个人中心
文章发布
退出
作者热门文章
- html - 出于某种原因，IE8 对我的 Sass 文件中继承的 html5 CSS 不友好？
- JMeter 在响应断言中使用 span 标签的问题
- html - 在 :hover and :active? 上具有不同效果的 CSS 动画
- html - 相对于居中的 html 内容固定的 CSS 重复背景？
26
4
当我尝试在 GPU 上使用 tf.matmul 函数时，出现以下错误:
InternalError: Blas xGEMMBatched launch failed
如果函数 calc() 中的 N 值设置为小于 15 的值，则该函数有效。
我正在运行tensorflow 1.8.0和Cuda V9.1.85。只有一个 Python 进程在 GPU 上工作，并且没有其他打开的 session 。另外，我还有足够的 GPU 内存可供使用 ( see attached image )。
更改 CUDA_VISIBLE_DEVICES 值不会显示任何效果。更改 ConfigProto() 设置也没有帮助。此外，使用 tf.matmul 并不能解决问题。
这是我正在运行的代码:
import tensorflow as tf
import numpy as np
import os
os.environ["CUDA_VISIBLE_DEVICES"]="0"
config = tf.ConfigProto()
config.gpu_options.allow_growth=True
tf.Session(config=config).close()
def calc():
N = 15 # works for N <= 14
a = 16
b = 8
X = np.random.rand(N, 11520, b, 1).astype(np.float32)
print(X.nbytes*1e-6, "MB")
W = np.random.rand(N, 11520, a, b).astype(np.float32)
print(W.nbytes*1e-6, "MB")
X_ = tf.constant(X, name="X-constant", dtype=tf.float32)
W_ = tf.constant(W, name="W-constant", dtype=tf.float32)
# return tf.matmul(W_, X_, name="mymatmul")
return W_ @ X_
tf.reset_default_graph()
a = calc()
sess = tf.Session()
sess.run(tf.global_variables_initializer())
b = sess.run(a)
sess.close()
print(b.shape)
这是我得到的输出:
5.529599999999999 MB
88.47359999999999 MB
---------------------------------------------------------------------------
InternalError Traceback (most recent call last)
/usr/local/lib/python3.6/dist-packages/tensorflow/python/client/session.py in _do_call(self, fn, *args)
1321 try:
-> 1322 return fn(*args)
1323 except errors.OpError as e:
/usr/local/lib/python3.6/dist-packages/tensorflow/python/client/session.py in _run_fn(feed_dict, fetch_list, target_list, options, run_metadata)
1306 return self._call_tf_sessionrun(
-> 1307 options, feed_dict, fetch_list, target_list, run_metadata)
1308
/usr/local/lib/python3.6/dist-packages/tensorflow/python/client/session.py in _call_tf_sessionrun(self, options, feed_dict, fetch_list, target_list, run_metadata)
1408 self._session, options, feed_dict, fetch_list, target_list,
-> 1409 run_metadata)
1410 else:
InternalError: Blas xGEMMBatched launch failed : a.shape=[172800,16,8], b.shape=[172800,8,1], m=16, n=1, k=8, batch_size=172800
[[Node: matmul = BatchMatMul[T=DT_FLOAT, adj_x=false, adj_y=false, _device="/job:localhost/replica:0/task:0/device:GPU:0"](W-constant, X-constant)]]
During handling of the above exception, another exception occurred:
InternalError Traceback (most recent call last)
<ipython-input-5-013153235a1a> in <module>()
3 sess = tf.Session()
4 sess.run(tf.global_variables_initializer())
----> 5 b = sess.run(a)
6 sess.close()
7 print(b.shape)
/usr/local/lib/python3.6/dist-packages/tensorflow/python/client/session.py in run(self, fetches, feed_dict, options, run_metadata)
898 try:
899 result = self._run(None, fetches, feed_dict, options_ptr,
--> 900 run_metadata_ptr)
901 if run_metadata:
902 proto_data = tf_session.TF_GetBuffer(run_metadata_ptr)
/usr/local/lib/python3.6/dist-packages/tensorflow/python/client/session.py in _run(self, handle, fetches, feed_dict, options, run_metadata)
1133 if final_fetches or final_targets or (handle and feed_dict_tensor):
1134 results = self._do_run(handle, final_targets, final_fetches,
-> 1135 feed_dict_tensor, options, run_metadata)
1136 else:
1137 results = []
/usr/local/lib/python3.6/dist-packages/tensorflow/python/client/session.py in _do_run(self, handle, target_list, fetch_list, feed_dict, options, run_metadata)
1314 if handle is None:
1315 return self._do_call(_run_fn, feeds, fetches, targets, options,
-> 1316 run_metadata)
1317 else:
1318 return self._do_call(_prun_fn, handle, feeds, fetches)
/usr/local/lib/python3.6/dist-packages/tensorflow/python/client/session.py in _do_call(self, fn, *args)
1333 except KeyError:
1334 pass
-> 1335 raise type(e)(node_def, op, message)
1336
1337 def _extend_graph(self):
InternalError: Blas xGEMMBatched launch failed : a.shape=[172800,16,8], b.shape=[172800,8,1], m=16, n=1, k=8, batch_size=172800
[[Node: matmul = BatchMatMul[T=DT_FLOAT, adj_x=false, adj_y=false, _device="/job:localhost/replica:0/task:0/device:GPU:0"](W-constant, X-constant)]]
Caused by op 'matmul', defined at:
File "/usr/lib/python3.6/runpy.py", line 193, in _run_module_as_main
"__main__", mod_spec)
File "/usr/lib/python3.6/runpy.py", line 85, in _run_code
exec(code, run_globals)
File "/usr/local/lib/python3.6/dist-packages/ipykernel_launcher.py", line 16, in <module>
app.launch_new_instance()
File "/usr/local/lib/python3.6/dist-packages/traitlets/config/application.py", line 658, in launch_instance
app.start()
File "/usr/local/lib/python3.6/dist-packages/ipykernel/kernelapp.py", line 486, in start
self.io_loop.start()
File "/usr/local/lib/python3.6/dist-packages/tornado/platform/asyncio.py", line 127, in start
self.asyncio_loop.run_forever()
File "/usr/lib/python3.6/asyncio/base_events.py", line 422, in run_forever
self._run_once()
File "/usr/lib/python3.6/asyncio/base_events.py", line 1432, in _run_once
handle._run()
File "/usr/lib/python3.6/asyncio/events.py", line 145, in _run
self._callback(*self._args)
File "/usr/local/lib/python3.6/dist-packages/tornado/platform/asyncio.py", line 117, in _handle_events
handler_func(fileobj, events)
File "/usr/local/lib/python3.6/dist-packages/tornado/stack_context.py", line 276, in null_wrapper
return fn(*args, **kwargs)
File "/usr/local/lib/python3.6/dist-packages/zmq/eventloop/zmqstream.py", line 450, in _handle_events
self._handle_recv()
File "/usr/local/lib/python3.6/dist-packages/zmq/eventloop/zmqstream.py", line 480, in _handle_recv
self._run_callback(callback, msg)
File "/usr/local/lib/python3.6/dist-packages/zmq/eventloop/zmqstream.py", line 432, in _run_callback
callback(*args, **kwargs)
File "/usr/local/lib/python3.6/dist-packages/tornado/stack_context.py", line 276, in null_wrapper
return fn(*args, **kwargs)
File "/usr/local/lib/python3.6/dist-packages/ipykernel/kernelbase.py", line 283, in dispatcher
return self.dispatch_shell(stream, msg)
File "/usr/local/lib/python3.6/dist-packages/ipykernel/kernelbase.py", line 233, in dispatch_shell
handler(stream, idents, msg)
File "/usr/local/lib/python3.6/dist-packages/ipykernel/kernelbase.py", line 399, in execute_request
user_expressions, allow_stdin)
File "/usr/local/lib/python3.6/dist-packages/ipykernel/ipkernel.py", line 208, in do_execute
res = shell.run_cell(code, store_history=store_history, silent=silent)
File "/usr/local/lib/python3.6/dist-packages/ipykernel/zmqshell.py", line 537, in run_cell
return super(ZMQInteractiveShell, self).run_cell(*args, **kwargs)
File "/usr/local/lib/python3.6/dist-packages/IPython/core/interactiveshell.py", line 2662, in run_cell
raw_cell, store_history, silent, shell_futures)
File "/usr/local/lib/python3.6/dist-packages/IPython/core/interactiveshell.py", line 2785, in _run_cell
interactivity=interactivity, compiler=compiler, result=result)
File "/usr/local/lib/python3.6/dist-packages/IPython/core/interactiveshell.py", line 2903, in run_ast_nodes
if self.run_code(code, result):
File "/usr/local/lib/python3.6/dist-packages/IPython/core/interactiveshell.py", line 2963, in run_code
exec(code_obj, self.user_global_ns, self.user_ns)
File "<ipython-input-5-013153235a1a>", line 2, in <module>
a = calc()
File "<ipython-input-4-bf0e6012e9e2>", line 13, in calc
return W_ @ X_
File "/usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/math_ops.py", line 847, in binary_op_wrapper
return func(x, y, name=name)
File "/usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/math_ops.py", line 1976, in matmul
a, b, adj_x=adjoint_a, adj_y=adjoint_b, name=name)
File "/usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/gen_math_ops.py", line 1236, in batch_mat_mul
"BatchMatMul", x=x, y=y, adj_x=adj_x, adj_y=adj_y, name=name)
File "/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/op_def_library.py", line 787, in _apply_op_helper
op_def=op_def)
File "/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/ops.py", line 3414, in create_op
op_def=op_def)
File "/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/ops.py", line 1740, in __init__
self._traceback = self._graph._extract_stack() # pylint: disable=protected-access
InternalError (see above for traceback): Blas xGEMMBatched launch failed : a.shape=[172800,16,8], b.shape=[172800,8,1], m=16, n=1, k=8, batch_size=172800
[[Node: matmul = BatchMatMul[T=DT_FLOAT, adj_x=false, adj_y=false, _device="/job:localhost/replica:0/task:0/device:GPU:0"](W-constant, X-constant)]]
最佳答案
以下是用 tf.einsum 替换 tf.matmul 的解决方法。但是，您的代码可以在我的配备 NVIDIA 840M (2004 MiB RAM)、cudnn 7.0.5.15 和 cuda 9.0.176 的机器上运行(也许降级有帮助？)。
import tensorflow as tf
import numpy as np
sess = tf.Session()
N = 20
M = 11520
a = 16
b = 8
W = np.random.rand(N, M, a, b).astype(np.float32)
X = np.random.rand(N, M, b, 1).astype(np.float32)
# tf.einsum does not support numpy arrays, so wrap W and X in tf.constants
W2 = tf.constant(W)
X2 = tf.constant(X)
# tf.einsum does not support "..." as seen later in np.einsum
WX = tf.einsum("uvik,uvkj->uvij", W2, X2)
# same as:
#WX = tf.matmul(W2, X2)
# calculate W@X using tf.einsum
result1 = sess.run(WX)
# calculate W@X using np.einsum
result2 = np.einsum("...ik,...kj->...ij", W, X)
# calculate W@X by hand (just for illustrative purpose, too slow for practical use)
result3 = np.zeros((N, M, a, 1), dtype=np.float32)
for i in range(a):
for j in range(1):
for k in range(b):
result3[..., i, j] += W[..., i, k] * X[..., k, j]
# ensure that everything is correct
assert(np.allclose(result1, result2))
assert(np.allclose(result1, result3))
print("everything ok")
sess.close()
关于python - TensorFlow matmul : Blas xGEMMBatched launch failed，我们在Stack Overflow上找到一个类似的问题： https://stackoverflow.com/questions/50911052/
26
4
0
代码: x = tf.constant([1.,2.,3.], shape = (3,2,4)) y = tf.constant([1.,2.,3.], shape = (3,21,4)) tf.ma
有人可以解释一下，TensorFlow 的 eager 模式是如何工作的吗？我正在尝试构建一个简单的回归，如下所示: import tensorflow as tf tfe = tf.contrib.
我对 Tensorflow 很陌生。我已经在搜索相同的问题，但我不明白。有代码。希望你能帮助我。 代码: import tensorflow as tf w1 = tf.Variable(tf.ran
我对使用 * 和 matmul 的两个张量之间的乘法感到困惑。下面是我的代码 import torch torch.manual_seed(7) features = torch.randn((2,
我有 3 个张量 X 形状(1, c, h, w)，假设(1, 20, 40, 50) Fx 形状(num, w, N)，假设(1000, 50, 10) Fy shape (num, N, h)，假
我已经计算了 Fortran 的 MATMUL 函数使用不同乘法大小(32 × 32、64 × 64，...)花费的时间，我对结果有疑问。 这些是结果: SIZE ----- TIME IN SECO
a = [1, 2, 3] b = [10, 10, 10] np.matmul(a, b) 结果是 60。 numpy 如何乘以 (3,) 和 (3,) 维度并返回点积而不是外积(3 * 3)或抛出
我看到许多机器学习教程通过构造两个矩阵、权重矩阵和输入(或激活)矩阵来解释全连接网络，并执行矩阵到矩阵乘法(matmul)以形成线性方程。 我看到的所有示例都将输入作为 matmul 的第一个参数，将
当我在代码中的某行调用 np.matmul 时出现此错误。这是我在解决 python 调试器错误时得到的信息: > /home/marcos/Desktop/Machine_Learning_for_
我想在等级 2 和等级 3 的两个张量之间广播 tf.matmul 运算，其中一个包含“未知”形状的维度(基本上是特定维度中的“无”值) )。 问题是动态尺寸 tf.reshape 和 tf.broa
我尝试在 tensorflow 中编写和(逻辑运算)，有两个输入和两个权重将它们相乘得到一个数字并将这个数字加到偏差中，我在 matmul 中的问题是发送 X(输入)和 W(权重) 以方法形。[[1]
我正在研究并行编程概念并尝试优化单核上的矩阵乘法示例。到目前为止，我想出的最快的实现如下: /* This routine performs a dgemm operation * C := C
我有一些由 input_x 表示的数据。它是一个未知大小的张量(应该批量输入)，每个项目的大小为 n。 input_x 经历 tf.nn.embedding_lookup，因此 embed 现在具有维
在 Python 中，@ 运算符传递给元素的 __matmul__ 属性。当实现一个与实际后端无关的方法时，这会派上用场。例如 def inner(x, y): return x @ y
关闭。这个问题是opinion-based .它目前不接受答案。 想要改进这个问题吗？ 更新问题，以便 editing this post 提供事实和引用来回答它. 关闭 2 年前。 Improve
我正在尝试使用 tf.matmul() 执行稀疏矩阵乘法。 但是，推理速度比密集矩阵乘法慢得多。 根据 tf.sparse_matmul() 中的描述: 在一个平台上使用此乘法与密集矩阵相乘的盈亏平衡
我有一个我一直在努力解决的问题。与 tf.matmul() 相关并且没有广播。 我在 https://github.com/tensorflow/tensorflow/issues/216 上发现了类
我有许多带有形状的矩阵 w1、w2、w3...wn (k*n1 、k*n2、k*n3...k*nn) 和 x1、x2、x3...xn 具有形状(n1*m、n2*m、n3*m...nn*m >). 我想
我阅读了tf.matmul的官方文档我理解第一个例子。这是一个简单的 [2,3] x [3,2] 操作: a = tf.constant([1, 2, 3, 4, 5, 6], shape=[2, 3
我正在尝试使用 4D-numpy 数组数据在 TensorFlow 中实现多层感知器我在 MatMul 函数上遇到了这个问题。我希望有人能在这里帮助我，非常感谢。 ValueError: Shape
我是一名优秀的程序员,十分优秀！