# CUDA 10.1 + Tensorflow 1.13

**URL:** https://forums.developer.nvidia.com/t/cuda-10-1-tensorflow-1-13/70940
**Category:** CUDA Setup and Installation
**Created:** [February 27, 2019, 6:42pm UTC](https://forums.developer.nvidia.com/t/cuda-10-1-tensorflow-1-13/70940 "2019-02-27T18:42:20Z")
**Posts on this page:** 16
**Page:** 1

<div class="post-metadata">

### Author: ![lukee2ni6](https://developer.download.nvidia.com/images/forums/profile-default-devtalk-84.png) [@lukee2ni6](https://forums.developer.nvidia.com/u/lukee2ni6)
#### Post date: [February 27, 2019, 6:42pm UTC](https://forums.developer.nvidia.com/t/cuda-10-1-tensorflow-1-13/70940/1 "2019-02-27T18:42:20Z")

</div>

The recently released TF 1.13 is built against CUDA 10 (.0 presumably?)

For a new build with TF 1.13 as the target application, is it recomennded to use CUDA 10.0 or 10.1?

---

<div class="post-metadata">

### Author: ![bernard.at.spark](https://developer.download.nvidia.com/images/forums/profile-default-devtalk-84.png) [@bernard.at.spark](https://forums.developer.nvidia.com/u/bernard.at.spark)
#### Post date: [March 2, 2019, 12:20am UTC](https://forums.developer.nvidia.com/t/cuda-10-1-tensorflow-1-13/70940/2 "2019-03-02T00:20:26Z")

</div>

10.0 for sure. tensorflow seems to be a fragile piece of software, everytime there is a cuda update it breaks. At this point apparently only the latest TF 1.13.1 released less than a week ago compiles with cuda 10.0, doubt that any tensorflow in release would work with 10.1, besides cuda 10.1 seems to be broken for other reason, see other threads.

---

<div class="post-metadata">

### Author: ![ddsawat](https://developer.download.nvidia.com/images/forums/profile-default-devtalk-84.png) [@ddsawat](https://forums.developer.nvidia.com/u/ddsawat)
#### Post date: [March 5, 2019, 10:34am UTC](https://forums.developer.nvidia.com/t/cuda-10-1-tensorflow-1-13/70940/3 "2019-03-05T10:34:05Z")

</div>

@bernard Yes Tensorflow 1.13.1 dosen’t work with CUDA 10.1. I am in process to compile 1.13.1 with CUDA 10.0. Hope it works

---

<div class="post-metadata">

### Author: ![ddsawat](https://developer.download.nvidia.com/images/forums/profile-default-devtalk-84.png) [@ddsawat](https://forums.developer.nvidia.com/u/ddsawat)
#### Post date: [March 5, 2019, 11:43am UTC](https://forums.developer.nvidia.com/t/cuda-10-1-tensorflow-1-13/70940/4 "2019-03-05T11:43:44Z")

</div>

@bernard Tensorflow 1.13.1 worked with CUDA 10.0, cudnn 7.3. Tested successfully

---

<div class="post-metadata">

### Author: ![lepersorium](https://sea2.discourse-cdn.com/nvidia/user_avatar/forums.developer.nvidia.com/lepersorium/32/11433_2.png) [@lepersorium](https://forums.developer.nvidia.com/u/lepersorium)
#### Post date: [March 7, 2019, 8:00am UTC](https://forums.developer.nvidia.com/t/cuda-10-1-tensorflow-1-13/70940/5 "2019-03-07T08:00:26Z")

</div>

I dont recommend tf 1.13 [https://github.com/tensorflow/tensorflow/issues/26218](https://github.com/tensorflow/tensorflow/issues/26218)

---

<div class="post-metadata">

### Author: ![a.khan](https://developer.download.nvidia.com/images/forums/profile-default-devtalk-84.png) [@a.khan](https://forums.developer.nvidia.com/u/a.khan)
#### Post date: [March 7, 2019, 8:31am UTC](https://forums.developer.nvidia.com/t/cuda-10-1-tensorflow-1-13/70940/6 "2019-03-07T08:31:26Z")

</div>

Yes Tensorflow 1.13.1 dosen’t work with CUDA 10.1.

---

<div class="post-metadata">

### Author: ![a.khan](https://developer.download.nvidia.com/images/forums/profile-default-devtalk-84.png) [@a.khan](https://forums.developer.nvidia.com/u/a.khan)
#### Post date: [March 7, 2019, 8:31am UTC](https://forums.developer.nvidia.com/t/cuda-10-1-tensorflow-1-13/70940/7 "2019-03-07T08:31:48Z")

</div>

How to down-grade CUDA 10.1 to CUDA 10.0?

---

<div class="post-metadata">

### Author: ![ddsawat](https://developer.download.nvidia.com/images/forums/profile-default-devtalk-84.png) [@ddsawat](https://forums.developer.nvidia.com/u/ddsawat)
#### Post date: [March 7, 2019, 4:00pm UTC](https://forums.developer.nvidia.com/t/cuda-10-1-tensorflow-1-13/70940/8 "2019-03-07T16:00:38Z")

</div>

Use the script generated by CUDA installer as

```auto
sudo /usr/local/cuda-X.Y/bin/uninstall_cuda_X.Y.p

```

Replace XY with your version of CUDA.

---

<div class="post-metadata">

### Author: ![bernard.at.spark](https://developer.download.nvidia.com/images/forums/profile-default-devtalk-84.png) [@bernard.at.spark](https://forums.developer.nvidia.com/u/bernard.at.spark)
#### Post date: [March 7, 2019, 6:14pm UTC](https://forums.developer.nvidia.com/t/cuda-10-1-tensorflow-1-13/70940/9 "2019-03-07T18:14:23Z")

</div>

@a.khan

```auto
How to down-grade CUDA 10.1 to CUDA 10.0?

```

What is your OS and how did you install cuda?

If you are on Ubuntu and installed cuda with Nvidia’s ubuntu repo then it is really simple

```auto
sudo apt uninstall cuda
sudo apt autoremove
sudo apt update
sudo apt install cuda-10-0

```

Here “cuda” and “cuda-10-0” are metapackages cuda always update to the latest, in this case cuda10.1. Whereas cuda-version-number doesn’t update (others available are cuda-9-2, cuda-9-1 etc, you can easily search for them if you have synaptic, or use the command “apt-cache search cuda”)

If you are on Windows or Mac then I have no idea.

---

<div class="post-metadata">

### Author: ![Andrey1984](https://sea2.discourse-cdn.com/nvidia/user_avatar/forums.developer.nvidia.com/andrey1984/32/14130_2.png) [@Andrey1984](https://forums.developer.nvidia.com/u/Andrey1984)
#### Post date: [March 16, 2019, 11:32pm UTC](https://forums.developer.nvidia.com/t/cuda-10-1-tensorflow-1-13/70940/10 "2019-03-16T23:32:15Z")

</div>

From the [documentation](https://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#removing-cuda-tk-and-driver)

```auto
Ubuntu
To remove CUDA Toolkit:
$ sudo apt-get --purge remove "*cublas*" "cuda*"
To remove NVIDIA Drivers:
$ sudo apt-get --purge remove "*nvidia*"

```

Though it defines different uninstall for runfile and debfile installations at some other paragraph.  
However, once removed - reboot and install cuda 10.0  
it can be downloaded from [CUDA Toolkit 10.0 Archive | NVIDIA Developer](https://developer.nvidia.com/cuda-10.0-download-archive)

---

<div class="post-metadata">

### Author: ![bernard.at.spark](https://developer.download.nvidia.com/images/forums/profile-default-devtalk-84.png) [@bernard.at.spark](https://forums.developer.nvidia.com/u/bernard.at.spark)
#### Post date: [March 17, 2019, 12:34am UTC](https://forums.developer.nvidia.com/t/cuda-10-1-tensorflow-1-13/70940/11 "2019-03-17T00:34:21Z")

</div>

```auto
Though it defines different uninstall for runfile and debfile installations at some other paragraph.
However, once removed - reboot and install cuda 10.0

```

You don’t need to download from archive again. Nvidia’s deb repository has multiple cuda

All you have to do, instead of installing cuda

```auto
sudo apt install cuda-10-0

```

---

<div class="post-metadata">

### Author: ![gbolcer](https://developer.download.nvidia.com/images/forums/profile-default-devtalk-84.png) [@gbolcer](https://forums.developer.nvidia.com/u/gbolcer)
#### Post date: [April 25, 2019, 5:10pm UTC](https://forums.developer.nvidia.com/t/cuda-10-1-tensorflow-1-13/70940/12 "2019-04-25T17:10:25Z")

</div>

They seem to have taken the cuda-10-0 target out of the latest repo.

cuda-repo-ubuntu1810/unknown,now 10.1.105-1 amd64 [installed]  
cuda repository configuration files

greg@salt:~$ sudo apt list cuda  
Listing… Done  
cuda/unknown,now 10.1.105-1 amd64 [installed]

---

<div class="post-metadata">

### Author: ![marlon\_evangelista](https://developer.download.nvidia.com/images/forums/profile-default-devtalk-84.png) [@marlon\_evangelista](https://forums.developer.nvidia.com/u/marlon_evangelista)
#### Post date: [May 13, 2019, 5:35pm UTC](https://forums.developer.nvidia.com/t/cuda-10-1-tensorflow-1-13/70940/13 "2019-05-13T17:35:48Z")

</div>

> [@](#):
>
> They seem to have taken the cuda-10-0 target out of the latest repo.
> 
> cuda-repo-ubuntu1810/unknown,now 10.1.105-1 amd64 [installed]  
> cuda repository configuration files
> 
> greg@salt:~$ sudo apt list cuda  
> Listing… Done  
> cuda/unknown,now 10.1.105-1 amd64 [installed]

The network .deb file from the 10.0 archive ([https://developer.nvidia.com/cuda-10.0-download-archive?target\_os=Linux&target\_arch=x86\_64&target\_distro=Ubuntu&target\_version=1804&target\_type=debnetwork](https://developer.nvidia.com/cuda-10.0-download-archive?target_os=Linux&target_arch=x86_64&target_distro=Ubuntu&target_version=1804&target_type=debnetwork)) should get you what you need. I just tried it, and I was able to run sudo apt-get install cuda-10-0 after uninstalling cuda 10.1.

---

<div class="post-metadata">

### Author: ![lukee2ni6](https://developer.download.nvidia.com/images/forums/profile-default-devtalk-84.png) [@lukee2ni6](https://forums.developer.nvidia.com/u/lukee2ni6)
#### Post date: [June 19, 2019, 9:00am UTC](https://forums.developer.nvidia.com/t/cuda-10-1-tensorflow-1-13/70940/14 "2019-06-19T09:00:02Z")

</div>

TF 1.1.4 has been released!

Anybody know which Cuda it has been built against? No mention in the release notes.

> **[Release TensorFlow 1.14.0 · tensorflow/tensorflow](https://github.com/tensorflow/tensorflow/releases/tag/v1.14.0)**
>
> Release 1.14.0
> Major Features and Improvements
> 
> This is the first 1.x release containing the compat.v2 module. This module is required to allow libraries to publish code which works in both 1.x and...

---

<div class="post-metadata">

### Author: ![aaron\_bond](https://sea2.discourse-cdn.com/nvidia/user_avatar/forums.developer.nvidia.com/aaron_bond/32/38555_2.png) [@aaron\_bond](https://forums.developer.nvidia.com/u/aaron_bond)
#### Post date: [July 12, 2019, 4:56am UTC](https://forums.developer.nvidia.com/t/cuda-10-1-tensorflow-1-13/70940/15 "2019-07-12T04:56:01Z")

</div>

> [@](#):
>
> TF 1.1.4 has been released!
> 
> Anybody know which Cuda it has been built against? No mention in the release notes.  
> [Release TensorFlow 1.14.0 · tensorflow/tensorflow · GitHub](https://github.com/tensorflow/tensorflow/releases/tag/v1.14.0)

I would also like to know.

---

<div class="post-metadata">

### Author: ![monocongo](https://sea2.discourse-cdn.com/nvidia/user_avatar/forums.developer.nvidia.com/monocongo/32/11802_2.png) [@monocongo](https://forums.developer.nvidia.com/u/monocongo)
#### Post date: [July 25, 2019, 1:18pm UTC](https://forums.developer.nvidia.com/t/cuda-10-1-tensorflow-1-13/70940/16 "2019-07-25T13:18:46Z")

</div>

TensorFlow 1.14 only works with CUDA 10.0, as I recently [learned the hard way](https://stackoverflow.com/questions/57192931/tensorflow-gpu-not-being-utilized-cuda-is-installed-but-the-gpu-appears-to-not).

This shows the dependency on CUDA version 10.0 as well: [使用 pip 安装 TensorFlow](https://www.tensorflow.org/install/gpu)
