I encountered the following warning message pop-up window of CUDA visual profilier
when I run a OpenCV-2.3.0 based program with gpu acceleration.
The profiler output and summary table were obtained but ‘GPU Timestamp(us)’ value
is inconsistent with ‘GPU Time’ of API Trace view.
== Compute Visual Profiler warning ==
The selected counter(s) “gld instructions 8bit”, “gld instructions 16bit”, “gld instructions 32bit”, “gld instructions 64bit”, “gld instructions 128bit”, “gst instructions 8bit”, “gst instructions 16bit”, “gst instructions 32bit”, “gst instructions 64bit”, “gst instructions 128bit” can cause GPU kernels to run longer than the driver’s watchdog timeout limit. In this case the driver will terminate the GPU kernel resulting in an application error and the profiling data will not be available. Setting the X Config option ‘Interactive’ to false is recommended when these counters are selected.
The last sentence says that Setting the X config option ‘Interactive’ to false is recommended …
I am wondering in which file "Interactive’ is.
I guessed .nvidia-setting-rc file at home directory, but
I don’t know how to modify it.
The environment used:
Fedora 13
CUDA 4.0
OpenCV-2.3.0
GPU card: GT440
P.S. My CUDA visual profiler works fine without any warning message
for a simple CUDA program such as two vector addition.
I am also getting this same popup; it seems that the profiler is only recording 32b loads and stores for me. 64 and 128 are empty, but if I profile the same program on OS X, I see 64 and 128 loads and stores. What exactly is the “Interactive” X config setting? Is it an Xresource?
This Visual Profiler message is expected when the “gld instructions …” or “gst instructions …” counters are enabled. You need to change the X Config option ‘Interactive’ to off only if the application is getting prematurely terminated.
The ‘Interactive’ driver option may be specified either in the Screen or Device sections of the X config file.
Option “Interactive” “boolean”
This option controls the behavior of the driver’s watchdog, which attempts to detect and terminate GPU programs that get stuck, in order to ensure that the GPU remains available for other processes. GPU compute applications, however, often have long-running GPU programs, and killing them would be undesirable. If you are using GPU compute applications and they are getting prematurely terminated, try turning this option off.
Default: on. The driver will attempt to detect and terminate GPU programs that cause excessive delays for other processes using the GPU.
e.g. Add Option “Interactive” “0” to the device section of your GPU in /etc/X11/xorg.conf. The device section looks like:
Section “Device”
Identifier “Device0”
Driver “nvidia”
VendorName “NVIDIA Corporation”
BoardName “GeForce GTX 460”
Option “Interactive” “0”
EndSection
What do you mean by “‘GPU Timestamp(us)’ value is inconsistent with ‘GPU Time’ of API Trace view.”?
I am also facing the same problem.The profiler is not able to complete the execution and i am getting the following error
[Program run #18 completed. Error : Application : "/home/cnode0/workspace/cuda/pred/pre".
Profiler data file '/home/cnode0/workspace/cuda/pred/temp_compute_profiler_0_0.csv' for
application run 0 not found.
I also got same warning during the execution
The selected counter(s) "gld instructions 8bit", .....
be available. Setting the X Config option 'Interactive' to false is recommended when
these counters are selected.
As mentioned in the above post, I changed my xorg.conf to set the interactive flag to false as follows and restarted the system
