ychding11 / cudastudy Goto Github PK

Error describe

*** WARNING - PE SIGNATURE VERIFICATION FAILED !!!
*** Filename: C:\Program Files\NVIDIA Corporation\Ansel\Tools\NvCameraWhitelisting64.dll
*** Error: 0x80092009
D3D11CreateDevice: Flags (0x2) were specified which require the D3D11 SDK Layers for Windows 10, but they are not present on the system.
These flags must be removed, or the Windows 10 SDK must be installed.
Flags include: D3D11_CREATE_DEVICE_DEBUG
Exception thrown at 0x00007FFA97863C58 in OGSTestApp.exe: Microsoft C++ exception: _com_error at memory location 0x0000000000148D50.

Kernel Time out

• Kernel fails if it takes more than 1.6s on my computer.
• Environment Windows 10
• CUDA Device

        Name                                                                GeForce GTX 965M
        Driver                                                              WDDM
        DeviceIndex                                                         0
        GPU Family                                                          GM206-A
        RmGpuId                                                             256
        Compute Major                                                       5
        Compute Minor                                                       2

• WDDM Devices

            GPU
            Name                                                            GeForce GTX 965M
            Architecture                                                    Unknown
            Architecture Number                                             288
            Architecture Implementation                                     6
            Architecture Revision                                           161
            Number of GPCs                                                  2
            Number of TPCs                                                  8
            Number of SMs                                                   8
            Warps per SM                                                    64
            Lanes per warp                                                  32
            Register file size                                              65536
            Max CTAs per SM                                                 32
            Max size of shared memory per CTA (bytes)                       49152
            SM Revision                                                     327680

It takes a long time to init curand state

• It takes 13s to generate a 1024x1024 curand states.
• breaks down whole kernel into sub-kernel of 256x256 because of big kernel time out.

• http://mrob.com/pub/muency/marianisilveralgorithm.html
• https://www.docenti.unina.it/webdocenti-be/allegati/materiale-didattico/517066