Hi,
Playing with Mathematica 12 and ImageRestyle and the system generates an error message:
The neural network is unavailabe. With MM11.3 it has always worked.
Stacktrace shows:
ImageRestyle::nnlibunavail
Running Windows 10
As a comment, before the imagerestyle started it first downloaded the GPU libraries. That was 900MB!
After the download Mathematica showed a message that a restart was needed. After the restart the message above was generated.
To clarify, are you trying to use ImageRestyle with TargetDevice->"GPU"? If so, can you make sure you have the latest drivers for your GPU installed? M12.0 requires newer drivers than M11.3.0.
Hi Stefan,
That is true. I want/need to use the GPU option!
I updated the NVidea driver to the latest hardware vendor supported version and that didn’t help.
Next I used the NVidea website to locate the latest and installed these.
I updated the CUDA paclets but it still doesn’t work.
The error message has changed though.
Any next suggestions or information needed?
CUDAResourcesInstall[Update -> True]
Out[2]= {
PacletManager`Paclet[
“Name” → “CUDAResources”, “Version” → “12.0.359”,
“MathematicaVersion” → “12.0”,
“Description” → “{ToolkitVersion → v10.1, MinimumDriver → 290}”,
“SystemID” → {“Windows-x86-64”}, “Qualifier” → “Win64”,
“Extensions” → {{
“Resource”,
“Resources” → {
“CUDAToolkit”, “ExampleData”, “LibraryResources”}}},
“Location” → “C:\Users\lvveen\AppData\Roaming\Mathematica\
Paclets\Repository\CUDAResources-Win64-12.0.359”]}
CUDADriverVersion[]
Out[4]= “425.31”
CUDAInformation[]
Out[5]= {1 → {“Name” → “Quadro M1000M”, “Clock Rate” → 1071500,
“Compute Capabilities” → 5., “GPU Overlap” → 1,
“Maximum Block Dimensions” → {1024, 1024, 64},
“Maximum Grid Dimensions” → {2147483647, 65535, 65535},
“Maximum Threads Per Block” → 1024,
“Maximum Shared Memory Per Block” → 49152,
“Total Constant Memory” → 65536, “Warp Size” → 32,
“Maximum Pitch” → 2147483647,
“Maximum Registers Per Block” → 65536, “Texture Alignment” → 512,
“Multiprocessor Count” → 4, “Core Count” → 128,
“Execution Timeout” → 1, “Integrated” → False,
“Can Map Host Memory” → True, “Compute Mode” → “Default”,
“Texture1D Width” → 65536, “Texture2D Width” → 65536,
“Texture2D Height” → 65536, “Texture3D Width” → 4096,
“Texture3D Height” → 4096, “Texture3D Depth” → 4096,
“Texture2D Array Width” → 16384,
“Texture2D Array Height” → 16384,
“Texture2D Array Slices” → 2048, “Surface Alignment” → 512,
“Concurrent Kernels” → True, “ECC Enabled” → False,
“TCC Enabled” → False, “Total Memory” → 2147483648}}
Hi Stefan,
That is true. I want/need to use the GPU option!
I updated the NVidea driver to the latest hardware vendor supported version and that didn’t help.
Next I used the NVidea website to locate the latest and installed these.
I updated the CUDA paclets but it still doesn’t work.
The error message has changed though.
Any next suggestions or information needed?
CUDAResourcesInstall[Update -> True]
Out[2]= {
PacletManager`Paclet[
“Name” → “CUDAResources”, “Version” → “12.0.359”,
“MathematicaVersion” → “12.0”,
“Description” → “{ToolkitVersion → v10.1, MinimumDriver → 290}”,
“SystemID” → {“Windows-x86-64”}, “Qualifier” → “Win64”,
“Extensions” → {{
“Resource”,
“Resources” → {
“CUDAToolkit”, “ExampleData”, “LibraryResources”}}},
“Location” → “C:\Users\lvveen\AppData\Roaming\Mathematica\
Paclets\Repository\CUDAResources-Win64-12.0.359”]}
CUDADriverVersion[]
Out[4]= “425.31”
CUDAInformation[]
Out[5]= {1 → {“Name” → “Quadro M1000M”, “Clock Rate” → 1071500,
“Compute Capabilities” → 5., “GPU Overlap” → 1,
“Maximum Block Dimensions” → {1024, 1024, 64},
“Maximum Grid Dimensions” → {2147483647, 65535, 65535},
“Maximum Threads Per Block” → 1024,
“Maximum Shared Memory Per Block” → 49152,
“Total Constant Memory” → 65536, “Warp Size” → 32,
“Maximum Pitch” → 2147483647,
“Maximum Registers Per Block” → 65536, “Texture Alignment” → 512,
“Multiprocessor Count” → 4, “Core Count” → 128,
“Execution Timeout” → 1, “Integrated” → False,
“Can Map Host Memory” → True, “Compute Mode” → “Default”,
“Texture1D Width” → 65536, “Texture2D Width” → 65536,
“Texture2D Height” → 65536, “Texture3D Width” → 4096,
“Texture3D Height” → 4096, “Texture3D Depth” → 4096,
“Texture2D Array Width” → 16384,
“Texture2D Array Height” → 16384,
“Texture2D Array Slices” → 2048, “Surface Alignment” → 512,
“Concurrent Kernels” → True, “ECC Enabled” → False,
“TCC Enabled” → False, “Total Memory” → 2147483648}}
So at the end I got it working.
I removed all 11.x versions an deleted the wolfram and mathematica directories in the
C:\Users\lvveen\AppData\Roaming directory.
After re-install I received an installation error:
I decided to ignore it and move on..
I went to the ImageRestyle documentation and started the example
Then it started to reload the GPU libraries
After this a restart was asked:
After the restart back on the page and a download was started again (this did not happen with previous installs)
Now it works for ImageRestyle:
and also for NetTrain:
I’m still a bit worried about the install error message but it seems to have no effect. Would love to hear if this is indeed save to ignore or not.
After the restart back on the page and a download was started again (this did not happen with previous installs)
There it is downloading the neural network that ImageRestyle uses. This should only happen once.
I’ve never seen that kind of installation warning before, I’ll ask around if anyone knows what it means…
After re-install I received an installation error:
Regarding this particular error this it may indicate a permission/firewall problem. I’d suggest rebooting first and trying the installation again otherwise make sure you have admin privileges and/or any firewall settings aren’t blocking this.
Hi Carlos, Oke thanks.
Trying the installation again is not working and I did the install in Admin mode.
So might be a firewall issue. Not looking forward to re-install again.. Any idea if this might cause problems? The installation gave the possibility to ignore it so I assume it not being a critical thing to solve. Thx!
This particular key probably won’t cause a problem but if it does feel free to reach out once again. However, it is due to some security/firewall feature enabled on your machine that is most likely the culprit.
