How to compile PTX code

Question

I need to modify the PTX code and compile it directly. The reason is that I want to have some specific instructions right after each other and it is difficult to write a cud

Answer 1

This sequence of nvcc commands seems to do the trick. Please see here for more details.

Create your ptx files to modify

nvcc file1.cu file2.cu file3.cu -rdc=true --ptx

Link ptx files into an object file

nvcc file1.ptx file2.ptx file3.ptx -dlink

I did this on Windows so it popped out a_dlink.obj. As the documentation points out host code has been discarded by this point. Run

nvcc file1.cu file2.cu file3.cu -rdc=true --compile

to create object files. They will be .obj for Windows or .o for Linux. Then create a library output file

nvcc file1.obj file2.obj file3.obj a_dlink.obj --lib -o myprogram.lib

Then run

nvcc myprogram.lib

which will pop out an exectuable a.exe on Windows or a.out on Linux. This procedure works for cubin and fatbin files too. Just substitute those names in place of ptx.