Can't install mxnet cuda on Windows Server 2016!

[Kandarp Rastey]

Description

OS = Windows Server 2016 (x64)
Python system version = 3.7.13 (default, Mar 28 2022, 08:03:21) [MSC v.1916 64 bit (AMD64)]
Python platform version = 3.7.13
PIP version = pip-22.2.2
CUDA = 11.7
Cuda compilation tools, release 11.7, V11.7.99
Build cuda_11.7.r11.7/compiler.31442593_0

Tried to use GPU to run inference giving context like:
ctx = mx.gpu()
net = gcv.model_zoo.get_model( name=*, pretrained=False, classes=*, ctx=ctx )
But it returned this error.

Error Message

Traceback (most recent call last):
File "main.py", line 24, in
net_object_detector = models.initialize_object_detector()
File "*\src\models.py", line 43, in initialize_object_detector
ctx=ctx
File "C:\Users*\Anaconda3\envs*\lib\site-packages\gluoncv\model_zoo\model_zoo.py", line 413, in get_model
net = _modelsname
File "C:\Users*\Anaconda3\envs*\lib\site-packages\gluoncv\model_zoo\ssd\presets.py", line 442, in ssd_512_resnet50_v1_custom
pretrained_base=pretrained_base, **kwargs)
File "C:\Users*\Anaconda3\envs*\lib\site-packages\gluoncv\model_zoo\ssd\ssd.py", line 417, in get_ssd
minimal_opset=pretrained, anchor_generator=anchor_generator, **kwargs)
File "C:\Users*\Anaconda3\envs*\lib\site-packages\gluoncv\model_zoo\ssd\ssd.py", line 150, in init
norm_layer=norm_layer, norm_kwargs=norm_kwargs, root=root)
File "C:\Users*\Anaconda3\envs*\lib\site-packages\gluoncv\nn\feature.py", line 135, in init
network, outputs, inputs, pretrained, ctx, **kwargs)
File "C:\Users*\Anaconda3\envs*\lib\site-packages\gluoncv\nn\feature.py", line 55, in _parse_network
network = get_model(network, pretrained=pretrained, ctx=ctx, **kwargs)
File "C:\Users*\Anaconda3\envs*\lib\site-packages\gluoncv\model_zoo\model_zoo.py", line 413, in get_model
net = _modelsname
File "C:\Users*\Anaconda3\envs*\lib\site-packages\gluoncv\model_zoo\resnet.py", line 635, in resnet50_v1
return get_resnet(1, 50, use_se=False, **kwargs)
File "C:\Users*\Anaconda3\envs*\lib\site-packages\gluoncv\model_zoo\resnet.py", line 560, in get_resnet
tag=pretrained, root=root), ctx=ctx)
File "C:\Users*\Anaconda3\envs*\lib\site-packages\mxnet\gluon\block.py", line 556, in load_parameters
params[name]._load_init(loaded[name], ctx, cast_dtype=cast_dtype, dtype_source=dtype_source)
File "C:\Users*\Anaconda3\envs*\lib\site-packages\mxnet\gluon\parameter.py", line 316, in _load_init
self._init_impl(data, ctx)
File "C:\Users*\Anaconda3\envs*\lib\site-packages\mxnet\gluon\parameter.py", line 365, in _init_impl
self._data = [data.copyto(ctx) for ctx in self._ctx_list]
File "C:\Users*\Anaconda3\envs*\lib\site-packages\mxnet\gluon\parameter.py", line 365, in
self._data = [data.copyto(ctx) for ctx in self._ctx_list]
File "C:\Users*\Anaconda3\envs*\lib\site-packages\mxnet\ndarray\ndarray.py", line 2666, in copyto
return _internal._copyto(self, out=hret)
File "", line 27, in _copyto
File "C:\Users*\Anaconda3\envs*\lib\site-packages\mxnet_ctypes\ndarray.py", line 91, in _imperative_invoke
ctypes.byref(out_stypes)))
File "C:\Users*\Anaconda3\envs*\lib\site-packages\mxnet\base.py", line 246, in check_call
raise get_last_ffi_error()
mxnet.base.MXNetError: Traceback (most recent call last):
File "C:\Jenkins\workspace\mxnet-tag\mxnet\src\ndarray\ndarray.cc", line 1298
MXNetError: GPU is not enabled

• So I tried installing mxnet-cu112 using pip but that returned this error:
  ERROR: Could not find a version that satisfies the requirement mxnet-cu112 (from versions: none)
ERROR: No matching distribution found for mxnet-cu112

• So instead I downloaded the mxnet_cu112-1.9.0-py3-none-manylinux2014_x86_64.whl from https://dist.mxnet.io/python/cu112 and tried installing it using pip but that returned this error:
  ERROR: mxnet_cu112-1.9.0-py3-none-manylinux2014_x86_64.whl is not a supported wheel on this platform.
  which is strange.

• Help me install with Cuda enabled mxnet so I can use GPU instead of CPU.

[github-actions]

Welcome to Apache MXNet (incubating)! We are on a mission to democratize AI, and we are glad that you are contributing to it by opening this issue.
Please make sure to include all the relevant context, and one of the @apache/mxnet-committers will be here shortly.
If you are interested in contributing to our project, let us know! Also, be sure to check out our guide on contributing to MXNet and our development guides wiki.

[relaxli00]

As far as I know, there is no binary wheel for windows released after 1.7.0. You'll have to build from source. I tried to build 1.9.1 for Windows 10 with MKL BLAS. Not sure if it's helpful, my steps can be found here https://medium.com/@relaxli/build-mxnet-1-9-1-with-cuda-11-7-and-mkl-blas-using-visual-studio-2019-on-windows-10-a9ebecb1941b .

[quxi0001]

Hi, have you solve this problrm? I got the same problem and don't know how to fix it.