How get GPU information in C#?

You can get basic information using this library:

https://github.com/falahati/NvAPIWrapper

---

For now, it doesn't support clock speed or GPU usage, nor it does support sensor information (temp), but it can get you the bus width, memory, used memory and cores. However, there is a solid ground for adding new functions to the library and as result, you can also expand it to fit your needs, probably in less than an hour or something if you know what function of NVAPI you need to add to the library and you are familiar with basics of marshaling.

Update 2018: It now supports GPU clock information, temp sensors, and usage information.

But for the information that is currently retrievable, you need to get a list of all currently connected physical GPUs. This is possible using the NvAPIWrapper.GPU.PhysicalGPU.GetPhysicalGPUs() static method. This method returns an array of NvAPIWrapper.GPU.PhysicalGPUs.

Now you can retrieve the information you desire using the properties of this class.

• NvAPIWrapper.GPU.PhysicalGPU.Bios: Gives you VBIOS version
• NvAPIWrapper.GPU.PhysicalGPU.Board: Gives you graphic board information
• NvAPIWrapper.GPU.PhysicalGPU.BusInfo: Gets GPU bus information
• NvAPIWrapper.GPU.PhysicalGPU.CUDACores: Gets total number of cores defined for this GPU
• NvAPIWrapper.GPU.PhysicalGPU.CurrentPCIEDownStreamWidth: Gets number of PCIE lanes being used for the PCIE interface downstream
• NvAPIWrapper.GPU.PhysicalGPU.FullName: Gets GPU full name
• NvAPIWrapper.GPU.PhysicalGPU.GPUType: Indicates if the GPU is integrated or discrete
• NvAPIWrapper.GPU.PhysicalGPU.IRQ: Gets GPU interrupt number
• NvAPIWrapper.GPU.PhysicalGPU.IsQuadro: Indicates if this GPU is of the Quadro line of products
• NvAPIWrapper.GPU.PhysicalGPU.MemoryInfo: Gives you every information about the memory and memory usage
• NvAPIWrapper.GPU.PhysicalGPU.PCIIdentifiers: Gives you information about the hardware PCI ids
• NvAPIWrapper.GPU.PhysicalGPU.PhysicalFrameBufferSize and NvAPIWrapper.GPU.PhysicalGPU.VirtualFrameBufferSize: Gets the size of frame buffer in KB for this GPU
• NvAPIWrapper.GPU.PhysicalGPU.ShaderSubPipeLines: Gets number of GPU Shader SubPipes or SM units

If you need a specific function you can't find in this version of the library, please feel free to open an issue.

I don't have any experience with the AMD tools, but we managed to use NVIDIA's NVAPI (https://developer.nvidia.com/nvapi) from C#

The version we use is only supplied as a static library do we can't p/invoke. We ended up create a thin wrapper library in C++/CLR that we could then call from our C# code.

Maybe the Win32_VideoController CLASS or the GPUinformation Class can help you.

Example:

using System.Management;
 
public partial class Win_Win32_VideoController : System.Web.UI.Page
{
    protected void Page_Load(object sender, EventArgs e)
    {
        using (var searcher = new ManagementObjectSearcher("select * from Win32_VideoController"))
        {
            foreach (ManagementObject obj in searcher.Get())
            {
                Response.Write("Name  -  " + obj["Name"] + "</br>");
                Response.Write("DeviceID  -  " + obj["DeviceID"] + "</br>");
                Response.Write("AdapterRAM  -  " + obj["AdapterRAM"] + "</br>");
                Response.Write("AdapterDACType  -  " + obj["AdapterDACType"] + "</br>");
                Response.Write("Monochrome  -  " + obj["Monochrome"] + "</br>");
                Response.Write("InstalledDisplayDrivers  -  " + obj["InstalledDisplayDrivers"] + "</br>");
                Response.Write("DriverVersion  -  " + obj["DriverVersion"] + "</br>");
                Response.Write("VideoProcessor  -  " + obj["VideoProcessor"] + "</br>");
                Response.Write("VideoArchitecture  -  " + obj["VideoArchitecture"] + "</br>");
                Response.Write("VideoMemoryType  -  " + obj["VideoMemoryType"] + "</br>");
            }
        }
    }
}

You can also consult the CUDAfy.net library.