Backtrace

The backtrace command displays a summary of how your program got where it is. Consider the following example numba_dpex/examples/debug/simple_dppy_func.py:

import dpctl
import numpy as np

import numba_dpex as dppy


@dppy.func(debug=True)
def func_sum(a_in_func, b_in_func):
    result = a_in_func + b_in_func
    return result


@dppy.kernel(debug=True)
def kernel_sum(a_in_kernel, b_in_kernel, c_in_kernel):
    i = dppy.get_global_id(0)
    c_in_kernel[i] = func_sum(a_in_kernel[i], b_in_kernel[i])


global_size = 10
a = np.arange(global_size, dtype=np.float32)
b = np.arange(global_size, dtype=np.float32)
c = np.empty_like(a)

device = dpctl.SyclDevice("opencl:gpu")
with dpctl.device_context(device):
    kernel_sum[global_size, dppy.DEFAULT_LOCAL_SIZE](a, b, c)

print("Done...")

The section presents two examples of using Intel Distribution for GDB* to generate backtrace from a numa_dppy.kernel function. The first example presents the case where the kernel function does not invoke any other function. The second example presents the case where the kernel function invokes a numba_dpex.func.

Example 1:

$ NUMBA_OPT=0 gdb-oneapi -q python
(gdb) set breakpoint pending on
(gdb) break simple_dppy_func.py:28
(gdb) run simple_dppy_func.py
...
Thread 2.2 hit Breakpoint 1, with SIMD lanes [0-7], __main__::kernel_sum () at simple_dppy_func.py:28
28          i = dppy.get_global_id(0)
(gdb) backtrace
#0  __main__::kernel_sum () at simple_dppy_func.py:28
(gdb) continue
...
[Switching to Thread 1.1073742080 lane 0]
Thread 2.3 hit Breakpoint 1, with SIMD lanes [0-1], __main__::kernel_sum () at simple_dppy_func.py:28
28          i = dppy.get_global_id(0)
(gdb) continue
...
Done...

Example 2:

$ NUMBA_OPT=0 gdb-oneapi -q python
(gdb) set breakpoint pending on
(gdb) break simple_dppy_func.py:22
(gdb) run simple_dppy_func.py
...
Thread 2.2 hit Breakpoint 1, with SIMD lanes [0-7], __main__::func_sum () at simple_dppy_func.py:22
22          result = a_in_func + b_in_func
(gdb) backtrace
#0  __main__::func_sum () at simple_dppy_func.py:22
#1  __main__::kernel_sum () at simple_dppy_func.py:29
(gdb) continue
...
[Switching to Thread 1.1073742080 lane 0]
Thread 2.3 hit Breakpoint 1, with SIMD lanes [0-1], __main__::func_sum () at simple_dppy_func.py:22
22          result = a_in_func + b_in_func
(gdb) continue
...
Done...

See also:

• Backtraces in GDB*