Jetson AGX Orin: Error running pycuda

[md.anowar]

I am trying to run the following program on NVIDIA AGX Orin, Ubuntu 20.04:

import NumPy
import pycuda.autoinit
import pycuda.driver as drive
from pycuda.compiler import SourceModule

SourceModule compiles C code for CUDA

mod = SourceModule(“”"
global void multiply_them(float *dest, float *a, float *b)
{
const int i = threadIdx.x; dest[i] = a[i] * b[i];
}
“”")
multiply_them = mod.get_function(“multiply_them”)
a = numpy.random.randn(400).astype(numpy.float32)
b = numpy.random.randn(400).astype(numpy.float32)
dest = numpy.zeros_like(a)

Perform the computation

multiply_them( drive.Out(dest), drive.In(a), drive.In(b), block=(400,1,1), grid=(1,1))
print(dest-a*b)

However, I am getting an error as shown below:

[AastaLLL]

Hi,

How do you install pyCUDA?
Do you use pip3 install pycuda --user?

Thanks.

[md.anowar]

Yes, I used pip3 install pycuda --user

[AastaLLL]

Hi,

Could you share a reproducible source and steps so we can give it a try?
More, does the same app work on other environments like an x86 desktop?

Thanks.

[md.anowar]

Thank you for your reply.
I was trying to run the attached Python script, using the following command:
python3 myGPUTest.py
myGPUTest.txt (630 Bytes)
I didn’t test on x86 desktop.
I have connected the NVIDIA AGX Orin directly to a monitor, and trying to run this pyCUDA program.

[AastaLLL]

Hi,

Could you help to check the format of the reproducible source?
We tried to run the source but failed in some format error rather than reproducing the issue.

...
pycuda.driver.CompileError: nvcc compilation of /tmp/tmpvqci04d2/kernel.cu failed
[command: nvcc --cubin -arch sm_87 -I/home/nvidia/.local/lib/python3.8/site-packages/pycuda/cuda kernel.cu]
[stderr:
kernel.cu(3): error: this declaration has no storage class or type specifier

kernel.cu(3): error: expected a ";"

kernel.cu(8): warning: parsing restarts here after previous syntax error

2 errors detected in the compilation of "kernel.cu".
]

Thanks.