Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

macbook pro GPU version, successfully opened CUDA library, but not found the GPU #4858

Closed
wwxFromTju opened this issue Oct 9, 2016 · 15 comments
Closed

macbook pro GPU version, successfully opened CUDA library, but not found the GPU #4858

wwxFromTju opened this issue Oct 9, 2016 · 15 comments

Comments

@wwxFromTju
Copy link

wwxFromTju commented Oct 9, 2016 •
edited
Loading

Hi ALL:
I am a chinese student, so may be English is bas. Sorry.
now I use the tensorflow GPU version. I user the pip to down the python3 GPU version, and down the all about the GPU file.
But i can successfully opened CUDA library, but not found the GPU, like this:
In [1]: import tensorflow as tf I tensorflow/stream_executor/dso_loader.cc:108] successfully opened CUDA library libcublas.dylib locally I tensorflow/stream_executor/dso_loader.cc:108] successfully opened CUDA library libcudnn.dylib locally I tensorflow/stream_executor/dso_loader.cc:108] successfully opened CUDA library libcufft.dylib locally I tensorflow/stream_executor/dso_loader.cc:108] successfully opened CUDA library libcuda.1.dylib locally I tensorflow/stream_executor/dso_loader.cc:108] successfully opened CUDA library libcurand.dylib locally

In [2]: tf.Session() E tensorflow/stream_executor/cuda/cuda_driver.cc:491] failed call to cuInit: CUDA_ERROR_NO_DEVICE I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:153] retrieving CUDA diagnostic information for host: wangxiaoweideWindows.local I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:160] hostname: wangxiaoweideWindows.local I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:185] libcuda reported version is: 346.3.6 I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:189] kernel reported version is: Invalid argument: expected %d.%d or %d.%d.%d form for driver version; got "" I tensorflow/core/common_runtime/gpu/gpu_init.cc:81] No GPU devices available on machine.

and I find the I can't deviceQuery the GPU:

➜ ~ nvcc -V nvcc: NVIDIA (R) Cuda compiler driver Copyright (c) 2005-2015 NVIDIA Corporation Built on Mon_Apr_11_13:23:40_CDT_2016 Cuda compilation tools, release 7.5, V7.5.26

➜ ~ ~/cuda-samples/bin/x86_64/darwin/release/deviceQuery /Users/codeMan/cuda-samples/bin/x86_64/darwin/release/deviceQuery Starting... CUDA Device Query (Runtime API) version (CUDART static linking) cudaGetDeviceCount returned 38 -> no CUDA-capable device is detected Result = FAIL

now I use the mac os, and Xcode8, how to solve it ?
@yaroslavvb
Copy link
Contributor

If you can't run deviceQuery, then the problem is not with TensorFlow. Perhaps your computer does not have a CUDA-capable GPU

@wwxFromTju
Copy link
Author

@yaroslavvb
My macbook pro 2014 have NVIDIA 750M, and I find the mac GPU course to download the library.
so how to solve? I don't know. THX

@wwxFromTju
Copy link
Author

wwxFromTju commented Oct 10, 2016 •
edited
Loading

OK. I solve this problem, is easy:NVIDIA CUDA 7.5 FOR MAC OS X RELEASE
download this, then OK!!!!!!
http://www.nvidia.com/object/macosx-cuda-7.5.30-driver.html

@yaroslavvb THX!

@jart
Copy link
Contributor

jart commented Oct 14, 2016

Glad you solved the problem @wwxFromTju. We hope TensorFlow serves you well in your studies. If you have any other questions in the future, the Stack Overflow community is a better forum for support. We try to keep this issue tracker focused on bugs and feature requests.

@jart jart closed this as completed Oct 14, 2016
@CosmosShadow
Copy link

Thanks, That helps me.

@laventura
Copy link

Hello @jart @wwxFromTju

I have similar problem as yours.:

Any help appreciated!! thanks!

"libcuda reported version is: 310.42.25;
kernel reported version is: Invalid argument: expected %d.%d or %d.%d.%d form for driver version; got ""
My detailed problem is here: #2882

In brief: CUDA Driver (from Apple > System Preferences > CUDA)

CUDA Driver Version: 8.0.57
GPU Driver Version: 10.10.14 310.42.25f02

My setup:
I'm on Mac OS X 10.11.6. Installed CUDA 8, and CUDNN 5.1. I've followed the install instructions from official TF site. (as of Jan 23, 2017)

./deviceQuery 
./deviceQuery Starting...

 CUDA Device Query (Runtime API) version (CUDART static linking)

cudaGetDeviceCount returned 38
-> no CUDA-capable device is detected
Result = FAIL

@yaroslavvb
Copy link
Contributor

@laventura sounds like you don't have CUDA-capable device (if your video card is AMD, it is not CUDA-capable)

@laventura
Copy link

@yaroslavvb

Here are the reports from System Report: (About This Mac > System Report > Graphics )

So - does this support CUDA or does it not? I 'm at a loss to figure this out now.

NVIDIA GeForce GT 750M:
  Chipset Model:	NVIDIA GeForce GT 750M
  Type:	GPU
  Bus:	PCIe
  PCIe Lane Width:	x8
  VRAM (Total):	2048 MB
  Vendor:	NVIDIA (0x10de)
  Device ID:	0x0fe9
  Revision ID:	0x00a2
  ROM Revision:	3776
  gMux Version:	4.0.8 [3.2.8]
  Displays:
Color LCD:
  Display Type:	Retina LCD
  Resolution:	2880 x 1800 Retina
  Retina:	Yes
  Pixel Depth:	32-Bit Color (ARGB8888)
  Main Display:	Yes
  Mirror:	Off
  Online:	Yes
  Built-In:	Yes

---- 
Intel Iris Pro:
  Chipset Model:	Intel Iris Pro
  Type:	GPU
  Bus:	Built-In
  VRAM (Dynamic, Max):	1536 MB
  Vendor:	Intel (0x8086)
  Device ID:	0x0d26
  Revision ID:	0x0008
  gMux Version:	4.0.8 [3.2.8]

@yaroslavvb
Copy link
Contributor

Yes it should. It seems your problem is outside of tensorflow since devicequery fails, maybe Nvidia support forums have tips

@laventura
Copy link

I've been running between CUDA/GPU and Tensorflow issues... 😩😩

TensorFlow on GPU worked for me earlier, with an older version of TensorFlow (0.10? cant recall) and perhaps older CUDA too (also can't recall)

I've seen the networks train faster with this same GPU...

Unfortunately, upgrading TF led to many of these errors.. and now I can't figure out if it is GPU/CUDA problem or TF.

Of course, nobody in the Nvidia DevTalk forums have had similar issues... 😟

[The older problem was that TF couldn't use all the GPU memory....]

(See here: aymericdamien/TensorFlow-Examples#38 (comment))

@yaroslavvb
Copy link
Contributor

yaroslavvb commented Jan 25, 2017 •
edited
Loading

./deviceQuery failing indicates this is is a general GPU/CUDA problem, not TF problem

@biomassives
Copy link

biomassives commented Jan 26, 2017 via email

@vrv
Copy link

vrv commented Jan 26, 2017

@biomassives you seem to have started spamming github threads, so I have blocked you. Please contact us offline once you've fixed whatever you're doing to stop spamming threads.

@richardxiong
Copy link

@laventura hello, I'm running exactly the same problem as you, with ./deviceQuery failing. Did you solve that problem? Thanks!

@richardxiong
Copy link

@laventura Hello, have you solved your problem? I have run into exactly the same problem, with ./deviceQuery failing indicates the CUDA-capable device, but I follow the instruction on here. Thanks!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
8 participants
@yaroslavvb @jart @laventura @vrv @biomassives @CosmosShadow @wwxFromTju @richardxiong