How to pass from Cuda 4.2 to 10.1 and replace compute_13,sm_13 by what?

Accelerated Computing CUDA CUDA Programming and Performance
1

I Pass from Cuda 4.2 to 10.1
I have the Error:
nvcc fatal : Unsupported gpu architecture ‘compute_13’

What is the Macro for Comput_xx et sm_xx ?

Thanks

2

you’re specifying it in the compile command somewhere. So change it.

3

I’m in Visual Studio 2019 :
Where is the Macro commande ?

I change It and it’s make alwais the same error,
nvcc fatal : Unsupported gpu architecture ‘compute_13’
but I’have put:
cumpu_53,sm_53

4

compute_35,sm_35 was put If I changer Visual Studio 219

But the Message compute_13 Error was Alwais send !

I don’t use GCC !

5

I don’t know what you mean by macro command.

In Visual Studio, you set the compute target here:

image
image700×494 60.1 KB

You need to edit the section to the right of “Code Generation”. Remove any unwanted targets. For a complex project, its possible that this is specified on a per-file basis. So if you remove compute_13,sm_13 and are still getting that error, it is specified (still) for one of the files in your project. You then have to right-click on that file in the project explorer, and adjust the properties there.

6

Thank’s,

I have put the Macro commande:
image
and it’s work’s (more simply then use the macro)

7

I can ask you a question:
I have suddently somme Ntdll exception with the Visual 2019 Version,
from part of code witch work’s some time, but alwais work in Visual 2010:
For exemple here when I click for setup Buttons in Mocrosoft source code:

image
image1938×946 144 KB

Why ?