Is Python 3 in dynamo use GPU or CPU?
I think this is an extra setting to decide if the machine uses GPU or not, I am using TensorFlow now and I have to install extra Cuda Toolkit to use GPU.
2 Likes
until now i cant acces GPU from Dynamo.
I installed Cudatoolkit but i can’t access to GPU from dynamo
you have use cudnn and tensorflow gpu?
https://developer.nvidia.com/cudnn
Let check version tensorflow before install to map with cuda, at the moment support 11.2
use command line :
pip install --upgrade tensorflow-gpu --user
and check again
# importing the tensorflow package
import tensorflow as tf
tf.test.is_built_with_cuda()
tf.test.is_gpu_available(cuda_only=False, min_cuda_compute_capability=None)
let try quick check in jupyter notebook first
import sys
import tensorflow.keras
import pandas as pd
import sklearn as sk
import tensorflow as tf
print(f"Tensor Flow Version: {tf.__version__}")
print(f"Keras Version: {tensorflow.keras.__version__}")
print()
print(f"Python {sys.version}")
print(f"Pandas {pd.__version__}")
print(f"Scikit-Learn {sk.__version__}")
gpu = len(tf.config.list_physical_devices('GPU'))>0
print("GPU is", "available" if gpu else "NOT AVAILABLE")
no, colab use gpu from server of google, we are setup in local to can run with local.
1 Like
Let complete setup cuda, install drive ndivia,cudnn to support first , next check again and use it in dynamo core, can show me info card vga in your computer ?
2 Likes
Try set some config like that
gpus = tf.config.list_physical_devices('GPU')
if gpus:
# Create 2 virtual GPUs with 1GB memory each
try:
tf.config.set_logical_device_configuration(
gpus[0],
[tf.config.LogicalDeviceConfiguration(memory_limit=1024),
tf.config.LogicalDeviceConfiguration(memory_limit=1024)])
logical_gpus = tf.config.list_logical_devices('GPU')
print(len(gpus), "Physical GPU,", len(logical_gpus), "Logical GPUs")
except RuntimeError as e:
# Virtual devices must be set before GPUs have been initialized
print(e)
if result return 1 Physical GPU, 2 Logical GPUs, you can run it normal in Dynamo.
This is a sample constant to check whether function run with GPU or not
tf.debugging.set_log_device_placement(True)
gpus = tf.config.list_logical_devices('GPU')
if gpus:
# Replicate your computation on multiple GPUs
c = []
for gpu in gpus:
with tf.device(gpu.name):
a = tf.constant([[1.0, 2.0, 3.0], [4.0, 5.0, 6.0]])
b = tf.constant([[1.0, 2.0], [3.0, 4.0], [5.0, 6.0]])
c.append(tf.matmul(a, b))
with tf.device('/CPU:0'):
matmul_sum = tf.add_n(c)
print(matmul_sum)
Result return a Tensor(“AddN:0”, shape=(2, 2), dtype=float32, device=/device:CPU:0)
You can use it in dynamo now.