Summary

NVIDIA Nsight Systems 2019.6 is a comprehensive tool for graphics debugging and profiling. This release introduces several key features and improvements, including enhanced support for graphics trace on Windows, improved user-defined layouts, and new support for various Vulkan extensions. Here, we’ll explore these updates and how they can benefit developers.

Enhanced Graphics Trace on Windows

NVIDIA Nsight Systems 2019.6 expands graphics trace on Windows by adding support for Direct3D 11 and WDDM CPU+GPU traces. This enhancement allows developers to capture and analyze traces from a wider range of applications, providing deeper insights into performance issues.

Key Features

  • Direct3D 11 Support: This release adds support for Direct3D 11, enabling developers to capture and analyze traces from applications using this API.
  • WDDM CPU+GPU Traces: Support for WDDM CPU+GPU traces provides a more comprehensive view of system performance, helping developers identify bottlenecks.

Improved User-Defined Layouts

Nsight Graphics 2019.6 introduces user-defined layouts, allowing developers to save and load custom named layouts that better match their workflow needs. This feature enhances productivity by enabling users to tailor their workspace to specific tasks.

Custom Layouts

  • Saving and Loading Layouts: Developers can now save and load custom layouts, making it easier to switch between different workflows.
  • Customization: This feature allows for greater customization, enabling developers to organize their workspace in a way that suits their needs.

New Support for Vulkan Extensions

Nsight Graphics 2019.6 adds support for several new Vulkan extensions, including VK_EXT_line_rasterization, VK_EXT_headless_surface, VK_KHR_create_renderpass2, VK_KHR_external_fence_win32, VK_KHR_external_memory_win32, VK_KHR_external_semaphore, VK_KHR_imageless_framebuffer, and VK_NVX_image_view_handle. These extensions provide developers with more tools to create and optimize Vulkan applications.

Supported Extensions

  • VK_EXT_line_rasterization: Enables line rasterization in Vulkan applications.
  • VK_EXT_headless_surface: Supports headless surfaces, allowing for rendering without a display.
  • VK_KHR_create_renderpass2: Provides an updated API for creating render passes.
  • VK_KHR_external_fence_win32: Enables the use of external fences on Windows.
  • VK_KHR_external_memory_win32: Supports external memory on Windows.
  • VK_KHR_external_semaphore: Enables the use of external semaphores.
  • VK_KHR_imageless_framebuffer: Allows for framebuffers without images.
  • VK_NVX_image_view_handle: Provides a handle for image views.

Enhanced Events View and Filtering

The Events View in Nsight Graphics 2019.6 has been enhanced with a detailed filter configuration dialog and the ability to right-click on an event and hide events of that type. This improves the readability and usability of the Events View.

Filtering and Hiding Events

  • Detailed Filter Configuration: A new dialog provides more precise control over event filtering.
  • Hiding Events: Developers can now hide specific types of events, reducing clutter and improving focus.

GPU Trace Enhancements

GPU Trace in Nsight Graphics 2019.6 includes several enhancements, such as the ability to pin rows to the top in the scrubber and a new Flat Rows view mode. These features make it easier to compare data and customize the scrubber.

Pinned Rows and Flat Rows Mode

  • Pinned Rows: Developers can now pin rows to the top of the scrubber, keeping them visible while scrolling.
  • Flat Rows Mode: This new view mode allows for greater customization of scrubber rows, enabling the removal and pinning of child rows.

Conclusion

NVIDIA Nsight Systems 2019.6 offers a range of improvements and new features that enhance the debugging and profiling experience for developers. With enhanced graphics trace on Windows, improved user-defined layouts, and new support for various Vulkan extensions, this release provides developers with powerful tools to optimize and debug their applications. By leveraging these features, developers can gain deeper insights into performance issues and create more efficient and robust applications.