If it’s content you’re looking for, you’ve come to the right place. From articles about graphics and 3D animation to lists of video games and digital media products, you’ll find all of it in here. Get reading!

You have an hour to kill before your next class. What to do? Study for your physics test? Nah, you whip out your lightweight laptop and continue the cool new game you were playing last night!

The shift from desktop PCs to laptops is creating new opportunities for game companies to expand their customer base. Gaming on the go—anytime, anywhere—is becoming commonplace. Game developers who once made games that were considered exclusively “high end” are now reaching a broader audience by making their games scale better across platforms. In 2008 mobile integrated chipset sales outsold discrete desktop graphics card sales for the first time in history. By 2013 mobile chipsets are expected to outsell discrete graphics cards by more than three to one.1 As these numbers continue to grow, developers need tools to help them quickly optimize gameplay to reach this broader market.

New tools, such as the Intel® Graphics Performance Analyzers (Intel® GPA), are helping developers optimize game code for integrated graphics, so they can hit the performance targets needed while still delivering the great visuals their customers want.

Unlike some other tools on the market that were really intended for internal use and don’t fit the needs of customers, Intel worked with a number of game companies to identify the features game developers needed most, and designed the tools specifically to fulfill those needs. “This is just the start,” said Dave Shinsel, Intel GPA engineering manager. “We have a bunch of cool features we’ll be implementing. We continue to listen to our customers; they tell us the features that are most important, and we build those first.” Mark Randel, president and chief technology officer of Terminal Reality, said his team used Intel GPA extensively on their upcoming title Ghostbusters*: The Video Game (image right). Based on the smash hit motion picture franchise, the new game reunites the original cast members not only with voice work, but also within the story line. The game uses graphics to enhance both humor and fright and is meant to appeal to all key market segments.

Early on, Randel knew he wanted to support more than just the high-end desktop space. “Intel GPA is a great first step into optimizing games for integrated graphics,” he said. “With the client/server approach, you can measure the system in real time with minimal overhead to the target application. For example, you can launch your app, move to a problem area, and see where your time is going from buffer locks to state changes, and even down to the chip level if that is what you need to get it running faster.” Taking advantage of the opportunity to sell games into the mobile chipset space makes sense only if the effort doesn’t substantially increase development time. Randel reported that his team was pleasantly surprised when Intel GPA helped make this a reality. “The Intel Graphics Performance Analyzers helped us mainly by reducing development time, since the tool provides a focus on specific problem areas. It also allowed our game to support the widest possible audience, by being able to run on Intel® Graphics. The benefit is mainly cost savings; the tool helps the engineer focus on the problem areas versus trial and error experiments. You can locate problem areas quicker than just turning visuals in the game on and off.”

The Intel GPA consists of two tools: the Frame Analyzer and the System Analyzer. Both tools sit on a network-based architecture, making data collection less intrusive and more suited to remote analysis than similar analyzers on the market. Because Intel GPA performs the calculations on a different machine than on the one running the game, developers don’t have to worry about tool overhead.

The Frame Analyzer (Figure 1) allows developers to inspect and adjust graphics API-level interactions on a frame-by-frame basis. This capture- and playback-based tool shows detailed frame performance, with draw calls visualized on a GPU duration graph. Using the scene-overview feature, which gives a spreadsheet view of the same GPU information, developers can drill down from the full frame to single draw calls. Full-render state overrides, shader overrides, and other high-level experiments are supported with real-time feedback. For example a developer can modify a shader directly in the tool and immediately see if that change affected the frame time, region time, or draw-call time. Similarly, the developer can modify the DX state or run a high-level experiment, such as a simple pixel shader, and immediately see if that change affected the frame time, region time, or draw-call time.

The SEGA development team working on Empire*: Total War used the Frame Analyzer extensively. Chris Southall, technical director at SEGA, said, “We have mainly used the frame level of Intel GPA. We have made changes to the level of detail in the game, made some art changes, optimized some shaders, changed draw order, and reduced overdraw. Using the system-level debugger, we have found a few bottlenecks with the way we were filling vertex buffers, which have also been fixed.”

The System Analyzer (Figure 2) is a high-level, real-time performance tool with game pause-and-resume capabilities that provides a system-level footprint of game performance as well as a single-frame-capture button that transitions to the Frame Analyzer. Developers can experiment with Microsoft DirectX* state overrides, as well as customizable drag-and-drop metrics.

Bartosz Kijanka, vice president of engineering for Gas Powered Games, said his team relied on the Intel GPA tools for the company’s forthcoming title: Demigod*. A fast-paced, real-time strategy game, Demigod blends role-playing elements with tactical combat. Players can choose from a variety of wildly customizable demigods, who fight in ancient battle arenas for the right to ascend into the pantheon of gods.

Kijanka was intent on supporting integrated graphics chipsets from the start, so the Intel GPA tool was invaluable. “The System Analyzer helped us identify bottlenecks in the graphics pipeline through a few simple and easy-to-understand graphs,” said Kijanka. “The Frame Analyzer is an amazing tool for drilling down deep into our engine’s render pipeline and identifying individual pieces of work that may be disproportionately expensive. What is particularly impressive about this module is that the Frame Analyzer allows us to test possible fixes and to see the results immediately.”

Using Intel GPA significantly reduced the cost of identifying and fixing graphics performance problems on the PC, according to Kijanka. “Many problems are costly to repair because they are very time consuming to diagnose,” he explained. “They are time-consuming because they often require creation of project-specific instrumentation or project-specific tools. Spending less time on such tools allows us to spend more time on actually making the games, which is something every game developer wants to do.”

The System Analyzer allows developers to run experiments that pinpoint common problems. The simple pixel-shader-override mode replaces every pixel shader with one that writes a constant color to the render target. A large increase in the frame rate after enabling this mode means that the game is spending a large proportion of time in either pixel shader compute or stall conditions.

The 1x1 scissor-override mode causes all pixels to be discarded after the pixel shader has run, before the pixel values are written to the render target. A significant increase in the frame rate after enabling this mode means the pixel-fill rate may be a potential bottleneck. Developers can then examine the operations, such as stencil and alpha blending, in the graphics pipeline to determine optimization potential.

Potential performance bottlenecks in the use of texture maps can be identified by using the 2x2 texture-override mode, in which all textures for a scene are replaced with a simple 2x2 pixel texture. If the override mode significantly improves the frame rate, the GPU may be bottlenecked on texture memory reads. If the total texture size is high for a scene, developers may want to consider reducing texture bandwidth in various ways.

The System Analyzer also collects various CPU and GPU metrics while the application is running. Developers can analyze the results and perform various “what if” scenarios to help isolate performance bottlenecks. The metrics data are stored in a local database and can be displayed in chart form for interactive analysis. Developers can pause the application and perform on-the-fly modifications without changing the application code.

Intel’s software team worked hard to create an intuitive interface for the tools. For example, the Frame Analyzer displays all DX data in visual form when possible. This includes the render targets, draw-call highlighting for the draw-call selection set, textures, and so on. The visual element of the tool suite allows any game developer to pick up the tool and use it effectively immediately.

“PC developers have never had such a reliable, flexible, and extensive tool set for identifying and resolving complex graphics performance issues,” said Kijanka. “Intel GPA collects and presents a breadth and depth of information about graphics performance that we have never previously seen for the PC platform. Our games perform much better, across a wider range of hardware than ever before, including many graphics parts with high market penetration that we have traditionally not had the resources to optimize for. This means more customers can play our games and have a great, high-frame-rate experience, even on older or lower-cost PCs, and we definitely benefit from that.”

As the Codemaster’s development team recently discovered, Intel® GPA offers in-depth frame and system analysis for graphics tasks. View Video »

For more details, go to: www.intel.com/software/gpa