Différences

Ci-dessous, les différences entre deux révisions de la page.

--- ressources:ressources:tensorflow [2018/08/21 09:56] – créée equemene
+++ ressources:ressources:tensorflow [2022/08/20 14:42] (Version actuelle) – equemene
@@ Ligne 1: / Ligne 1: @@
-====== Tensorflow ======
+<note warning>Documentation obsolète : elle correspond à un ancien système et ne doit pas être exploitée en ce moment. Pour exploiter le Machine Learning au CBP, suivez ce lien [[ressources:miniconda4cbp|Exploitation ou Installation de Miniconda au Centre Blaise Pascal]]</note>
-L'environnement [[https://www.tensorflow.org/|Tensorflow]] a été installé cet été au Centre Blaise Pascal.
-Deux environnements étaient disponibles mais seul l'environnement permettant d'exploiter les cartes GPU à base de circuit Nvidia a été installée. La version courante déployée est la 1.10.
+====== Tensorflow ======
+Un premier environnement [[https://www.tensorflow.org/|Tensorflow]] a été installé au printemps au Centre Blaise Pascal. Il exploite l'environnement Anaconda3-2019.3 installé dans le dossier ''/opt/anaconda3''
+Un second environnement [[https://www.tensorflow.org/|Tensorflow]] a été installé cet automne au Centre Blaise Pascal. Il exploite l'environnement Anaconda3-2019.10 installé dans le dossier ''/opt/anaconda3-2019.10''
 ===== Configuration d'exploitation =====
-L'environnement Tensorflow disponible exige une version spécifique de CUDA, la 9.0 minimum.
+Deux environnements Tensorflow sont installés dans deux Anaconda3 différents :
+  * l'environnement Tensorflow installé dans Anaconda3-2019.03 exploite une version de TensorFlow 1.12
+  * l'environnement Tensorflow installé dans Anaconda3-2019.10 exploite une version de TensorFlow 2.0
-Pour la charge dans le SIDUS standard du CBP:<code>module load cuda/9.0
+Pour charger l'environnement Tensorflow 1.12
+Pour la charge dans le SIDUS standard du CBP:<code>source /etc/tensorflow.init
 </code>
-===== Exemple =====
+Pour charger l'environnement Tensorflow 2.0
+Pour la charge dans le SIDUS standard du CBP:<code>source /etc/tensorflow2.init
+</code>
+Lorsqu'il est activé l'invite de commande est alors préfixée de ''(base)''. Par exemple, l'utilisateur ''einstein'' sur la machine ''ascenseur'' aura comme invite de commande :<code>(base) einstein@ascenseur:~$</code>
+===== Exemple pour Tensorflow 1.12 =====
 L'exemple suivant, fourni par le [[https://www.tensorflow.org/|site officiel]], permet de rapidement juger du fonctionnement de l'environnement. Il nécessite le lancement de l'interpréteur python :
@@ Ligne 23: / Ligne 36: @@
 </code>
-A l'exécution de la première ligne, vous pouvez avoir un message d'erreur sans importance :
+A l'exécution de la troisième ligne, l'environnement détecte les cartes graphiques susceptibles d'être exploitées :<code>
-<code>/usr/lib/python2.7/dist-packages/h5py/__init__.py:34: FutureWarning: Conversion of the second argument of issubdtype from `float` to `np.floating` is deprecated. In future, it will be treated as `np.float64 == np.dtype(float).type`.
+-06-11 18:03:17.928752: I tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: SSE4.1 SSE4.2 AVX
-  from ._conv import register_converters as _register_converters
+-06-11 18:03:17.938098: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 1995195000 Hz
+-06-11 18:03:17.938873: I tensorflow/compiler/xla/service/service.cc:150] XLA service 0x562f3dba5a30 executing computations on platform Host. Devices:
+-06-11 18:03:17.938924: I tensorflow/compiler/xla/service/service.cc:158]   StreamExecutor device (0): <undefined>, <undefined>
+-06-11 18:03:18.167596: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:998] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
+-06-11 18:03:18.203546: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:998] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
+-06-11 18:03:18.205179: I tensorflow/compiler/xla/service/service.cc:150] XLA service 0x562f3dc62f80 executing computations on platform CUDA. Devices:
+-06-11 18:03:18.205294: I tensorflow/compiler/xla/service/service.cc:158]   StreamExecutor device (0): GeForce RTX 2080, Compute Capability 7.5
+-06-11 18:03:18.205385: I tensorflow/compiler/xla/service/service.cc:158]   StreamExecutor device (1): GeForce GT 730, Compute Capability 3.5
+-06-11 18:03:18.206798: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1433] Found device 0 with properties:
+name: GeForce RTX 2080 major: 7 minor: 5 memoryClockRate(GHz): 1.71
+pciBusID: 0000:04:00.0
+totalMemory: 7.77GiB freeMemory: 7.65GiB
+-06-11 18:03:18.207245: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1433] Found device 1 with properties:
+name: GeForce GT 730 major: 3 minor: 5 memoryClockRate(GHz): 0.9015
+pciBusID: 0000:03:00.0
+totalMemory: 1.95GiB freeMemory: 1.90GiB
+-06-11 18:03:18.207374: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1497] Ignoring visible gpu device (device: 1, name: GeForce GT 730, pci bus id: 0000:03:00.0, compute capability: 3.5) with Cuda multiprocessor count: 2. The minimum required count is 8. You can adjust this requirement with the env var TF_MIN_GPU_MULTIPROCESSOR_COUNT.
+-06-11 18:03:18.207456: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1512] Adding visible gpu devices: 0
+-06-11 18:03:18.210563: I tensorflow/core/common_runtime/gpu/gpu_device.cc:984] Device interconnect StreamExecutor with strength 1 edge matrix:
+-06-11 18:03:18.210632: I tensorflow/core/common_runtime/gpu/gpu_device.cc:990]      0 1
+-06-11 18:03:18.210687: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1003] 0:   N N
+-06-11 18:03:18.210734: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1003] 1:   N N
+-06-11 18:03:18.211630: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1115] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 7439 MB memory) -> physical GPU (device: 0, name: GeForce RTX 2080, pci bus id: 0000:04:00.0, compute capability: 7.5)
 </code>
+Dans l'exemple ci-dessus, il s'agit d'une carte  d'une carte GeForce GT 730 avec 1.95GiB de RAM et d'une carte GeForce RTX 2080 avec 7.65GiB de RAM
-A l'exécution de la seconde ligne, l'environnement détecte la carte graphique susceptible d'être exploitée :<code>
--08-21 11:24:28.475954: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:897] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
--08-21 11:24:28.476656: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1405] Found device 0 with properties:
-name: GeForce GTX TITAN major: 3 minor: 5 memoryClockRate(GHz): 0.8755
-pciBusID: 0000:06:00.0
-totalMemory: 5.94GiB freeMemory: 5.83GiB
--08-21 11:24:28.476704: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1484] Adding visible gpu devices: 0
--08-21 11:24:34.039086: I tensorflow/core/common_runtime/gpu/gpu_device.cc:965] Device interconnect StreamExecutor with strength 1 edge matrix:
--08-21 11:24:34.039883: I tensorflow/core/common_runtime/gpu/gpu_device.cc:971]      0
--08-21 11:24:34.039922: I tensorflow/core/common_runtime/gpu/gpu_device.cc:984] 0:   N
--08-21 11:24:34.040245: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1097] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 5626 MB memory) -> physical GPU (device: 0, name: GeForce GTX TITAN, pci bus id: 0000:06:00.0, compute capability: 3.5)
-</code>
-Dans l'exemple ci-dessus, il s'agit d'une carte GeForce GTX TITAN avec 5.94GiB de RAM.
 La dernière ligne permet s'assurer que la session fonctionne de manière nominal en affichant :<code>
@@ Ligne 47: / Ligne 69: @@
 Une grande variété de [[https://www.tensorflow.org/tutorials/|tutoriels]] en ligne permettent de vérifier le bon fonctionnement.
+===== Exemple pour Tensorflow 2.0 =====
+L'exemple précédent ne fonctionne pas : voici un petit exemple qui fonctionne pour tester son tensorflow 2.0
+<code>
+# Python
+from __future__ import absolute_import, division, print_function, unicode_literals
+import tensorflow as tf
+# Create some tensors
+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 = tf.matmul(a, b)
+print(c)
+</code>
+A l'exécution de la troisième ligne, l'environnement détecte les cartes graphiques susceptibles d'être exploitées :<code>
+-11-04 17:58:31.722039: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcuda.so.1
+-11-04 17:58:31.749152: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1006] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
+-11-04 17:58:31.749646: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1618] Found device 0 with properties:
+name: Tesla V100-PCIE-16GB major: 7 minor: 0 memoryClockRate(GHz): 1.38
+pciBusID: 0000:07:00.0
+-11-04 17:58:31.750304: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.0
+-11-04 17:58:31.752157: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10.0
+-11-04 17:58:31.753851: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcufft.so.10.0
+-11-04 17:58:31.755073: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcurand.so.10.0
+-11-04 17:58:31.757047: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusolver.so.10.0
+-11-04 17:58:31.758713: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusparse.so.10.0
+-11-04 17:58:31.762588: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
+-11-04 17:58:31.762748: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1006] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
+-11-04 17:58:31.763286: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1006] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
+-11-04 17:58:31.763714: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1746] Adding visible gpu devices: 0
+-11-04 17:58:31.763967: I tensorflow/core/platform/cpu_feature_guard.cc:142] Your CPU supports instructions that this TensorFlow binary was not compiled to use: SSE4.1 SSE4.2 AVX AVX2 AVX512F FMA
+-11-04 17:58:31.768995: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 3192495000 Hz
+-11-04 17:58:31.769414: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x55a533b0ac90 executing computations on platform Host. Devices:
+-11-04 17:58:31.769443: I tensorflow/compiler/xla/service/service.cc:175]   StreamExecutor device (0): Host, Default Version
+-11-04 17:58:31.769643: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1006] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
+-11-04 17:58:31.770095: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1618] Found device 0 with properties:
+name: Tesla V100-PCIE-16GB major: 7 minor: 0 memoryClockRate(GHz): 1.38
+pciBusID: 0000:07:00.0
+-11-04 17:58:31.770124: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.0
+-11-04 17:58:31.770139: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10.0
+-11-04 17:58:31.770150: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcufft.so.10.0
+-11-04 17:58:31.770164: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcurand.so.10.0
+-11-04 17:58:31.770178: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusolver.so.10.0
+-11-04 17:58:31.770190: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcusparse.so.10.0
+-11-04 17:58:31.770202: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudnn.so.7
+-11-04 17:58:31.770292: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1006] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
+-11-04 17:58:31.770799: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1006] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
+-11-04 17:58:31.771228: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1746] Adding visible gpu devices: 0
+-11-04 17:58:31.771260: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcudart.so.10.0
+-11-04 17:58:31.859125: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1159] Device interconnect StreamExecutor with strength 1 edge matrix:
+-11-04 17:58:31.859173: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1165]      0
+-11-04 17:58:31.859183: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1178] 0:   N
+-11-04 17:58:31.859388: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1006] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
+-11-04 17:58:31.859930: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1006] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
+-11-04 17:58:31.860424: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:1006] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
+-11-04 17:58:31.860871: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1304] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 14961 MB memory) -> physical GPU (device: 0, name: Tesla V100-PCIE-16GB, pci bus id: 0000:07:00.0, compute capability: 7.0)
+-11-04 17:58:31.862849: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x55a534edf7d0 executing computations on platform CUDA. Devices:
+-11-04 17:58:31.862884: I tensorflow/compiler/xla/service/service.cc:175]   StreamExecutor device (0): Tesla V100-PCIE-16GB, Compute Capability 7.0
+-11-04 17:58:31.863918: I tensorflow/stream_executor/platform/default/dso_loader.cc:44] Successfully opened dynamic library libcublas.so.10.0
+tf.Tensor(
+[[22. 28.]
+ [49. 64.]], shape=(2, 2), dtype=float32)
+</code>
+On reconnait la carte **Tesla V100-PCIE-16GB** détectée.
+===== Utilisation de TensorFlow2 pour du TensorFlow1 =====
+Il est possible, comme le précise la [[https://www.tensorflow.org/guide/migrate|documentation de TensorFlow]], d'exécuter un code de l'ancienne version sur la nouvelle avec le chargement suivant :
+<code>
+import tensorflow.compat.v1 as tf
+tf.disable_v2_behavior()
+</code>
+===== Astuces en cas de plantage incompréhensible =====
+Il se peut que dans l'utilisation, des plantages apparaissent avec comme source de première erreur ''CUDNN_STATUS_INTERNAL_ERROR''
+<code>2019-09-28 05:35:09.764756: E tensorflow/stream_executor/cuda/cuda_dnn.cc:334] Could not create cudnn handle: CUDNN_STATUS_INTERNAL_ERROR
+-09-28 05:35:09.766851: E tensorflow/stream_executor/cuda/cuda_dnn.cc:334] Could not create cudnn handle: CUDNN_STATUS_INTERNAL_ERROR
+---
+tensorflow.python.framework.errors_impl.UnknownError: Failed to get convolution algorithm. This is probably because cuDNN failed to initialize, so try looking to see if a warning log message was printed above.
+---
+</code>
+Cette erreur aura cela de désarmant qu'elle ne va pas apparaître sur toutes les machines exploitables avec TensorFlow.
+Une solution peut être de définir une variable d'environnement ''$TF_FORCE_GPU_ALLOW_GROWTH'' à ''true''
+<code>
+export TF_FORCE_GPU_ALLOW_GROWTH=true
+</code>
+Cette option (comme son nom l'indique) permet au GPU de conserver la mémoire déjà allouée. Le souci, c'est que cette allocation ne prend fin que lorsque l'exécution est terminée.
 <note important>L'application TensorFlow sur GPU requiert une version de GPU avec des capacités minimales pour fonctionner (compute capability de 3.5) : sur les 50 GPU accessibles au CBP, certains ne supportent pas cette capacité</note>
@@ Ligne 55: / Ligne 168: @@
   * circuits Maxwell : GTX 980Ti
   * circuits Pascal  : GTX 1060, GTX 1070, GTX 1080, GTX 1080Ti, Tesla P100
+  * circuits Turing  : RTX Titan, RTX 2080 Ti, RTX 2080 Super
+  * circuits Volta   : Tesla V100
 ===== GPU du CBP invalidés pour TensorFlow  =====
@@ Ligne 60: / Ligne 175: @@
   * circuits GT200   : Quadro FX4000, Tesla C1060
   * circuits Fermi   : GTX 560Ti, Quadro 4000
-  * circuits Kepler  : GTX 680, GTX 690,
-  * circuits Maxwell : GTX 960, GTX 970, GTX980
-  * circuits Pascal  :

ressources/ressources/tensorflow.1534845398.txt.gz · Dernière modification : 2025/05/14 07:38 (modification externe)

Rechercher

Translations

Piste :

toolbox