There are a few simple options available to the GPU PerfClient application. These can be used from a command prompt, or by creating a custom shortcut to launch the GPU PerfClient application.
The Client command line options that are available are as follows (note that double dashes ("--") are required):
GPUPerfClient.exe [--server <IPv4 or HostName> [--port <port number>] [--app <path to app> [--appargs <app arguments>]]]
- --server <IPv4 or HostName>: Attempt to automatically connect to this server on startup
- --port <port number>: Only valid with --server or --app arguments. Use this port instead of the default for the connection.
- --app <path to app>: Only valid for local 32 bit app debugging. Launch the specified application with GPU PerfServer automatically when starting GPU PerfClient
- --appargs <app arguments>: Specify any arguments that should be passed to the application being launched with the --app command
For quick help on any of these options, type
GPUPerfServer.exe --help in a command box.
- --help Print all available command line options along with descriptions
- --openwith Adds an "Open with GPU PerfServer" option to the right-click context menu in Windows.
- --clean Removes any changes made to the system by GPUPerfServer. This includes undoing the effect of the --openwith option.
- --version Print the GPU PerfServer version number
The following options can also be specified in a GPUPerfServer.cfg configuration file which should be located in the same place as GPUPerfServer.exe
- port=<port number> Specify the TCP/IP Port to use for communicating with the GPU PerfServer.
- no-logfile=true Disable writing to the pslog.txt log file.
- logfile=<path to logfile> Path to logfile to use for PerfServer error logging.
- loglevel=<value> Verbosity of LogFile messages. The default value is 0. The valid values are:
- 0 = Errors only
- 1 = Errors + Warnings
- 2 = Errors + Warnings + Messages
- real-pause=true Use 'Freeze' Time Spoofing method. When this mode is enabled, GPU PerfServer will always return a delta-time value of 0 to the application. Some applications may crash with this mode enabled. Note: The user will typically control the Time Spoofing mode through the client-side Settings. This server option should only be used by advanced users who wish to set up application specific GPU PerfServer launch scripts. See Pausing your Application for more information on the time spoofing setting.
- speed=<value> Set the playback speed of the application. This is a relative speed (i.e. a value of 2 means to run at twice normal speed). Set this option to 0 and use the --real-pause option to pause on the first frame of the application.
- debug-runtime=true Enable/Disable the D3D Debug Runtime. When this option is enabled, GPU PerfServer will force the use of the D3D 10 Debug Runtime when running the application. Any D3D Debug messages will appear in the API Trace window in the GPU PerfClient application.
- warp-device=true Use the D3D WARP device. When this option is enabled, GPU PerfStudio server will force the D3D WARP device to be used.
- hopcountmax=<value> Controls the maximum number of processes that the GPU PerfStudio server should track before binding to a process.
The following example shows how a "GPUPerfServer.cfg" file may be used to temporarily change the default configuration of GPU PerfServer
## This file is intended to provide sample usage of the parameters for the
## config file.
## All of these options also have a command line equivalent
## Logfile options
# disable writing to logfile
no-logfile=true
# specify level of logging, currently valid values are 0-2 corresponding to
# logERROR, logWARNING, logMESSAGE
# for example loglevel=1 will write errors and warnings, but not messages or traces
loglevel=2
# specify webserver port
port=80
# Applications that can handle delta t=0 can set this option to true, meaning
# pause will do absolute pausing, but some applications can't handle absolute pausing,
# so set it to false here
real-pause=false
# Run the application at double speed
speed=2
# specify that the launched application should break - awaiting debugger attach # break=true
# Turn on the use of the D3D Debug Runtime
debug-runtime=true
# Controls the maximum number of processes that the GPU PerfStudio server should track
# before binding to a process. This can be useful if you are using Steam in situations where a
# process is created after the actual game exe is started. The hopcountmax value can be
# used to stop GPU PerfStudio from jumping to these late starting processes.
hopcountmax=2