Nvidia memory test linux
MemtestCL open source edition
- Description
- How to build
- Using MemtestCL as a library
- CLI Standalone Basic Usage
- CLI Standalone Advanced Usage
- FAQ
- Licensing
MemtestCL is a program to test the memory and logic of OpenCL-enabled GPUs, CPUs, and accelerators for errors. It is an OpenCL port of our CUDA- based tester for NVIDIA GPUs, MemtestG80.
This is the open-source version of MemtestCL, implementing the same memory tests as the closed-source version. The intended usage is as a library so that other software developers can use the MemtestCL tests to validate the correct operation of GPUs or accelerators in their own code. In addition to the core memory testing libraries, this package contains the source code to a limited version of the command-line interface standalone tester included in the closed- source build; certain capabilities, such as the ability to transmit results back to Stanford, are not present in the open-source version.
Up-to-date versions of both the closed- and open-source versions of MemtestCL may be downloaded at https://simtk.org/home/memtest. The closed-source version is available as precompiled binaries; the open-source version is available only as a source package.
This document concerns the open-source version.
First, ensure that you have installed an OpenCL SDK (typically, either that in the NVIDIA CUDA toolkits (from 3.0 onwards) or the ATI Stream SDK. Binaries built using either SDK should execute on any OpenCL implementation (ie, it is not required that the binary be built using the ATI Stream SDK to run on ATI GPUs). Common paths for SDK installations are included in the Makefiles; it may be necessary to adjust such paths to correspond to your installation.
Makefiles for 32- and 64-bit Linux, Mac OS X, and 32-bit Windows are included On Linux and OSX, it should be possible to build MemtestCL by executing the following from the root of the source package:
where OS is one of linux32, linux64, osx. On Windows, the Visual Studio C++ compiler and make system is required (tested under VS2005); execute the following command to build:
The resulting executable, memtestCL, should be immediately executable on Linux and OS X platforms. On Windows, libiconv-2.dll, libintl-2.dll, and popt1.dll must be copied from the popt/win32 subdirectory into a directory in the DLL search path (most conveniently, the root of the source distribution). MemtestCL uses the MIT/X licensed popt library to handle command line arguments; precompiled static libraries are provided for Linux and OS X, but dynamic libraries for Windows.
USING MEMTESTCL AS A LIBRARY
We encourage software developers to use MemtestCL as a code library in their programs to verify the correct operation of hardware on which they execute. The code is licensed under the LGPL, so developers of both open- and closed-source software can use it — developers of closed-source software are required to link to MemtestCL via a shared library (.so, .dll) mechanism; open-source software can integrate it via static linkage.
The API for the memory tests is defined in memtestCL_core.h. There are two APIs — a low-level API defined by the memtestFunctions class (which is a thin wrapper around the underlyinh OpenCL kernel invocations), and a high-level API defined by the memtestState and memtestMultiTester classes. At the lowest level the tests are implemented by the kernels in memtestCL_kernels.cl.
The recommended interface is the memtestMultiTester class, which automatically encapsulates details such as the maximum per-buffer allocation in a particular OCL library. An example of the API’s usage can be found in the standalone tester, memtestCL_cli.cu.
CLI STANDALONE BASIC USAGE
MemtestCL is available for Windows, Linux, and Mac OS X-based machines. In the following directions, please replace «MemtestCL» with the name of the program included in the distribution for your operating system.
MemtestCL is a command line application; to run it, start it from a command prompt (Start->Run->cmd in Windows, Terminal.app in OS X). For basic operation, just run it from the command prompt:
By default, MemtestCL will test 128 megabytes of memory on the first OpenCL device on the first OpenCL platform found,running 50 iterations of its tests. On typical machines, each iteration will complete in under 10 seconds with these parameters (the speed will vary both with the speed of the card tested and the amount of memory tested). The amount of memory tested and number of test iterations can be modified by adding command line parameters as follows:
For example, to run MemtestCL over 256 megabytes of RAM, with 100 test iterations, execute the following command:
Be aware that not all of the memory on your video card can be tested, as part of it is reserved for use by the operating system, and (as of this writing) both ATI and NVIDIA OpenCL drivers severely restrict the amount of memory available to an OpenCL program running on a GPU. If too large a test region is specified, the program will print a warning and quit. Also, if the tested GPU is currently driving a graphical desktop, the driver may impose time limits on test execution such that tests over very large test regions will time out. Timeouts or other execution errors will be trapped and will cause the test to terminate early. Due to the currently immature state of OpenCL implementations, they may also cause the program to crash.
If you suspect that your graphics card is having issues (for example, it fails running Folding@home work units), we strongly recommend that you test as large a memory region as is practical, and run thousands of test iterations. In our testing, we have found that even «problematic» cards may only fail sporadically (e.g., once every 50,000 test iterations). Like other stress testing tools, to properly verify stability MemtestCL should be run for an extended period of time.
CLI STANDALONE ADVANCED USAGE
MemtestCL supports the use of various command line flags to enable advanced functionality. Flags may be issued in any order, and may precede or follow the memory size and iteration count parameters (but the memory size must always precede the iteration count).
To run MemtestCL on an OpenCL platform other than the first (e.g., if you have both the AMD and NVIDIA OpenCL implementations installed), use the —platform or -p flags, passing the index of the platform to test (starting at zero). If you do not know the index of the OpenCL platform you want, just run MemtestCL with no parameters — a list of all platforms found will be printed immediately following the usage summary. For example, to run MemtestCL on the second platform in a system:
To run MemtestCL on an OpenCL device (e.g., GPU) other than the first one on the selected platform, use the —gpu or -g flags, passing the index of the device to test (starting at zero). MemtestCL prints a list of all devices on the selected platform (and their indices) before running tests. For example, to run MemtestCL on the third OpenCL device on the default platform (platform index 0):
The —platform and —gpu flags may be combined, to select a different platform and device. This may be necessary in multi-vendor or multi-GPU configurations. Refer to the list of platforms and devices on the current platform printed at program start to determine the right combination. To select the third GPU on the second platform:
Finally, to display the license agreement for MemtestCL, provide the —license or -l options:
To check lots of memory on AMD cards extra environment variables can be useful, depending on the driver version. For example with the below settings testing 7672MB was succesfully done on a 8GB AMD gpu w/o display on 64-bit windows.
Frequently Asked Questions
I have an
- Only OpenCL-capable hardware is supported. As of this writing, this includes any NVIDIA GeForce 8-series or newer graphics cards (e.g., GeForce 8-, 9-, GT- GTS-, GTX-series; Quadro FX, and Tesla boards), and the ATI Radeon 4xxx and 5xxx-series graphics cards. MemtestCL has also been tested on Intel and AMD CPUs using the AMD OpenCL implementation. OpenCL implementations exist for the Cell processor, but this has not been tested.
I have an OpenCL-enabled card, but it still doesn’t work!
- You must have an OpenCL-enabled graphics driver and OpenCL runtime installed. For NVIDIA hardware, this requires a version 195 ForceWare driver or newer; ATI requires BOTH a v9.12 or newer Catalyst driver, as well as (at the time of writing) the ATI Stream SDK.
I get an error complaining about a missing «OpenCL.dll» on Windows!
- You must install OpenCL. NVIDIA bundles OpenCL.dll with recent (v195 or newer) drivers. For ATI GPUs, you must install an OpenCL- capable video driver, and also install the ATI Stream SDK (http://developer.amd.com/gpu/atistreamsdk/). For CPU support, installing only the ATI Stream SDK is sufficient.
The source code to the open-source edition of MemtestCL is Copyright 2010, Stanford University, and is licensed under the terms of the GNU Lesser General Public License, version 3, reproduced below:
Источник
Nvidia memory test linux
README for MemtestG80 open source edition
- Description
- How to build
- Using MemtestG80 as a library
- CLI Standalone Basic Usage
- CLI Standalone Advanced Usage
- FAQ
- Licensing
MemtestG80 is a program to test the memory and logic of NVIDIA CUDA-enabled GPUs for errors.
This is the open-source version of MemtestG80, implementing the same memory tests as the closed-source version. The intended usage is as a library so that other software developers can use the MemtestG80 tests to validate the correct operation of GPUs in their own code. In addition to the core memory testing libraries, this package contains the source code to a limited version of the command-line interface standalone tester included in the closed-source build; certain capabilities, such as the ability to transmit results back to Stanford, are not present in the open-source version.
The closed-source version can be found at https://simtk.org/home/memtest. The open-source version lives at http://github.com/ihaque/memtestG80.
This document concerns the open-source version.
First, ensure that the CUDA toolkit binaries and libraries are included in the appropriate path variables for you system so that you can run the nvcc toolchain and successfully execute CUDA programs.
Makefiles for 32- and 64-bit Linux, Mac OS X, and 32-bit Windows are included From the root of the source package, it should be possible to build MemtestG80 by executing the following:
where OS is one of linux32, linux64, osx, or windows. Note that on Windows, GNU make (included, for example, in Cygwin) is assumed, not Microsoft nmake.
The resulting executable, memtestG80, should be immediately executable on Linux and OS X platforms. On Windows, libiconv-2.dll, libintl-2.dll, and popt1.dll must be copied from the popt/win32 subdirectory into a directory in the DLL search path (most conveniently, the root of the source distribution). MemtestG80 uses the MIT/X licensed popt library to handle command line arguments; precompiled static libraries are provided for Linux and OS X, but dynamic libraries for Windows.
USING MEMTESTG80 AS A LIBRARY
We encourage software developers to use MemtestG80 as a code library in their programs to verify the correct operation of GPUs on which they execute. The code is licensed under the LGPL, so developers of both open- and closed-source software can use it — developers of closed-source software are required to link to MemtestG80 via a shared library (.so, .dll) mechanism; open-source software can integrate it via static linkage.
The API for the memory tests is defined in memtestG80_core.h. There are two APIs — a low-level API defined by CUDA __host__ functions, and a high-level API defined by the memtestState class. At an even lower level, the individual tests are implemented by CUDA __global__ functions. Naming conventions are explained in comments in memtestG80_core.cu.
In general, for ease of use, we recommend the use of the high-level (object- oriented) API. An example of the API’s usage can be found in the standalone tester, memtestG80_cli.cu.
CLI STANDALONE BASIC USAGE
MemtestG80 is available for Windows, Linux, and Mac OS X-based machines. In the following directions, please replace «MemtestG80» with the name of the program included in the distribution for your operating system.
MemtestG80 is a command line application; to run it, start it from a command prompt (Start->Run->cmd in Windows, Terminal.app in OS X). For basic operation, just run it from the command prompt:
By default, MemtestG80 will test 128 megabytes of memory on your first video card, running 50 iterations of its tests. On typical machines, each iteration will complete in under 10 seconds with these parameters (the speed will vary both with the speed of the card tested and the amount of memory tested). The amount of memory tested and number of test iterations can be modified by adding command line parameters as follows:
For example, to run MemtestG80 over 256 megabytes of RAM, with 100 test iterations, execute the following command:
Be aware that not all of the memory on your video card can be tested, as part of it is reserved for use by the operating system. If too large a test region is specified, the program will print a warning and quit. Also, if the tested GPU is currently driving a graphical desktop, the NVIDIA driver may impose time limits on test execution such that tests over very large test regions will time out. This can be easily recognized by seeing a number of test errors larger than 4 billion, which go away when a smaller region is tested.
If you suspect that your graphics card is having issues (for example, it fails running Folding@home work units), we strongly recommend that you test as large a memory region as is practical, and run thousands of test iterations. In our testing, we have found that even «problematic» cards may only fail sporadically (e.g., once every 50,000 test iterations). Like other stress testing tools, to properly verify stability MemtestG80 should be run for an extended period of time.
CLI STANDALONE ADVANCED USAGE
MemtestG80 supports the use of various command line flags to enable advanced functionality. Flags may be issued in any order, and may precede or follow the memory size and iteration count parameters (but the memory size must always precede the iteration count).
To run MemtestG80 on a GPU other than the first, use the —gpu or -g flags, passing the index of the GPU to test (starting at zero). For example, to run MemtestG80 on the third GPU in a system:
Finally, to display the license agreement for MemtestG80, provide the —license or -l options:
I have an
I have a CUDA-enabled card, but it still doesn’t work! * You must have a CUDA-enabled graphics driver installed. See the Downloads section of http://nvidia.com/cuda to obtain a CUDA driver.
I get an error complaining about a missing «cudart.dll» on Windows! * This is a CUDA runtime file which we currently cannot redistribute with MemtestG80. However, a version of the file is bundled with the Folding@home GPU client; that file will work if copied into the MemtestG80 runtime directory.
The source code to the open-source edition of MemtestG80 is Copyright 2009, Stanford University, and is licensed under the terms of the GNU Lesser General Public License, version 3, reproduced below:
Источник
