Replies: 1 comment 2 replies
-
|
Hello @571991380, Thank you for sharing the full stack trace, modules installed and details about your system, which are all quite helpful. Since the error seems to be occuring in sionna/test/unit/mimo/test_mimo_utils.py Lines 111 to 132 in 7eb04f4 Another thing to check: does a simple script using only Tensorflow work correctly? |
Beta Was this translation helpful? Give feedback.
-
|
Thanks for your help @merlinND .Following your advice, I ran this test separately。 #
# SPDX-FileCopyrightText: Copyright (c) 2021-2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
# SPDX-License-Identifier: Apache-2.0
#
try:
import sionna
except ImportError as e:
import sys
sys.path.append("../")
import unittest
import numpy as np
import tensorflow as tf
from sionna.utils.tensors import matrix_inv
gpus = tf.config.list_physical_devices('GPU')
print('Number of GPUs available :', len(gpus))
if gpus:
gpu_num = 0 # Number of the GPU to be used
try:
tf.config.set_visible_devices(gpus[gpu_num], 'GPU')
print('Only GPU number', gpu_num, 'used.')
tf.config.experimental.set_memory_growth(gpus[gpu_num], True)
except RuntimeError as e:
print(e)
from sionna.mimo.utils import complex2real_vector
from sionna.mimo.utils import complex2real_covariance
from sionna.utils import matrix_sqrt, complex_normal
from sionna.channel.utils import exp_corr_mat
class Complex2Real(unittest.TestCase):
def test_covariance_statistics(self):
"""Test that the statisics of the real-valued equivalent random
vector match the target statistics"""
batch_size = 1000000
num_batches = 100
n = 8
r = exp_corr_mat(0.8, 8)
rr = complex2real_covariance(r)
r_12 = matrix_sqrt(r)
@tf.function(jit_compile=True)
def fun():
w = tf.matmul(r_12, complex_normal([n, batch_size]))
w = tf.transpose(w, perm=[1,0])
wr = complex2real_vector(w)
r_hat = tf.matmul(wr, wr, transpose_a=True)/tf.cast(batch_size, wr.dtype)
return r_hat
r_hat = tf.zeros_like(rr)
for _ in range(num_batches):
r_hat += fun()/num_batches
self.assertTrue(np.max(np.abs(rr-r_hat))<1e-3)When i run it in terminal in pycharm, the result is: But When I click to run directly in PyCharm, the result is: Another thing is I try to run .pynb files in examples folder. I can run these files successfully in jupyter notebook, but there are a few files where the kernel crashes when using my env py38_ft210, although they run successfully on the default env python3.This should demonstrate that TensorFlow and the GPU are functioning properly. |
Beta Was this translation helpful? Give feedback.
-
|
Hello @571991380, The test script that you ran executes the imports succesfully (which is good), but the actual test code is not run. if __name__ == "__main__":
test_covariance_statistics() |
Beta Was this translation helpful? Give feedback.
-
Hello, I am a beginner in coding. I installed sionna and when running all the '. py' files in the test folder in pycharm, there will always be the following errors.Could you help me.
My Environment List：
My computer：
Beta Was this translation helpful? Give feedback.
All reactions