PyOpenCL June 2012

pyopencl@tiker.net

10 participants
10 discussions

by Gregor Thalhammer

Dear Andreas, I am currently working on a cython based wrapper for the OpenCL FFT library from AMD: https://github.com/geggo/gpyfft For this I need to create a pyopencl Event instance from a cl_event returned by the library. I attached a patch against recent pyopencl that adds this possibility, similar to the from_cl_mem_as_int() method of the MemoryObject class. Could you please add this to pyopencl. Thanks for your help Gregor

7 years, 8 months

EuroPython anyone?

by Tomasz Rybak

Hello list. Will any of you attend EuroPython in Florence, Italy, starting Monday 2012-07-02? I'll be there all week and will be glad to meet other PyOpenCL users. Regards. -- Tomasz Rybak <tomasz.rybak(a)post.pl> GPG/PGP key ID: 2AD5 9860 Fingerprint A481 824E 7DD3 9C0E C40A 488E C654 FB33 2AD5 9860 http://member.acm.org/~tomaszrybak

8 years, 9 months

Problems with complex GPU arrays

by Andrew Miller

Hello, I've found some unexpected errors while working with PyOpenCL, specifically when using one-dimensional, complex arrays. pycuda.array.dot(a,b) doesn't work with a complex vector. pycuda.array.sum(a) also gives an error when a is complex Are these features that will be added in the future, or is there just not full support for complex vectors? Thanks! -Andrew Miller

8 years, 9 months

Working on PyOpenCL compatible wrapper for OpenVideo

by Keith Brafford

In case anyone else is interested in such a thing, I have started working on a Python wrapper for AMD's OpenVideo library. Here's a page that talks about it: http://developer.amd.com/sdks/AMDAPPSDK/samples/Pages/default.aspx Here's a sample project that supposedly shows how to use OpenVideo with GL-CL interop: http://developer.amd.com/downloads/OVDecodeRender.zip And here's the repo for my project: https://github.com/kbrafford/pyopenvideo --Keith Brafford

8 years, 9 months

Invitation to connect on LinkedIn

by Pedro Marcal

LinkedIn ------------ I'd like to add you to my professional network on LinkedIn. - Pedro Pedro Marcal CEO at MPACT Greater San Diego Area Confirm that you know Pedro Marcal: https://www.linkedin.com/e/ld6ujy-h3pvzmbt-6y/isd/7581759114/S5MdU_8H/?hs=f… -- You are receiving Invitation to Connect emails. Click to unsubscribe: http://www.linkedin.com/e/ld6ujy-h3pvzmbt-6y/uNhYjjG4yLjqIh4VayhPmihv1lWKZz… (c) 2012 LinkedIn Corporation. 2029 Stierlin Ct, Mountain View, CA 94043, USA.

8 years, 9 months

Re: [PyOpenCL] In-place Elementwise Array Multiplication

by Andreas Kloeckner

Hi Andrew, Andrew Miller <ajm09c(a)acu.edu> writes: > I have run into a question while using 1d arrays in OpenCL. I am able to > multiply a 1d array*=scalar. When I try to multiply a 1d array by another > 1d array, it works fine as: newarray=a*b but when I try a*=b, I run into > problems. Should be fixed in git, thanks for the report. Andreas

8 years, 10 months

Binary Array ===> Vectorization

by Timo Fleckenstein

8 years, 10 months

Binary Array ===> Vectorization

by Timo Burggraf

Hello Mr. Klöckner, I work with your PyOpenCL Package for my own PhD at TU-Darmstadt in the field of applied mathematics. It is great! Thank you for your effords! Right know I try to handle a problem, and my google-Research did not work out. I do some calculations of Voronoi-Cells (nearest neighbor-problem). Therefore, I write into my Result array, which is quite big (Number of Points > 250.000). I only want to store True/False to keep the information of closest neighbors. (i.e. Pnt 12 is Neighbour of Pnt 13 => 12/13 = 1) I compressed this information to a uint8 linear array (Values 0-255) and I do bit manipulations to store 8 Neighbor Informations. Furthermore the array grows with the Gaussian sum: 0 1 2 3 0 1 0 2 0 1 3 0 1 0 Source: nachbar = zeros(array_size, dtype=uint8) nachbar_buf = cl.Buffer(ctx, mf.READ_WRITE | mf.USE_HOST_PTR, hostbuf=nachbar) int index = (cur_compare_pnt-1)*cur_compare_pnt/2 + cur_pnt; int main_index = index / 8; int sub_index = index % 8; nachbar_buf[main_index] |= 1<<sub_index; I am not sure, if this was understandable I hope so. My problem is, that this process is not vectorizable, since I write with different Point Tupels into the same uint8 Entry. My question is: Is there a way to use a bool-linear- array in OpenCL? Or is there a better way at all? Thank you very much. I started with OpenCL at all one year ago and your work help a lot. Cheers Timo

8 years, 10 months

In-place Elementwise Array Multiplication

by Andrew Miller

Hello, I have run into a question while using 1d arrays in OpenCL. I am able to multiply a 1d array*=scalar. When I try to multiply a 1d array by another 1d array, it works fine as: newarray=a*b but when I try a*=b, I run into problems. Here is the output: a_gpu*=b_gpu File "/home/andrewmiller/dev/root/pyopencl/lib/python2.6/site-packages/pyopencl-2011.2-py2.6-linux-x86_64.egg/pyopencl/array.py", line 631, in __imul__ self._axpbz(self, scalar, self, self.dtype.type(0)) File "/home/andrewmiller/dev/root/pyopencl/lib/python2.6/site-packages/pyopencl-2011.2-py2.6-linux-x86_64.egg/pyopencl/array.py", line 155, in kernel_runner knl = kernel_getter(*args) File "/home/andrewmiller/dev/root/pyopencl/lib/python2.6/site-packages/pyopencl-2011.2-py2.6-linux-x86_64.egg/pyopencl/array.py", line 407, in _axpbz a.dtype, x.dtype, b.dtype, out.dtype) File "<string>", line 2, in get_axpbz_kernel File "/home/andrewmiller/dev/root/pyopencl/lib/python2.6/site-packages/pyopencl-2011.2-py2.6-linux-x86_64.egg/pyopencl/tools.py", line 83, in first_arg_dependent_memoize result = func(cl_object, *args) File "/home/andrewmiller/dev/root/pyopencl/lib/python2.6/site-packages/pyopencl-2011.2-py2.6-linux-x86_64.egg/pyopencl/elementwise.py", line 417, in get_axpbz_kernel "tp_a": dtype_to_ctype(dtype_a), File "/home/andrewmiller/dev/root/pyopencl/lib/python2.6/site-packages/pyopencl-2011.2-py2.6-linux-x86_64.egg/pyopencl/compyte/dtypes.py", line 114, in dtype_to_ctype raise ValueError, "unable to map dtype '%s'" % dtype ValueError: unable to map dtype 'object' a_gpu and b_gpu are both pyopencl arrays, and they are both 1d, 2-element, float32 Is there a way to do this with pyopencl, and I'm just going about it wrong? Thanks for the help! -Andrew Miller

8 years, 10 months

strange bus error

by Massimo DiPierro

I am getting a strange bus error. It is related to opencl or numpy but it does not manifest itself inside opencl code. My code looks like this: …. os.system("top -l 1 | awk '/PhysMem:/ {print $10}'") pry.run(….) # run my kernel (constant memory) os.system("top -l 1 | awk '/PhysMem:/ {print $10}'") print 'done' The output is …. 1430M 1413M Bus error: 10 It does not print 'done'. It is 100% reproducible. It hits when the total size of my numpy arrays is 80Mbytes. I tried to rearrange the python code in many ways and I always get the bus error in python code executed after a call to pyopencl when the size hits 80M. I am running on OSX Lion with 8GB Ram. Any idea? Process: Python [20697] Path: /System/Library/Frameworks/Python.framework/Versions/2.7/Resources/Python.app/Contents/MacOS/Python Identifier: Python Version: ??? (???) Code Type: X86-64 (Native) Parent Process: bash [163] PlugIn Path: cl_kernels PlugIn Identifier: cl_kernels PlugIn Version: ??? (???) Date/Time: 2012-06-03 23:47:35.913 -0500 OS Version: Mac OS X 10.7.4 (11E53) Report Version: 9 Interval Since Last Report: 125828 sec Crashes Since Last Report: 18 Per-App Crashes Since Last Report: 18 Anonymous UUID: 6B52EA49-91A6-491B-844A-18883AD108BA Crashed Thread: 3 Dispatch queue: com.apple.root.low-overcommit-priority Exception Type: EXC_BAD_ACCESS (SIGBUS) Exception Codes: KERN_PROTECTION_FAILURE at 0x0000000105563b68 VM Regions Near 0x105563b68: Stack 00000001054e1000-0000000105563000 [ 520K] rw-/rwx SM=COW thread 2 --> STACK GUARD 0000000105563000-0000000105564000 [ 4K] ---/rwx SM=NUL stack guard for thread 3 Stack 0000000105564000-00000001055e6000 [ 520K] rw-/rwx SM=COW thread 3 Application Specific Information: objc[20697]: garbage collection is OFF Thread 0:: Dispatch queue: com.apple.main-thread 0 org.python.python 0x00000001046aad93 PyEval_EvalFrameEx + 1601 1 org.python.python 0x00000001046b0df7 0x104626000 + 568823 2 org.python.python 0x00000001046ade0a PyEval_EvalFrameEx + 14008 3 org.python.python 0x00000001046b0cd8 PyEval_EvalCodeEx + 1996 4 org.python.python 0x000000010464eabf 0x104626000 + 166591 5 org.python.python 0x000000010462dd32 PyObject_Call + 97 6 org.python.python 0x000000010463c6e9 0x104626000 + 91881 7 org.python.python 0x000000010462dd32 PyObject_Call + 97 8 org.python.python 0x0000000104678484 0x104626000 + 337028 9 org.python.python 0x0000000104672b7a 0x104626000 + 314234 10 org.python.python 0x000000010462dd32 PyObject_Call + 97 11 org.python.python 0x00000001046adf63 PyEval_EvalFrameEx + 14353 12 org.python.python 0x00000001046b0df7 0x104626000 + 568823 13 org.python.python 0x00000001046ade0a PyEval_EvalFrameEx + 14008 14 org.python.python 0x00000001046b0cd8 PyEval_EvalCodeEx + 1996 15 org.python.python 0x000000010464eabf 0x104626000 + 166591 16 org.python.python 0x000000010462dd32 PyObject_Call + 97 17 org.python.python 0x00000001046ae5ec PyEval_EvalFrameEx + 16026 18 org.python.python 0x00000001046b0cd8 PyEval_EvalCodeEx + 1996 19 org.python.python 0x00000001046b0e6c 0x104626000 + 568940 20 org.python.python 0x00000001046ade0a PyEval_EvalFrameEx + 14008 21 org.python.python 0x00000001046b0df7 0x104626000 + 568823 22 org.python.python 0x00000001046ade0a PyEval_EvalFrameEx + 14008 23 org.python.python 0x00000001046b0cd8 PyEval_EvalCodeEx + 1996 24 org.python.python 0x00000001046b0d4d PyEval_EvalCode + 54 25 org.python.python 0x00000001046c808f 0x104626000 + 663695 26 org.python.python 0x00000001046c814f PyRun_FileExFlags + 157 27 org.python.python 0x00000001046c92a2 PyRun_SimpleFileExFlags + 392 28 org.python.python 0x00000001046d92af Py_Main + 2715 29 org.python.python 0x0000000104621e88 0x104621000 + 3720 Thread 1:: Dispatch queue: com.apple.libdispatch-manager 0 libsystem_kernel.dylib 0x00007fff85ab37e6 kevent + 10 1 libdispatch.dylib 0x00007fff864ae78a _dispatch_mgr_invoke + 923 2 libdispatch.dylib 0x00007fff864ad31a _dispatch_mgr_thread + 54 Thread 2:: Dispatch queue: com.apple.root.low-overcommit-priority 0 cl_kernels 0x000000011357db1c __aux_loop_Vectorized_wrapper + 1292 1 libcldcpuengine.dylib 0x0000000105412ef9 cld_exec_workitem + 789 2 libdispatch.dylib 0x00007fff864b0688 _dispatch_apply2 + 71 3 libdispatch.dylib 0x00007fff864b05ef dispatch_apply_f + 388 4 libcldcpuengine.dylib 0x0000000105412bc6 cld_exec_kernel + 84 5 libdispatch.dylib 0x00007fff864ae2d6 _dispatch_queue_drain + 264 6 libdispatch.dylib 0x00007fff864ae132 _dispatch_queue_invoke + 54 7 libdispatch.dylib 0x00007fff864ad92c _dispatch_worker_thread2 + 198 8 libsystem_c.dylib 0x00007fff867f33da _pthread_wqthread + 316 9 libsystem_c.dylib 0x00007fff867f4b85 start_wqthread + 13 Thread 3 Crashed:: Dispatch queue: com.apple.root.low-overcommit-priority 0 cl_kernels 0x000000011357db1c __aux_loop_Vectorized_wrapper + 1292 1 libcldcpuengine.dylib 0x0000000105412ef9 cld_exec_workitem + 789 2 libdispatch.dylib 0x00007fff864b0688 _dispatch_apply2 + 71 3 libdispatch.dylib 0x00007fff864ad965 _dispatch_worker_thread2 + 255 4 libsystem_c.dylib 0x00007fff867f33da _pthread_wqthread + 316 5 libsystem_c.dylib 0x00007fff867f4b85 start_wqthread + 13 Thread 4:: Dispatch queue: com.apple.root.low-overcommit-priority 0 cl_kernels 0x000000011357db1c __aux_loop_Vectorized_wrapper + 1292 1 libcldcpuengine.dylib 0x0000000105412ef9 cld_exec_workitem + 789 2 libdispatch.dylib 0x00007fff864b0688 _dispatch_apply2 + 71 3 libdispatch.dylib 0x00007fff864ad965 _dispatch_worker_thread2 + 255 4 libsystem_c.dylib 0x00007fff867f33da _pthread_wqthread + 316 5 libsystem_c.dylib 0x00007fff867f4b85 start_wqthread + 13 Thread 5: 0 libsystem_c.dylib 0x00007fff867f4b78 start_wqthread + 0 Thread 6: 0 libsystem_kernel.dylib 0x00007fff85ab3192 __workq_kernreturn + 10 1 libsystem_c.dylib 0x00007fff867f3594 _pthread_wqthread + 758 2 libsystem_c.dylib 0x00007fff867f4b85 start_wqthread + 13 Thread 3 crashed with X86 Thread State (64-bit): rax: 0x0000000100000000 rbx: 0x000000011342d000 rcx: 0x0000000100000000 rdx: 0x0000000000000000 rdi: 0x00000000000013c4 rsi: 0x00000001055e3c58 rbp: 0x00000001055e3d20 rsp: 0x0000000105563b70 r8: 0x0000000000000000 r9: 0x0000000040c00000 r10: 0x0000000100000000 r11: 0x0000000000000000 r12: 0x00000001055e3c58 r13: 0x00000000000013c3 r14: 0x000013c400000000 r15: 0x0000000105563b70 rip: 0x000000011357db1c rfl: 0x0000000000010202 cr2: 0x0000000105563b68 Logical CPU: 0

8 years, 10 months

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

PyOpenCL June 2012