Closed ghost closed 8 years ago
@bscully Can you tell me how you installed arrayfire python wrapper?
pip install arrayfire
Should I try another way?
On Apr 26, 2016, at 11:57 PM, Pavan Yalamanchili notifications@github.com<mailto:notifications@github.com> wrote:
@bscullyhttps://github.com/bscully Can you tell me how you installed arrayfire python wrapper?
You are receiving this because you were mentioned. Reply to this email directly or view it on GitHubhttps://github.com/arrayfire/arrayfire-python/issues/77#issuecomment-214963886
@bscully can you print out the output of isinstance(data.ctypes.data, int)
isinstance(data.ctypes.data, int) Out[58]: False
From: Pavan Yalamanchili [mailto:notifications@github.com] Sent: Wednesday, April 27, 2016 8:50 AM To: arrayfire/arrayfire-python arrayfire-python@noreply.github.com Cc: Ben Scully bscully@alphaparity.com; Mention mention@noreply.github.com Subject: Re: [arrayfire/arrayfire-python] NumPy Array to AF Array (#77)
@bscullyhttps://github.com/bscully can you print out the output of isinstance(data.ctypes.data, int)
— You are receiving this because you were mentioned. Reply to this email directly or view it on GitHubhttps://github.com/arrayfire/arrayfire-python/issues/77#issuecomment-215071475
@bscully output of type(data.ctypes.data)
please.
Can you also let me know if you are using a different python distribution ?
@ZacDiggum I assume you are also on windows but are not seeing this issue?
type(data.ctypes.data) Out[61]: long
I'm running everything on windows and it's the standalone version, ie not anaconda or enthought etc
Update:
The calculation examples work but not the graphics.
convolve.py
full 2D convolution time: 1.67066 ms
separable 2D convolution time: 1.58442 ms
intro.py (only kept the first several lines) ---- Intro to ArrayFire using signed(s32) arrays ---- ---- Sub referencing and sub assignment
ArrayFire v3.3.2 (CUDA, 64-bit Windows, build f65dd97) Platform: CUDA Toolkit 7.5, Driver: CUDA Driver Version: 7050 [0] GeForce GT 610, 2048 MB, CUDA Compute 2.1 ArrayFire v3.3.2 (CUDA, 64-bit Windows, build f65dd97) Platform: CUDA Toolkit 7.5, Driver: CUDA Driver Version: 7050 [0] GeForce GT 610, 2048 MB, CUDA Compute 2.1 A [3 3 1 1] 1 -1 4 2 2 2 4 0 3
surface.py
Traceback (most recent call last):
File "C:\Python27\lib\site-packages\IPython\core\interactiveshell.py", line 3066, in run_code
exec(code_obj, self.user_global_ns, self.user_ns)
File "
more info: I do have a 2nd distribution with Anaconda. Switching between the 2 is easy which is why I said the standard python earlier. I started using PyCuda and could only get that to work with Anaconda.
Might be related to this error..
looks like the original error only happens on windows + python 2.7. I will try to patch this in and make a release today.
Fixed in master and latest release.
@bscully This does not fix the graphics issues. If you are still having those errors, please create a new issue for it.
Uninstalled, ran pip install and now it works!! Interestingly when i tried compiling the master branch which looks to be version 3.4, it did not work. Does that make sense?
Either way, the current pip install is working great - thanks so much!
And as for the graphics error, will log another case.
@bscully If you are using the command from the README (or use git clone), it uses develop
branch not master
. I patched this only in master
so far. I will merge it into devel
soon.
@bscully FYI, I just merged the fixes into devel branch as well.
data=np.random.random(size=(100,100)) af.np_to_af_array(data)
gets error: af.np_to_af_array(data) File "C:\Python27\lib\site-packages\arrayfire\interop.py", line 53, in np_to_af_array res = Array(in_ptr, shape, in_dtype) File "C:\Python27\lib\site-packages\arrayfire\array.py", line 428, in init raise TypeError("src is an object of unsupported class") TypeError: src is an object of unsupported class
My Environment: Python27 64 bit Windows 7 64 AF version: 3.3.2 or sys.version_info(major=2, minor=7, micro=10, releaselevel='final', serial=0) CUDA kit 7.5
More background: I just installed the package so likely a setup issue on my side... The examples work for me, both cpu & cuda.
Any ideas?