more complex animations you can orchestrate your own animations as we did for scoring.

To get this right it is important to make sure all elements are carefully choreographed. The AbsQuizView class performs a handful of carefully crafted animations when a question has been answered:

The animation starts with a color change for the floating action button, depending on the provided answer. After this has finished, the button shrinks out of view with a scale animation. The view holding the question itself also moves offscreen. We scale this view to a small green square before sliding it up behind the app bar. During the scaling the foreground of the view changes color to match the color of the fab that just disappeared. This establishes continuity across the various quiz question states.

All this takes place in less than a second’s time. We introduced a number of minor pauses (start delays) to keep the animation from being too overwhelming, while ensuring it’s still fast.

The code responsible for this exists within AbsQuizView’s performScoreAnimation method.

FAB placement

The recently announced Floating Action Buttons are great for executing promoted actions. In the case of Topeka, we use it to submit an answer. The FAB also straddles two surfaces with variable heights; like this:

To achieve this we query the height of the top view (R.id.question_view) and then set padding on the FloatingActionButton once the view hierarchy has been laid out:

private void addFloatingActionButton() {
    final int fabSize = getResources().getDimensionPixelSize(R.dimen.fab_size);
    int bottomOfQuestionView = findViewById(R.id.question_view).getBottom();
    final LayoutParams fabLayoutParams = new LayoutParams(fabSize, fabSize,
            Gravity.END | Gravity.TOP);
    final int fabPadding = getResources().getDimensionPixelSize(R.dimen.padding_fab);
    final int halfAFab = fabSize / 2;
    fabLayoutParams.setMargins(0, // left
        bottomOfQuestionView - halfAFab, //top
        0, // right
        fabPadding); // bottom
    addView(mSubmitAnswer, fabLayoutParams);
}

To make sure that this only happens after the initial layout, we use an OnLayoutChangeListener in the AbsQuizView’s constructor:

addOnLayoutChangeListener(new OnLayoutChangeListener() {
    @Override
    public void onLayoutChange(View v, int l, int t, int r, int b,
            int oldLeft, int oldTop, int oldRight, int oldBottom) {
        removeOnLayoutChangeListener(this);
        addFloatingActionButton();
    }
});

Round OutlineProvider

Creating circular masks on API 21 onward is now really simple. Just extend the ViewOutlineProvider class and override the getOutline() method like this:

@Override
public final void getOutline(View view, Outline outline) {
    final int size = view.getResources().
        getDimensionPixelSize(R.id.view_size);
    outline.setOval(0, 0, size, size);
}

and setClipToOutline(true) on the target view in order to get the right shadow shape.

Check out more details within the outlineprovider package within Topeka for Android.

Vector Drawables

We use vector drawables to display icons in several places throughout the app. You might be aware of our collection of Material Design Icons on GitHub which contains about 750 icons for you to use. The best thing for Android developers: As of Lollipop you can use these VectorDrawables within your apps so they will look crisp no matter what density the device’s screen. For example, the back arrow ic_arrow_back from the icons repository has been adapted to Android’s vector drawable format.

<vector xmlns:android="http://schemas.android.com/apk/res/android"
    android:width="24dp"
    android:height="24dp"
    android:viewportWidth="48"
    android:viewportHeight="48">
    <path
        android:pathData="M40 22H15.66l11.17-11.17L24 8 8 24l16 16 2.83-2.83L15.66 26H40v-4z"
        android:fillColor="?android:attr/textColorPrimary" />
</vector>

The vector drawable only has to be stored once within the res/drawable folder. This means less disk space is being used for drawable assets.

Property Animations

Did you know that you can easily animate any property of a View beyond the standard transformations offered by the ViewPropertyAnimator class (and it’s handy View#animate syntax)? For example in AbsQuizView we define a property for animating the view’s foreground color.

// Property for animating the foreground
public static final Property FOREGROUND_COLOR =
        new IntProperty("foregroundColor") {

            @Override
            public void setValue(FrameLayout layout, int value) {
                if (layout.getForeground() instanceof ColorDrawable) {
                    ((ColorDrawable) layout.getForeground()).setColor(value);
                } else {
                    layout.setForeground(new ColorDrawable(value));
                }
            }

            @Override
            public Integer get(FrameLayout layout) {
                return ((ColorDrawable) layout.getForeground()).getColor();
            }
        };

This can later be used to animate changes to said foreground color from one value to another like this:

final ObjectAnimator foregroundAnimator = ObjectAnimator
        .ofArgb(this, FOREGROUND_COLOR, Color.WHITE, backgroundColor);

This is not particularly new, as it has been added with API 12, but still can come in quite handy when you want to animate color changes in an easy fashion.

Tests

In addition to exemplifying material design components, Topeka for Android also features a set of unit and instrumentation tests that utilize the new testing APIs, namely “Gradle Unit Test Support” and the “Android Testing Support Library.” The implemented tests make the app resilient against changes to the data model. This catches breakages early, gives you more confidence in your code and allows for easy refactoring. Take a look at the androidTest and test folders for more details on how these tests are implemented within Topeka. For a deeper dive into Testing on Android, start reading about the Testing Tools.

What’s next?

With Topeka for Android, you can see how material design lets you create a more consistent experience across Android and the web. The project also highlights some of the best material design features of the Android 5.0 SDK and the new Android Design Library.

While the project currently only supports API 21+, there’s already a feature request open to support earlier versions, using tools like AppCompat and the new Android Design Support Library.

Have a look at the project and let us know in the project issue tracker if you’d like to contribute, or on Google+ or Twitter if you have questions.

12 June 2015

Updates to Unity, C++, and iOS tools for Play game services

Posted by Benjamin Frenkel, Product Manager

To further support all you game developers, we've updated our popular developer tools to give you a consistent set of game services across platforms for a better, more stable experience, with a particular focus on improvements to the Play game services Unity plugin. In addition, we added support for the Nearby Connections API, launched earlier this year at GDC, to our C++ SDK and Unity plugin.

Let’s take a look a closer look!

Unity plugin feature parity and stability improvements

We’ve added full support for Events and Quests in the Unity plugin. If you’re a Unity developer, you can now incorporate Quests into your games and take full advantage of Player Analytics natively within the Unity IDE.

We’ve also listened to feedback from our community of Unity plugin users and made stability improvements to Play game services Multiplayer, Saved Games, and to sign-in. You’ll now have a much better experience integrating with these Play game services, with fewer crashes and glitches.

C++ SDK and Unity support for the Nearby Connections API

We have added support for the Nearby Connections API to our C++ SDK and Unity plugin. You can now easily build awesome second screen and local multiplayer experiences, like this Beach Bugging Racing example, with the development tools you are most comfortable with.

Easier and more stable iOS builds with CocoaPods

We’ve also made major improvements to our Play game services CocoaPods, which simplify dependency management and building App Store packages from Xcode. The CocoaPods will improve building for iOS with the Play game services iOS and C++ SDKs, and the Unity plugin. We also improved the stability of multiplayer on iOS, eliminating many of the issues around accepting match invitations.

Finally, we improved our support for iOS 8, making it easier to set up multiplayer push notifications, and fixing UI compatibility issues.

Quick links to get you started

Play game services developer page: https://developers.google.com/games/services/
Case studies: http://developer.android.com/distribute/stories/games.html

Downloads