Open topic with navigation
This quickstart guide is designed to assist a new user in configuring PerfHUD ES to profile applications on a Tegra device.
PerfHUD ES enables experimentation and in-depth analysis of OpenGL ES applications. Through data provided by the OpenGL ES driver and the Tegra SoC, developers are provided live system status, including GPU performance and bottleneck information. This can help target optimizations where they are most needed.
The debugging capabilities of PerfHUD ES give full access to the state of the OpenGL ES pipeline, related textures and shaders, and all rendering states. By doing so, the developer is able to find the causes for improper setup, rendering anomalies, and performance issues.
Note that all downloads are available on the Tegra Developer's site:
Further detailed network setup instructions are outside the scope of this document.
Before you begin, make sure the following steps have been completed:
The PerfHUD ES launcher shortcut should be added by the installer to the desktop. The PerfHUD ES Host may also be accessed through the start menu at the following location:
Programs\NVIDIA Corporation\NVIDIA PerfHUD ES Tegra
The default installed location will be
On Linux, the latest NVIDIA drivers for your graphics card should be installed. If this was not done automatically, you can use the following command to install the driver:
sudo apt-get install nvidia-current
The PerfHUD ES launcher shortcuts should be added by the installer to the system menu, in one of the following locations:
NVIDIA PerfHUD ES
NVIDIA PerfHUD ES
The default installation location is
adb shell setprop debug.perfhudes 1
adb shell setprop debug.perfhudes 0
At this time, LDK customers will need the
tegra_linux_perfhud_10.8_10.9.tgz support pack for this next step – if you don’t already have it, please contact your support representative.
Copy the archive to the Tegra device (either copy to boot media from a host Linux PC, or use a remote file transfer like SCP). Then, simply unpack it on-device using a remote shell, using the following syntax:
tar -xzvf tegra_linux_perfhud_10.8_10.9.tgz
If you execute
ls, you should see find now have
perfhud.sh, as well as two PerfHUD libraries (
.so files). The first time you run the
perfhud.sh script, it will detect if the libraries are missing from the system, and attempt to install them properly (moving them to
|Note: If you are using the L4T 10.8.2 pack or later, you can skip this step and go directly to the “Enable or Disable PerfHUD Target Support” section below.|
To enable PerfHUD support, open a command shell to the device. From the default user home directory, execute:
If you later want to turn off PerfHUD support for performance or compatibility reasons, simply execute:
Note that you must re-enable PerfHUD ES after the device has been rebooted.
Enable PerfHUD support on the target device (detailed above), then follow the steps outlined in this section.
Launch the application to be profiled on the Tegra device. The following should be displayed in the log output (remote shell on a Linux device, or adb logcat on an Android device):
NVPerfHUD ES is active.
Communication layer init succeed, listen on port 7876.
This indicates the PerfHUD support code is running and waiting for the host app to connect.
Run the PerfHUD ES Host program on your host PC. You will then be prompted with a dialog requesting the Target device’s IP address.
If you are running just a single application, in a non-GLES-accelerated window manager:
If you are running multiple GLES applications simultaneously on the target, or a window manager that also leverages OpenGL ES in addition to applications, you’ll need to use the following steps:
|Note that when there’s only a single GLES application running, if you quit the application and then start it (or another GLES app) without exiting the PerfHUD ES Host, the Host will automatically re-establish a connection to the new app instance on the same target and port without user intervention. This is useful for repeated test runs as you find and fix issues.|
If ADB exists on your host system, then the PerfHUD ES Host will automatically attempt to connect to running applications on the target device using ADB. Note that if you have problems with ADB port forwarding, you may need to check that a local firewall is not blocking the port.
NVIDIA® GameWorks™ Documentation Rev. 1.0.200601 ©2014-2020. NVIDIA Corporation. All Rights Reserved.