urutu / Urutu

A Python based parallel programming library for GPUs
Apache License 2.0
7 stars 1 forks source link

Errors #23

Open fermicro opened 7 years ago

fermicro commented 7 years ago

I'm having the following errors trying to run the sample code in your git main page:

Python 3 (done 2to3):

In Urutu\cl__init__.py line 422 if sentence.split('\t'))>1: ##TypeError: '>' not supported between instances of 'list' and 'int'

if bypassed, will also give the same error below.

Python 2:

Traceback (most recent call last): File "D:\tcc\urutuWinPython-64bit-3.6.0.1Qt5\sample.py.bak", line 35, in print "Running on OpenCL.. \n", divmul([100,1,1], [1,1,1], a, b, c, d) File "D:\tcc\WinPython-64bit-2.7.10.3\python-2.7.10.amd64\lib\Urutu__init__.py", line 30, in inner return cl.execute() File "D:\tcc\WinPython-64bit-2.7.10.3\python-2.7.10.amd64\lib\Urutu\cl__init__.py", line 612, in execute tmp.start(self.args,self.arg_nam, self.arg_mask) File "D:\tcc\WinPython-64bit-2.7.10.3\python-2.7.10.amd64\lib\Urutu\cl\execl.py", line 61, in start self.cl_args.append(cl.Buffer(self.ctx,self.mf.READ_WRITE | self.mf.COPY_HOST_PTR, hostbuf = self.args[i])) NameError: global name 'cl' is not defined

Can you help?

aditya4d commented 7 years ago

Can you try changing this line https://github.com/urutu/Urutu/blob/master/Urutu/cl/__init__.py#L422

to

if len(sentence.split('\t'))>1:
fermicro commented 7 years ago

That will work to bypass the first error. So the second error will show.

On Mar 2, 2017 11:43, "Aditya Atluri" notifications@github.com wrote:

Can you try changing this line https://github.com/urutu/ Urutu/blob/master/Urutu/cl/init.py#L422

to

if len(sentence.split('\t'))>1:

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/urutu/Urutu/issues/23#issuecomment-283671505, or mute the thread https://github.com/notifications/unsubscribe-auth/AQl9WmCj8gwfJ7mSJ14CSWDy0Dlcgbc9ks5rhtV1gaJpZM4MQfvC .

aditya4d commented 7 years ago

I haven't tried opencl in a while and I didn't actively updated its backend to work efficiently as CUDA. I'll try to reproduce the error.