NVIDIA PerfHUD ES Release Notes

• Support for NVIDIA SHIELD (requires OTA OS update 59 or newer)
• Google Nexus 7 support

• Support for Tegra4
• Google Nexus 7 support

• The Frame Profiler is much more accurate and efficient.
• New “user friendly” configuration dialog replaces manually edited XML.
• Draw call ranges can now be selected dynamically using new scroll area.
• The performance of the texture viewer has been significantly improved.
• The texture viewer now displays textures on all texture units.
• The shader performance report now considers GPU frequency and multi-epoch.

Important Information for Google Nexus 7 Users

• Numerous bug fixes, performance and stability improvements. 

• PerfHUD ES functions perfectly on the ASUS Transformer Prime (after the ICS update).
• Added the ability to easily advance to the next frame simplifying debugging workflow.
• Significant increase in performance.

• Root access no longer required for Android.
• New counters added that quantify the time spent waiting each from by the OpenGL ES driver (2D HW, 3D HW and CPU).
• Added the “Disable Texture Upload” experiment, which indicates if an application is texture-data bound.
• Added the “Disable Buffer Data” experiment, which indicates if an application has issues with data loads.
• Added the “Disable Uniform Upload” experiment, which helps determine if data uploads are a bottleneck.
• New GPU Idle counter added to the Performance Dashboard.
• Line and draw call numbers now associated with OpenGL ES call trace.
• After compiling a shader, double-clicking an error message will take you to the source code of that error.

• Added the “Disable Filtering” experiment, which indicates if an application is limited by filtering.
• New counter added that measures vertex cache hits.
• Dynamic Shader Editing now applies to all applications that use that shader.
• The Frame Debugger now displays all active attributes.
• The CU load information on the Performance Dashboard has been moved to a separate view.
• The host to target connection process has been greatly simplified.

• Modify a GLSL shader at runtime and immediately see the results on the target
• Multi-core processor support
• New directed test for differentiating between the cost of clear and draw calls
• User defined frame delimiters allow the user to specify which call ends a frame