I have recently tried to update my system from CUDA 3.2 to CUDA 4.0
This seems to not be a very smooth update. First of all, a lot of the SDK examples fail now. matrixMul and the FFT problems and such all come back saying the errors were too big, so they give a fail.
Code that I have written also seems to have some problems. I had to track down some errors caused by NaN that I never had before, and I am getting occaisional kernel launch failures now too. Usually it says
cudaSafeCall() Runtime API error 4: unspecified launch failure.
or
cutilCheckMsg() CUTIL CUDA error : fftshift() execution failed
: (4) unspecified launch failure.
(fftshift being my function that has never failed before, in 3.2)
Do I perhaps have something misconfigured here?
I have tried using the 270 driver that was released with 4.0 for development, after having downgraded from the latest driver that I thoguht was maybe causing problems.
Thanks for any tips
I'm going to assume (since you didn't specify) that you are running Windows(7) and are trying to use 3.2 SDK samples with nvcc 3.2 compiler but with 4.0 capable drivers. If so, you should know that version 4.0 has a major revision in regards memory addressing (namely Unified memory addressing on 64-bit systems) in addition to a whole bunch of other features.
What you want is to uninstall CUDA 3.2 driver, SDK and Toolkit (purge them for good). Then download and install all the 4.0 versions of the above. The new SDK samples should work, provided everything is installed correctly.
Related
Since a couple of weeks I'm trying to profile a piece of numerical software and I'm unable to get useful results.
The code I'm profiling results in a huge function (__attribute__((flatten))) created out of many inlined functions and a few calls to std::exp/std::log/std::pow. This function is located inside a shared library and loaded via dlopen().
I've used
the google CPU profiler (hangs in the first fork() (interrupted by SIGPROF and restarted and interrupted and...) -- same problem with g++ option -pg)
linux tool perf (caused a reboot of the machine, I complained and they upgraded the OS (CENTOS 6.5). The results only highlight two assembler instructions out of above mentioned huge function. I don't have permissions to read accurate event sources (*:ppp))
some old version of vtune (difficult to operate, results are unreliable, no hardware drivers loaded)
sprof (results do not tell me anything as there is only a single function to profile -- when avoiding to use attribute flatten then the behavior is fully different)
I'm running
CENTOS 6.5
and
g++ (GCC) 5.3.0
I don't have any influence over the version of the OS or the compiler version.
I complained about the ancient OS some weeks ago, and they upgraded me to what I mentioned above.
In some former live I successfully used the google profiler -- when it was working (and not crashing or hanging due to signal handling problems) it provided useful results.
Anybody any comment?
Could all these unclear results be the result of the fixes for SPECTRE?
Do I need to insist, that certain profiling options are enabled on the machine?
Do I need to insist on the vtune drivers loaded?
Do I need to insist on an uptodate copy of vtune installed?
Compile with -fno-omit-frame-pointers?
I was trying to code with unreal engine 4 (version 4.24.2), but I don't know why, all of a sudden this happen and since then I couldn't do nothing no matter what the version or projects this keeping happening and I don't know what is it or how to fix it
output log error here:
I had the same issue, but it was fixed once I updated to the latest Intel graphics driver.
Check your display driver version.
(In my case, I used gtx1060 and intel graphics630, and only intel driver was not lastet version)
Visit https://downloadcenter.intel.com and update your drivers
I've installed the Intel OpenCL SDK. I'm able to find the header and link the library, but none of the functions work. The very first function called in an OpenCL program, clGetPlatformIDs(), returns -1001 - an undefined error code from what I can find.
Other people have gotten the same issue with the Nvidia SDK for OpenCL and fixed it, but I'm finding it hard to get an answer about how to solve the issue with an Intel installation, which has a slightly different setup.
Please direct your questions about Intel OpenCL SDK to Intel OpenCL Forum: https://software.intel.com/en-us/forums/opencl
To ensure fast service:
Please let us know what Processor, Operating System, Graphics Driver Version, and Tool Version you are using
Please state steps to reproduce the issue as precisely as you possibly could
Before posting, search the forum to see if someone already answered a similar question
I’m a C++ application developer on Windows system. I’m developing an application using EGL 1.4 and OpenGL ES 2.0.
In a certain kind of a OpenGL32.DLL version, I found that a PVR error message occurred. The versions of OpenGL32.DLL that cause this problem are 6.3.9600.16384 on Win8.1 and 4.0.1379.1. And the message says :
PVR Message
PVR: VFrame attempted to use one of these functions:
PVR: wglChoosePixelFormatARB
PVR: but they are not present on you machine.
Ignore?
This message appeared when I call eglChooseConfig() or eglGetConfigs() functions in EGL API to get system configs.
I think this problem occurred because OpenGL32.dll versions mentioned above don’t support wglChoosePixelFormatARB() function.
Most of versions that I tested worked well. List of working version is :
5.2.3790.3959
6.1.7600.16385
6.3.9600.16384(1000kb)
10.0.10130.0
And list of not working version is :
4.0.1379.1
5.1.2600.1280
5.1.2600.5512
6.3.9600.16384 (720kb)
To solve this problem, I want to include this dll file in my product for a commercial use. I want to know that distributing a working version of OpenGL32.dll is possible. And can you know me the list of OpenGL32.dll version without this error and license information about OpenGL32.dll?
The issue is not about the opengl32.dll (in fact its inner working hasn't changed since Windows NT-4 some 19 years ago). Your system is simply missing a graphics driver with full OpenGL support:
The graphics drivers that are installed by default on Windows (and Windows Update) have only a very limited, stripped down OpenGL support (in fact they hardly support GPU acceleration). To get proper, full OpenGL support you must download the driver package for your GPU from the GPU vendor's website and install it manually.
Using Windows standard, automatic driver installation and Windows Update will not give you proper OpenGL support.
Using the AMD C++ binding and SDK (the most recent one) running an OpenCL program that gets a platform, a GPU, then compiles 4 kernels has the above error upon startup. It works fine on my computer, whose GPU only supports up to 1.1, but other computers seem to have the above error. Is this a problem in the compilation (As in, I have to define some macros), in the lack of a driver, the C++ binding, or something else? I don't explicitly call clRetainDevice in my own code—is it part of the binding somewhere?
It happens when you use the C++ bindings header file with OpenCL 1.2 header. For instance, when you run an application compiled with AMD SDK (OpenCL 1.2) on NVIDIA platform (OpenCL 1.1 only).
As fast and dirty work around, you can just edit the AMD SDK cl.h header and undef "CL_VERSION_1_2" preprocessor symbol. If you are not interested to 1.2 features, it should fix your problem.