`SIGSEGV` at `~CudaDeviceCompilationContext()` when using Nvidia GPU

[bagrorg]

Hi everyone!

I have the following code in test.py:

import pyhdk 
import pyarrow as pa
import numpy as np

pyhdk.initLogger(debug_logs=True)

arr = np.linspace(0, 100000, 30)
arr2 = np.linspace(10, 300000, 30)
arr3 = np.linspace(5000, 600000, 30)

np.random.shuffle(arr)
np.random.shuffle(arr2)
np.random.shuffle(arr3)


hdk = pyhdk.init(cpu_only=False)
ht = hdk.import_pydict(
            {
                "a": arr, 
                "b": arr2, 
                "c": arr3, 
            }
        )

print(ht.sort("a", "b").run(device_type="GPU"))

When running it with python test.py i have the following error:

$ python test.py 
Schema:
  a: FP64
  b: FP64
  c: FP64
Data:
0|72421.4|517931
3448.28|10|435862
6896.55|186211|312759
10344.8|227589|271724
13793.1|51732.4|230690
17241.4|113799|333276
20689.7|82765.9|210172
24137.9|196555|353793
27586.2|144833|415345
31034.5|20699|579483
34482.8|206900|374310
37931|134488|189655
41379.3|217244|476897
44827.6|41387.9|394828
48275.9|300000|128103
51724.1|248278|5000
55172.4|258622|558966
58620.7|155177|25517.2
62069|268967|538448
65517.2|10354.5|46034.5
68965.5|289656|66551.7
72413.8|165522|169138
75862.1|237933|456379
79310.3|93110.3|497414
82758.6|31043.4|107586
86206.9|279311|600000
89655.2|62076.9|251207
93103.4|124144|292241
96551.7|175866|148621
100000|103455|87069

#
# A fatal error has been detected by the Java Runtime Environment:
#
#  SIGSEGV (0xb) at pc=0xffffffffffffffff, pid=48818, tid=48818
#
# JRE version: OpenJDK Runtime Environment (17.0.3) (build 17.0.3-internal+0-adhoc..src)
# Java VM: OpenJDK 64-Bit Server VM (17.0.3-internal+0-adhoc..src, mixed mode, sharing, tiered, compressed oops, compressed class ptrs, g1 gc, linux-amd64)
# Problematic frame:
# C  0xffffffffffffffff
#
# Core dump will be written. Default location: Core dumps may be processed with "/usr/share/apport/apport -p%p -s%s -c%c -d%d -P%P -u%u -g%g -- %E" (or dumping to /home/bagrorg/Documents/thesis/hdk_notebooks/core.48818)
#
# An error report file with more information is saved as:
# /home/bagrorg/Documents/thesis/hdk_notebooks/hs_err_pid48818.log
#
# If you would like to submit a bug report, please visit:
#   https://bugreport.java.com/bugreport/crash.jsp
#
Aborted (core dumped)

Im am installing HDK with Coda environment, provided at omniscidb/scripts/mapd-deps-conda-dev-env.yml. I am building project as follows:

cmake -DENABLE_CUDA=on -DCMAKE_BUILD_TYPE=Debug ..
make -j 12
make install

My GPU:

$ nvidia-smi -L
GPU 0: NVIDIA GeForce GTX 1650 SUPER (UUID: GPU-94a184c0-9d0d-9ae1-7ae4-4925c6e7bb77)

$ nvidia-smi
Sat Apr  1 20:24:28 2023       
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 510.108.03   Driver Version: 510.108.03   CUDA Version: 11.6     |
|-------------------------------+----------------------+----------------------+
...

Error and executions logs are also attached.

This issue is reproduced in the main branch.

Maybe I am using/installing something wrong?

Thanks!

hs_err_pid48818.log PyHDK.INFO.20230401-201215.log