Re: [PyCUDA] Install issue
by Andreas Kloeckner
Mike,
Mike McFarlane <mike.mcfarlane(a)iproov.com> writes:
>> I'm using Numpy 1.4.1. Thanks for your help.
First, please keep the list cc'd. That makes answers searchable.
Second, could you try upgrading that? You don't even need to mess with
your system to do so--just use a virtualenv.
Andreas
4 years, 7 months
Re: [PyCUDA] Install issue
by Andreas Kloeckner
Mike McFarlane <mike.mcfarlane(a)iproov.com> writes:
> Hi
>
> I've installed pycuda following
> http://wiki.tiker.net/PyCuda/Installation/Linux
>
> When I try to run test/test_driver.py it fails many tests, mainly with
> 'TypeError: 'numpy.ndarray' does not have the buffer interface'. The output
> is below for test_driver.py and the initial make.
>
> Can anyone explain what is wrong please?
>
> Thanks, and apologies if this mailing list isn't the right place for such
> Qs.
It is. What's your version of numpy?
Andreas
4 years, 7 months
Install issue
by Mike McFarlane
Hi
I've installed pycuda following
http://wiki.tiker.net/PyCuda/Installation/Linux
When I try to run test/test_driver.py it fails many tests, mainly with
'TypeError: 'numpy.ndarray' does not have the buffer interface'. The output
is below for test_driver.py and the initial make.
Can anyone explain what is wrong please?
Thanks, and apologies if this mailing list isn't the right place for such
Qs.
Mike
python test_driver.py
================================================= test session starts
=================================================
platform linux2 -- Python 2.6.6 -- pytest-2.3.5
collected 23 items
test_driver.py FFF..F.F.FFFF.FFsFFF...
====================================================== FAILURES
=======================================================
___________________________________________ TestDriver.test_simple_kernel_2
___________________________________________
args = (<test_driver.TestDriver instance at 0x22ceea8>,), kwargs = {}
pycuda = <module 'pycuda' from
'/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/__init__.pyc'>
ctx = <pycuda._driver.Context object at 0x229bb18>, clear_context_caches =
<function clear_context_caches at 0x1af6ed8>
collect = <built-in function collect>
def f(*args, **kwargs):
import pycuda.driver
# appears to be idempotent, i.e. no harm in calling it more than
once
pycuda.driver.init()
ctx = make_default_context()
try:
assert isinstance(ctx.get_device().name(), str)
assert isinstance(ctx.get_device().compute_capability(), tuple)
assert isinstance(ctx.get_device().get_attributes(), dict)
> inner_f(*args, **kwargs)
/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/tools.py:453:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
self = <test_driver.TestDriver instance at 0x22ceea8>
@mark_cuda_test
def test_simple_kernel_2(self):
mod = SourceModule("""
__global__ void multiply_them(float *dest, float *a, float *b)
{
const int i = threadIdx.x;
dest[i] = a[i] * b[i];
}
""")
multiply_them = mod.get_function("multiply_them")
a = np.random.randn(400).astype(np.float32)
b = np.random.randn(400).astype(np.float32)
> a_gpu = drv.to_device(a)
test_driver.py:66:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
bf_obj = array([-0.4856573 , 0.45444968, 1.2984767 , -0.02761065,
0.41128042,
...0.56486648, -0.43269619, 0.55279583, -1.11403978, -0.85410458],
dtype=float32)
def to_device(bf_obj):
import sys
if sys.version_info >= (2, 7):
bf = memoryview(bf_obj).tobytes()
else:
bf = buffer(bf_obj)
result = mem_alloc(len(bf))
> memcpy_htod(result, bf)
E TypeError: 'buffer' does not have the buffer interface
/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/driver.py:776:
TypeError
_______________________________________________ TestDriver.test_memory
________________________________________________
args = (<test_driver.TestDriver instance at 0x228dd88>,), kwargs = {}
pycuda = <module 'pycuda' from
'/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/__init__.pyc'>
ctx = <pycuda._driver.Context object at 0x23031b8>, clear_context_caches =
<function clear_context_caches at 0x1af6ed8>
collect = <built-in function collect>
def f(*args, **kwargs):
import pycuda.driver
# appears to be idempotent, i.e. no harm in calling it more than
once
pycuda.driver.init()
ctx = make_default_context()
try:
assert isinstance(ctx.get_device().name(), str)
assert isinstance(ctx.get_device().compute_capability(), tuple)
assert isinstance(ctx.get_device().get_attributes(), dict)
> inner_f(*args, **kwargs)
/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/tools.py:453:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
self = <test_driver.TestDriver instance at 0x228dd88>
@mark_cuda_test
def test_memory(self):
z = np.random.randn(400).astype(np.float32)
> new_z = drv.from_device_like(drv.to_device(z), z)
test_driver.py:28:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
bf_obj = array([ -4.73744690e-01, 1.29374540e+00, -6.42129302e-01,
3.951009...1, -1.27567184e+00, 1.12105417e+00,
-6.89846277e-01], dtype=float32)
def to_device(bf_obj):
import sys
if sys.version_info >= (2, 7):
bf = memoryview(bf_obj).tobytes()
else:
bf = buffer(bf_obj)
result = mem_alloc(len(bf))
> memcpy_htod(result, bf)
E TypeError: 'buffer' does not have the buffer interface
/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/driver.py:776:
TypeError
______________________________________________ TestDriver.test_gpuarray
_______________________________________________
args = (<test_driver.TestDriver instance at 0x2487b90>,), kwargs = {}
pycuda = <module 'pycuda' from
'/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/__init__.pyc'>
ctx = <pycuda._driver.Context object at 0x23030c8>, clear_context_caches =
<function clear_context_caches at 0x1af6ed8>
collect = <built-in function collect>
def f(*args, **kwargs):
import pycuda.driver
# appears to be idempotent, i.e. no harm in calling it more than
once
pycuda.driver.init()
ctx = make_default_context()
try:
assert isinstance(ctx.get_device().name(), str)
assert isinstance(ctx.get_device().compute_capability(), tuple)
assert isinstance(ctx.get_device().get_attributes(), dict)
> inner_f(*args, **kwargs)
/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/tools.py:453:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
self = <test_driver.TestDriver instance at 0x2487b90>
@mark_cuda_test
def test_gpuarray(self):
a = np.arange(200000, dtype=np.float32)
b = a + 17
import pycuda.gpuarray as gpuarray
> a_g = gpuarray.to_gpu(a)
test_driver.py:148:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
ary = array([ 0.00000000e+00, 1.00000000e+00, 2.00000000e+00, ...,
1.99997000e+05, 1.99998000e+05, 1.99999000e+05], dtype=float32)
allocator = <Boost.Python.function object at 0x1a8f7c0>
def to_gpu(ary, allocator=drv.mem_alloc):
"""converts a numpy array to a GPUArray"""
result = GPUArray(ary.shape, ary.dtype, allocator,
strides=ary.strides)
> result.set(ary)
/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/gpuarray.py:913:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
self = <[TypeError("'numpy.ndarray' does not have the buffer interface")
raised in repr()] SafeRepr object at 0x2477ea8>
ary = array([ 0.00000000e+00, 1.00000000e+00, 2.00000000e+00, ...,
1.99997000e+05, 1.99998000e+05, 1.99999000e+05], dtype=float32)
def set(self, ary):
assert ary.size == self.size
assert ary.dtype == self.dtype
if ary.strides != self.strides:
from warnings import warn
warn("Setting array from one with different strides/storage
order. "
"This will cease to work in 2013.x.",
stacklevel=2)
assert self.flags.forc
if self.size:
> drv.memcpy_htod(self.gpudata, ary)
E TypeError: 'numpy.ndarray' does not have the buffer interface
/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/gpuarray.py:228:
TypeError
____________________________________________ TestDriver.test_simple_kernel
____________________________________________
args = (<test_driver.TestDriver instance at 0x1e73908>,), kwargs = {}
pycuda = <module 'pycuda' from
'/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/__init__.pyc'>
ctx = <pycuda._driver.Context object at 0x24b2488>, clear_context_caches =
<function clear_context_caches at 0x1af6ed8>
collect = <built-in function collect>
def f(*args, **kwargs):
import pycuda.driver
# appears to be idempotent, i.e. no harm in calling it more than
once
pycuda.driver.init()
ctx = make_default_context()
try:
assert isinstance(ctx.get_device().name(), str)
assert isinstance(ctx.get_device().compute_capability(), tuple)
assert isinstance(ctx.get_device().get_attributes(), dict)
> inner_f(*args, **kwargs)
/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/tools.py:453:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
self = <test_driver.TestDriver instance at 0x1e73908>
@mark_cuda_test
def test_simple_kernel(self):
mod = SourceModule("""
__global__ void multiply_them(float *dest, float *a, float *b)
{
const int i = threadIdx.x;
dest[i] = a[i] * b[i];
}
""")
multiply_them = mod.get_function("multiply_them")
a = np.random.randn(400).astype(np.float32)
b = np.random.randn(400).astype(np.float32)
dest = np.zeros_like(a)
multiply_them(
drv.Out(dest), drv.In(a), drv.In(b),
> block=(400, 1, 1))
test_driver.py:49:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
func = <pycuda._driver.Function object at 0x229bd70>
args = (<pycuda.driver.Out object at 0x227dcd0>, <pycuda.driver.In object
at 0x227ded0>, <pycuda.driver.In object at 0x227ddd0>)
kwargs = {}, grid = (1, 1), stream = None, block = (400, 1, 1), shared = 0,
texrefs = [], time_kernel = False
handlers = [<pycuda.driver.Out object at 0x227dcd0>, <pycuda.driver.In
object at 0x227ded0>, <pycuda.driver.In object at 0x227ddd0>]
arg_buf =
'\x00\x00!\x00\x00\x00\x00\x00\x00\x07!\x00\x00\x00\x00\x00\x00\x0e!\x00\x00\x00\x00\x00'
handler = <pycuda.driver.In object at 0x227ded0>
def function_call(func, *args, **kwargs):
grid = kwargs.pop("grid", (1, 1))
stream = kwargs.pop("stream", None)
block = kwargs.pop("block", None)
shared = kwargs.pop("shared", 0)
texrefs = kwargs.pop("texrefs", [])
time_kernel = kwargs.pop("time_kernel", False)
if kwargs:
raise ValueError(
"extra keyword arguments: %s"
% (",".join(kwargs.iterkeys())))
if block is None:
raise ValueError("must specify block size")
func._set_block_shape(*block)
handlers, arg_buf = _build_arg_buf(args)
for handler in handlers:
> handler.pre_call(stream)
/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/driver.py:380:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
self = <pycuda.driver.In object at 0x227ded0>, stream = None
def pre_call(self, stream):
if stream is not None:
memcpy_htod(self.get_device_alloc(), self.array)
else:
> memcpy_htod(self.get_device_alloc(), self.array)
E TypeError: 'numpy.ndarray' does not have the buffer interface
/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/driver.py:82:
TypeError
_______________________________________
TestDriver.test_multichannel_2d_texture
_______________________________________
args = (<test_driver.TestDriver instance at 0x2478d88>,), kwargs = {}
pycuda = <module 'pycuda' from
'/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/__init__.pyc'>
ctx = <pycuda._driver.Context object at 0x24b2f50>, clear_context_caches =
<function clear_context_caches at 0x1af6ed8>
collect = <built-in function collect>
def f(*args, **kwargs):
import pycuda.driver
# appears to be idempotent, i.e. no harm in calling it more than
once
pycuda.driver.init()
ctx = make_default_context()
try:
assert isinstance(ctx.get_device().name(), str)
assert isinstance(ctx.get_device().compute_capability(), tuple)
assert isinstance(ctx.get_device().get_attributes(), dict)
> inner_f(*args, **kwargs)
/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/tools.py:453:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
self = <test_driver.TestDriver instance at 0x2478d88>
@mark_cuda_test
def test_multichannel_2d_texture(self):
mod = SourceModule("""
#define CHANNELS 4
texture<float4, 2, cudaReadModeElementType> mtx_tex;
__global__ void copy_texture(float *dest)
{
int row = threadIdx.x;
int col = threadIdx.y;
int w = blockDim.y;
float4 texval = tex2D(mtx_tex, row, col);
dest[(row*w+col)*CHANNELS + 0] = texval.x;
dest[(row*w+col)*CHANNELS + 1] = texval.y;
dest[(row*w+col)*CHANNELS + 2] = texval.z;
dest[(row*w+col)*CHANNELS + 3] = texval.w;
}
""")
copy_texture = mod.get_function("copy_texture")
mtx_tex = mod.get_texref("mtx_tex")
shape = (5, 6)
channels = 4
a = np.asarray(
np.random.randn(*((channels,)+shape)),
dtype=np.float32, order="F")
drv.bind_array_to_texref(
> drv.make_multichannel_2d_array(a, order="F"), mtx_tex)
test_driver.py:261:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
ndarray = array([[[-0.87390321, 0.78430617, 0.80751866, 0.0188568 ,
0.31150779,
... -1.94733143, -1.69470119, 0.88077658,
0.6898424 ]]], dtype=float32)
order = 'F'
def make_multichannel_2d_array(ndarray, order):
"""Channel count has to be the first dimension of the C{ndarray}."""
descr = ArrayDescriptor()
if order.upper() == "C":
h, w, num_channels = ndarray.shape
stride = 0
elif order.upper() == "F":
num_channels, w, h = ndarray.shape
stride = 2
else:
raise LogicError("order must be either F or C")
descr.width = w
descr.height = h
descr.format = dtype_to_array_format(ndarray.dtype)
descr.num_channels = num_channels
ary = Array(descr)
copy = Memcpy2D()
> copy.set_src_host(ndarray)
E TypeError: 'numpy.ndarray' does not have the buffer interface
/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/driver.py:744:
TypeError
_____________________________________________ TestDriver.test_2d_texture
______________________________________________
args = (<test_driver.TestDriver instance at 0x2478bd8>,), kwargs = {}
pycuda = <module 'pycuda' from
'/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/__init__.pyc'>
ctx = <pycuda._driver.Context object at 0x24b25f0>, clear_context_caches =
<function clear_context_caches at 0x1af6ed8>
collect = <built-in function collect>
def f(*args, **kwargs):
import pycuda.driver
# appears to be idempotent, i.e. no harm in calling it more than
once
pycuda.driver.init()
ctx = make_default_context()
try:
assert isinstance(ctx.get_device().name(), str)
assert isinstance(ctx.get_device().compute_capability(), tuple)
assert isinstance(ctx.get_device().get_attributes(), dict)
> inner_f(*args, **kwargs)
/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/tools.py:453:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
self = <test_driver.TestDriver instance at 0x2478bd8>
@mark_cuda_test
def test_2d_texture(self):
mod = SourceModule("""
texture<float, 2, cudaReadModeElementType> mtx_tex;
__global__ void copy_texture(float *dest)
{
int row = threadIdx.x;
int col = threadIdx.y;
int w = blockDim.y;
dest[row*w+col] = tex2D(mtx_tex, row, col);
}
""")
copy_texture = mod.get_function("copy_texture")
mtx_tex = mod.get_texref("mtx_tex")
shape = (3, 4)
a = np.random.randn(*shape).astype(np.float32)
> drv.matrix_to_texref(a, mtx_tex, order="F")
test_driver.py:188:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
matrix = array([[ 1.00075412, -1.55745828, -0.64481455, 1.14575994],
[-3.110294...],
[-2.00341797, 0.95837516, 0.69810289, -0.08028961]], dtype=float32)
texref = <pycuda._driver.TextureReference object at 0x24b0910>, order = 'F'
def matrix_to_texref(matrix, texref, order):
> bind_array_to_texref(matrix_to_array(matrix, order), texref)
/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/driver.py:764:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
matrix = array([[ 1.00075412, -1.55745828, -0.64481455, 1.14575994],
[-3.110294...],
[-2.00341797, 0.95837516, 0.69810289, -0.08028961]], dtype=float32)
order = 'F', allow_double_hack = False
def matrix_to_array(matrix, order, allow_double_hack=False):
if order.upper() == "C":
h, w = matrix.shape
stride = 0
elif order.upper() == "F":
w, h = matrix.shape
stride = -1
else:
raise LogicError("order must be either F or C")
matrix = np.asarray(matrix, order=order)
descr = ArrayDescriptor()
descr.width = w
descr.height = h
if matrix.dtype == np.float64 and allow_double_hack:
descr.format = array_format.SIGNED_INT32
descr.num_channels = 2
else:
descr.format = dtype_to_array_format(matrix.dtype)
descr.num_channels = 1
ary = Array(descr)
copy = Memcpy2D()
> copy.set_src_host(matrix)
E TypeError: 'numpy.ndarray' does not have the buffer interface
/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/driver.py:712:
TypeError
_____________________________________________ TestDriver.test_fp_textures
_____________________________________________
args = (<test_driver.TestDriver instance at 0x2582b00>,), kwargs = {}
pycuda = <module 'pycuda' from
'/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/__init__.pyc'>
ctx = <pycuda._driver.Context object at 0x24b2500>, clear_context_caches =
<function clear_context_caches at 0x1af6ed8>
collect = <built-in function collect>
def f(*args, **kwargs):
import pycuda.driver
# appears to be idempotent, i.e. no harm in calling it more than
once
pycuda.driver.init()
ctx = make_default_context()
try:
assert isinstance(ctx.get_device().name(), str)
assert isinstance(ctx.get_device().compute_capability(), tuple)
assert isinstance(ctx.get_device().get_attributes(), dict)
> inner_f(*args, **kwargs)
/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/tools.py:453:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
self = <test_driver.TestDriver instance at 0x2582b00>
@mark_cuda_test
def test_fp_textures(self):
if drv.Context.get_device().compute_capability() < (1, 3):
return
for tp in [np.float32, np.float64]:
from pycuda.tools import dtype_to_ctype
tp_cstr = dtype_to_ctype(tp)
mod = SourceModule("""
#include <pycuda-helpers.hpp>
texture<fp_tex_%(tp)s, 1, cudaReadModeElementType> my_tex;
__global__ void copy_texture(%(tp)s *dest)
{
int i = threadIdx.x;
dest[i] = fp_tex1Dfetch(my_tex, i);
}
""" % {"tp": tp_cstr})
copy_texture = mod.get_function("copy_texture")
my_tex = mod.get_texref("my_tex")
import pycuda.gpuarray as gpuarray
shape = (384,)
a = np.random.randn(*shape).astype(tp)
> a_gpu = gpuarray.to_gpu(a)
test_driver.py:522:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
ary = array([ 4.60776985e-01, 7.17113495e-01, 1.01722872e+00,
-1.146432...1,
2.04511788e-02, 1.11698084e-01, -2.13677096e+00], dtype=float32)
allocator = <Boost.Python.function object at 0x1a8f7c0>
def to_gpu(ary, allocator=drv.mem_alloc):
"""converts a numpy array to a GPUArray"""
result = GPUArray(ary.shape, ary.dtype, allocator,
strides=ary.strides)
> result.set(ary)
/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/gpuarray.py:913:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
self = <[TypeError("'numpy.ndarray' does not have the buffer interface")
raised in repr()] SafeRepr object at 0x2300dd0>
ary = array([ 4.60776985e-01, 7.17113495e-01, 1.01722872e+00,
-1.146432...1,
2.04511788e-02, 1.11698084e-01, -2.13677096e+00], dtype=float32)
def set(self, ary):
assert ary.size == self.size
assert ary.dtype == self.dtype
if ary.strides != self.strides:
from warnings import warn
warn("Setting array from one with different strides/storage
order. "
"This will cease to work in 2013.x.",
stacklevel=2)
assert self.flags.forc
if self.size:
> drv.memcpy_htod(self.gpudata, ary)
E TypeError: 'numpy.ndarray' does not have the buffer interface
/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/gpuarray.py:228:
TypeError
____________________________________________ TestDriver.test_vector_types
_____________________________________________
args = (<test_driver.TestDriver instance at 0x25a02d8>,), kwargs = {}
pycuda = <module 'pycuda' from
'/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/__init__.pyc'>
ctx = <pycuda._driver.Context object at 0x24b29b0>, clear_context_caches =
<function clear_context_caches at 0x1af6ed8>
collect = <built-in function collect>
def f(*args, **kwargs):
import pycuda.driver
# appears to be idempotent, i.e. no harm in calling it more than
once
pycuda.driver.init()
ctx = make_default_context()
try:
assert isinstance(ctx.get_device().name(), str)
assert isinstance(ctx.get_device().compute_capability(), tuple)
assert isinstance(ctx.get_device().get_attributes(), dict)
> inner_f(*args, **kwargs)
/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/tools.py:453:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
self = <test_driver.TestDriver instance at 0x25a02d8>
@mark_cuda_test
def test_vector_types(self):
mod = SourceModule("""
__global__ void set_them(float3 *dest, float3 x)
{
const int i = threadIdx.x;
dest[i] = x;
}
""")
set_them = mod.get_function("set_them")
a = gpuarray.vec.make_float3(1, 2, 3)
dest = np.empty((400), gpuarray.vec.float3)
> set_them(drv.Out(dest), a, block=(400,1,1))
test_driver.py:98:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
func = <pycuda._driver.Function object at 0x24b27d0>
args = (<pycuda.driver.Out object at 0x1e6fcd0>, array((1.0, 2.0, 3.0),
dtype=[('x', '<f4'), ('y', '<f4'), ('z', '<f4')]))
kwargs = {}, grid = (1, 1), stream = None, block = (400, 1, 1), shared = 0,
texrefs = [], time_kernel = False
handlers = [<pycuda.driver.Out object at 0x1e6fcd0>]
arg_buf = '\x00\x00!\x00\x00\x00\x00\x00\x00\x00\x80?\x00\x00\x00@\x00\x00@
@'
handler = <pycuda.driver.Out object at 0x1e6fcd0>
def function_call(func, *args, **kwargs):
grid = kwargs.pop("grid", (1, 1))
stream = kwargs.pop("stream", None)
block = kwargs.pop("block", None)
shared = kwargs.pop("shared", 0)
texrefs = kwargs.pop("texrefs", [])
time_kernel = kwargs.pop("time_kernel", False)
if kwargs:
raise ValueError(
"extra keyword arguments: %s"
% (",".join(kwargs.iterkeys())))
if block is None:
raise ValueError("must specify block size")
func._set_block_shape(*block)
handlers, arg_buf = _build_arg_buf(args)
for handler in handlers:
handler.pre_call(stream)
for texref in texrefs:
func.param_set_texref(texref)
post_handlers = [handler
for handler in handlers
if hasattr(handler, "post_call")]
if stream is None:
if time_kernel:
Context.synchronize()
from time import time
start_time = time()
func._launch_kernel(grid, block, arg_buf, shared, None)
if post_handlers or time_kernel:
Context.synchronize()
if time_kernel:
run_time = time()-start_time
for handler in post_handlers:
> handler.post_call(stream)
/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/driver.py:405:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
self = <pycuda.driver.Out object at 0x1e6fcd0>, stream = None
def post_call(self, stream):
if stream is not None:
memcpy_dtoh(self.array, self.get_device_alloc())
else:
> memcpy_dtoh(self.array, self.get_device_alloc())
E TypeError: 'numpy.ndarray' does not have the buffer interface
/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/driver.py:90:
TypeError
________________________________________
TestDriver.test_multiple_2d_textures
_________________________________________
args = (<test_driver.TestDriver instance at 0x280f950>,), kwargs = {}
pycuda = <module 'pycuda' from
'/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/__init__.pyc'>
ctx = <pycuda._driver.Context object at 0x24b2e60>, clear_context_caches =
<function clear_context_caches at 0x1af6ed8>
collect = <built-in function collect>
def f(*args, **kwargs):
import pycuda.driver
# appears to be idempotent, i.e. no harm in calling it more than
once
pycuda.driver.init()
ctx = make_default_context()
try:
assert isinstance(ctx.get_device().name(), str)
assert isinstance(ctx.get_device().compute_capability(), tuple)
assert isinstance(ctx.get_device().get_attributes(), dict)
> inner_f(*args, **kwargs)
/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/tools.py:453:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
self = <test_driver.TestDriver instance at 0x280f950>
@mark_cuda_test
def test_multiple_2d_textures(self):
mod = SourceModule("""
texture<float, 2, cudaReadModeElementType> mtx_tex;
texture<float, 2, cudaReadModeElementType> mtx2_tex;
__global__ void copy_texture(float *dest)
{
int row = threadIdx.x;
int col = threadIdx.y;
int w = blockDim.y;
dest[row*w+col] =
tex2D(mtx_tex, row, col)
+
tex2D(mtx2_tex, row, col);
}
""")
copy_texture = mod.get_function("copy_texture")
mtx_tex = mod.get_texref("mtx_tex")
mtx2_tex = mod.get_texref("mtx2_tex")
shape = (3,4)
a = np.random.randn(*shape).astype(np.float32)
b = np.random.randn(*shape).astype(np.float32)
> drv.matrix_to_texref(a, mtx_tex, order="F")
test_driver.py:223:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
matrix = array([[-0.60509264, -0.69680727, -1.962533 , 0.67714691],
[-1.094154...],
[ 0.40417606, -0.97287714, 0.10696917, -1.54577947]], dtype=float32)
texref = <pycuda._driver.TextureReference object at 0x24b0d00>, order = 'F'
def matrix_to_texref(matrix, texref, order):
> bind_array_to_texref(matrix_to_array(matrix, order), texref)
/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/driver.py:764:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
matrix = array([[-0.60509264, -0.69680727, -1.962533 , 0.67714691],
[-1.094154...],
[ 0.40417606, -0.97287714, 0.10696917, -1.54577947]], dtype=float32)
order = 'F', allow_double_hack = False
def matrix_to_array(matrix, order, allow_double_hack=False):
if order.upper() == "C":
h, w = matrix.shape
stride = 0
elif order.upper() == "F":
w, h = matrix.shape
stride = -1
else:
raise LogicError("order must be either F or C")
matrix = np.asarray(matrix, order=order)
descr = ArrayDescriptor()
descr.width = w
descr.height = h
if matrix.dtype == np.float64 and allow_double_hack:
descr.format = array_format.SIGNED_INT32
descr.num_channels = 2
else:
descr.format = dtype_to_array_format(matrix.dtype)
descr.num_channels = 1
ary = Array(descr)
copy = Memcpy2D()
> copy.set_src_host(matrix)
E TypeError: 'numpy.ndarray' does not have the buffer interface
/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/driver.py:712:
TypeError
___________________________________________ TestDriver.test_streamed_kernel
___________________________________________
args = (<test_driver.TestDriver instance at 0x23007e8>,), kwargs = {}
pycuda = <module 'pycuda' from
'/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/__init__.pyc'>
ctx = <pycuda._driver.Context object at 0x24b2e60>, clear_context_caches =
<function clear_context_caches at 0x1af6ed8>
collect = <built-in function collect>
def f(*args, **kwargs):
import pycuda.driver
# appears to be idempotent, i.e. no harm in calling it more than
once
pycuda.driver.init()
ctx = make_default_context()
try:
assert isinstance(ctx.get_device().name(), str)
assert isinstance(ctx.get_device().compute_capability(), tuple)
assert isinstance(ctx.get_device().get_attributes(), dict)
> inner_f(*args, **kwargs)
/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/tools.py:453:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
self = <test_driver.TestDriver instance at 0x23007e8>
@mark_cuda_test
def test_streamed_kernel(self):
# this differs from the "simple_kernel" case in that *all*
computation
# and data copying is asynchronous. Observe how this necessitates
the
# use of page-locked memory.
mod = SourceModule("""
__global__ void multiply_them(float *dest, float *a, float *b)
{
const int i = threadIdx.x*blockDim.y + threadIdx.y;
dest[i] = a[i] * b[i];
}
""")
multiply_them = mod.get_function("multiply_them")
shape = (32, 8)
a = drv.pagelocked_zeros(shape, dtype=np.float32)
b = drv.pagelocked_zeros(shape, dtype=np.float32)
a[:] = np.random.randn(*shape)
b[:] = np.random.randn(*shape)
a_gpu = drv.mem_alloc(a.nbytes)
b_gpu = drv.mem_alloc(b.nbytes)
strm = drv.Stream()
> drv.memcpy_htod_async(a_gpu, a, strm)
E TypeError: 'numpy.ndarray' does not have the buffer interface
test_driver.py:127: TypeError
_________________________________________
TestDriver.test_prepared_invocation
_________________________________________
args = (<test_driver.TestDriver instance at 0x24951b8>,), kwargs = {}
pycuda = <module 'pycuda' from
'/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/__init__.pyc'>
ctx = <pycuda._driver.Context object at 0x24a20c8>, clear_context_caches =
<function clear_context_caches at 0x1af6ed8>
collect = <built-in function collect>
def f(*args, **kwargs):
import pycuda.driver
# appears to be idempotent, i.e. no harm in calling it more than
once
pycuda.driver.init()
ctx = make_default_context()
try:
assert isinstance(ctx.get_device().name(), str)
assert isinstance(ctx.get_device().compute_capability(), tuple)
assert isinstance(ctx.get_device().get_attributes(), dict)
> inner_f(*args, **kwargs)
/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/tools.py:453:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
self = <test_driver.TestDriver instance at 0x24951b8>
@mark_cuda_test
def test_prepared_invocation(self):
a = np.random.randn(4,4).astype(np.float32)
a_gpu = drv.mem_alloc(a.size * a.dtype.itemsize)
> drv.memcpy_htod(a_gpu, a)
E TypeError: 'numpy.ndarray' does not have the buffer interface
test_driver.py:450: TypeError
___________________________________________ TestDriver.test_constant_memory
___________________________________________
args = (<test_driver.TestDriver instance at 0x24963f8>,), kwargs = {}
pycuda = <module 'pycuda' from
'/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/__init__.pyc'>
ctx = <pycuda._driver.Context object at 0x24a2500>, clear_context_caches =
<function clear_context_caches at 0x1af6ed8>
collect = <built-in function collect>
def f(*args, **kwargs):
import pycuda.driver
# appears to be idempotent, i.e. no harm in calling it more than
once
pycuda.driver.init()
ctx = make_default_context()
try:
assert isinstance(ctx.get_device().name(), str)
assert isinstance(ctx.get_device().compute_capability(), tuple)
assert isinstance(ctx.get_device().get_attributes(), dict)
> inner_f(*args, **kwargs)
/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/tools.py:453:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
self = <test_driver.TestDriver instance at 0x24963f8>
@mark_cuda_test
def test_constant_memory(self):
# contributed by Andrew Wagner
module = SourceModule("""
__constant__ float const_array[32];
__global__ void copy_constant_into_global(float*
global_result_array)
{
global_result_array[threadIdx.x] = const_array[threadIdx.x];
}
""")
copy_constant_into_global =
module.get_function("copy_constant_into_global")
const_array, _ = module.get_global('const_array')
host_array = np.random.randint(0,255,(32,)).astype(np.float32)
global_result_array = drv.mem_alloc_like(host_array)
> drv.memcpy_htod(const_array, host_array)
E TypeError: 'numpy.ndarray' does not have the buffer interface
test_driver.py:551: TypeError
_____________________________________
TestDriver.test_multichannel_linear_texture
_____________________________________
args = (<test_driver.TestDriver instance at 0x1e61cb0>,), kwargs = {}
pycuda = <module 'pycuda' from
'/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/__init__.pyc'>
ctx = <pycuda._driver.Context object at 0x24a2398>, clear_context_caches =
<function clear_context_caches at 0x1af6ed8>
collect = <built-in function collect>
def f(*args, **kwargs):
import pycuda.driver
# appears to be idempotent, i.e. no harm in calling it more than
once
pycuda.driver.init()
ctx = make_default_context()
try:
assert isinstance(ctx.get_device().name(), str)
assert isinstance(ctx.get_device().compute_capability(), tuple)
assert isinstance(ctx.get_device().get_attributes(), dict)
> inner_f(*args, **kwargs)
/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/tools.py:453:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
self = <test_driver.TestDriver instance at 0x1e61cb0>
@mark_cuda_test
def test_multichannel_linear_texture(self):
mod = SourceModule("""
#define CHANNELS 4
texture<float4, 1, cudaReadModeElementType> mtx_tex;
__global__ void copy_texture(float *dest)
{
int i = threadIdx.x+blockDim.x*threadIdx.y;
float4 texval = tex1Dfetch(mtx_tex, i);
dest[i*CHANNELS + 0] = texval.x;
dest[i*CHANNELS + 1] = texval.y;
dest[i*CHANNELS + 2] = texval.z;
dest[i*CHANNELS + 3] = texval.w;
}
""")
copy_texture = mod.get_function("copy_texture")
mtx_tex = mod.get_texref("mtx_tex")
shape = (16, 16)
channels = 4
a = np.random.randn(*(shape+(channels,))).astype(np.float32)
> a_gpu = drv.to_device(a)
test_driver.py:297:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
bf_obj = array([[[ 0.06351238, -1.50826311, 0.99232352, -0.19848856],
[ 0.1630...
[-0.59885794, 0.58485585, 0.93704814, 0.02101008]]],
dtype=float32)
def to_device(bf_obj):
import sys
if sys.version_info >= (2, 7):
bf = memoryview(bf_obj).tobytes()
else:
bf = buffer(bf_obj)
result = mem_alloc(len(bf))
> memcpy_htod(result, bf)
E TypeError: 'buffer' does not have the buffer interface
/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/driver.py:776:
TypeError
_____________________________________________ TestDriver.test_3d_texture
______________________________________________
args = (<test_driver.TestDriver instance at 0x2585ea8>,), kwargs = {}
pycuda = <module 'pycuda' from
'/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/__init__.pyc'>
ctx = <pycuda._driver.Context object at 0x24a2848>, clear_context_caches =
<function clear_context_caches at 0x1af6ed8>
collect = <built-in function collect>
def f(*args, **kwargs):
import pycuda.driver
# appears to be idempotent, i.e. no harm in calling it more than
once
pycuda.driver.init()
ctx = make_default_context()
try:
assert isinstance(ctx.get_device().name(), str)
assert isinstance(ctx.get_device().compute_capability(), tuple)
assert isinstance(ctx.get_device().get_attributes(), dict)
> inner_f(*args, **kwargs)
/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg/pycuda/tools.py:453:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
self = <test_driver.TestDriver instance at 0x2585ea8>
@mark_cuda_test
def test_3d_texture(self):
# adapted from code by Nicolas Pinto
w = 2
h = 4
d = 8
shape = (w, h, d)
a = np.asarray(
np.random.randn(*shape),
dtype=np.float32, order="F")
descr = drv.ArrayDescriptor3D()
descr.width = w
descr.height = h
descr.depth = d
descr.format = drv.dtype_to_array_format(a.dtype)
descr.num_channels = 1
descr.flags = 0
ary = drv.Array(descr)
copy = drv.Memcpy3D()
> copy.set_src_host(a)
E TypeError: 'numpy.ndarray' does not have the buffer interface
test_driver.py:412: TypeError
=================================== 14 failed, 8 passed, 1 skipped in 13.38
seconds ===================================
--
------------------------- make output -----------------
*** WARNING: nvcc not in path.
running install
running bdist_egg
running egg_info
writing requirements to pycuda.egg-info/requires.txt
writing pycuda.egg-info/PKG-INFO
writing top-level names to pycuda.egg-info/top_level.txt
writing dependency_links to pycuda.egg-info/dependency_links.txt
reading manifest file 'pycuda.egg-info/SOURCES.txt'
reading manifest template 'MANIFEST.in'
warning: no files found matching '*.cpp' under directory
'bpl-subset/bpl_subset/boost'
warning: no files found matching '*.html' under directory
'bpl-subset/bpl_subset/boost'
warning: no files found matching '*.inl' under directory
'bpl-subset/bpl_subset/boost'
warning: no files found matching '*.txt' under directory
'bpl-subset/bpl_subset/boost'
warning: no files found matching '*.h' under directory
'bpl-subset/bpl_subset/libs'
warning: no files found matching '*.ipp' under directory
'bpl-subset/bpl_subset/libs'
warning: no files found matching '*.pl' under directory
'bpl-subset/bpl_subset/libs'
writing manifest file 'pycuda.egg-info/SOURCES.txt'
installing library code to build/bdist.linux-x86_64/egg
running install_lib
running build_py
running build_ext
creating build/bdist.linux-x86_64/egg
creating build/bdist.linux-x86_64/egg/pycuda
copying build/lib.linux-x86_64-2.6/pycuda/autoinit.py ->
build/bdist.linux-x86_64/egg/pycuda
copying build/lib.linux-x86_64-2.6/pycuda/compiler.py ->
build/bdist.linux-x86_64/egg/pycuda
copying build/lib.linux-x86_64-2.6/pycuda/tools.py ->
build/bdist.linux-x86_64/egg/pycuda
copying build/lib.linux-x86_64-2.6/pycuda/__init__.py ->
build/bdist.linux-x86_64/egg/pycuda
creating build/bdist.linux-x86_64/egg/pycuda/compyte
copying build/lib.linux-x86_64-2.6/pycuda/compyte/__init__.py ->
build/bdist.linux-x86_64/egg/pycuda/compyte
copying build/lib.linux-x86_64-2.6/pycuda/compyte/array.py ->
build/bdist.linux-x86_64/egg/pycuda/compyte
copying build/lib.linux-x86_64-2.6/pycuda/compyte/dtypes.py ->
build/bdist.linux-x86_64/egg/pycuda/compyte
copying build/lib.linux-x86_64-2.6/pycuda/gpuarray.py ->
build/bdist.linux-x86_64/egg/pycuda
copying build/lib.linux-x86_64-2.6/pycuda/_mymako.py ->
build/bdist.linux-x86_64/egg/pycuda
copying build/lib.linux-x86_64-2.6/pycuda/debug.py ->
build/bdist.linux-x86_64/egg/pycuda
creating build/bdist.linux-x86_64/egg/pycuda/sparse
copying build/lib.linux-x86_64-2.6/pycuda/sparse/__init__.py ->
build/bdist.linux-x86_64/egg/pycuda/sparse
copying build/lib.linux-x86_64-2.6/pycuda/sparse/pkt_build.py ->
build/bdist.linux-x86_64/egg/pycuda/sparse
copying build/lib.linux-x86_64-2.6/pycuda/sparse/cg.py ->
build/bdist.linux-x86_64/egg/pycuda/sparse
copying build/lib.linux-x86_64-2.6/pycuda/sparse/packeted.py ->
build/bdist.linux-x86_64/egg/pycuda/sparse
copying build/lib.linux-x86_64-2.6/pycuda/sparse/coordinate.py ->
build/bdist.linux-x86_64/egg/pycuda/sparse
copying build/lib.linux-x86_64-2.6/pycuda/sparse/operator.py ->
build/bdist.linux-x86_64/egg/pycuda/sparse
copying build/lib.linux-x86_64-2.6/pycuda/sparse/inner.py ->
build/bdist.linux-x86_64/egg/pycuda/sparse
copying build/lib.linux-x86_64-2.6/pycuda/_pvt_struct.so ->
build/bdist.linux-x86_64/egg/pycuda
copying build/lib.linux-x86_64-2.6/pycuda/cumath.py ->
build/bdist.linux-x86_64/egg/pycuda
copying build/lib.linux-x86_64-2.6/pycuda/_cluda.py ->
build/bdist.linux-x86_64/egg/pycuda
copying build/lib.linux-x86_64-2.6/pycuda/characterize.py ->
build/bdist.linux-x86_64/egg/pycuda
creating build/bdist.linux-x86_64/egg/pycuda/cuda
copying build/lib.linux-x86_64-2.6/pycuda/cuda/pycuda-complex.hpp ->
build/bdist.linux-x86_64/egg/pycuda/cuda
copying build/lib.linux-x86_64-2.6/pycuda/cuda/pycuda-complex-impl.hpp ->
build/bdist.linux-x86_64/egg/pycuda/cuda
copying build/lib.linux-x86_64-2.6/pycuda/cuda/pycuda-helpers.hpp ->
build/bdist.linux-x86_64/egg/pycuda/cuda
copying build/lib.linux-x86_64-2.6/pycuda/reduction.py ->
build/bdist.linux-x86_64/egg/pycuda
creating build/bdist.linux-x86_64/egg/pycuda/gl
copying build/lib.linux-x86_64-2.6/pycuda/gl/autoinit.py ->
build/bdist.linux-x86_64/egg/pycuda/gl
copying build/lib.linux-x86_64-2.6/pycuda/gl/__init__.py ->
build/bdist.linux-x86_64/egg/pycuda/gl
copying build/lib.linux-x86_64-2.6/pycuda/driver.py ->
build/bdist.linux-x86_64/egg/pycuda
copying build/lib.linux-x86_64-2.6/pycuda/curandom.py ->
build/bdist.linux-x86_64/egg/pycuda
copying build/lib.linux-x86_64-2.6/pycuda/elementwise.py ->
build/bdist.linux-x86_64/egg/pycuda
copying build/lib.linux-x86_64-2.6/pycuda/scan.py ->
build/bdist.linux-x86_64/egg/pycuda
copying build/lib.linux-x86_64-2.6/pycuda/_driver.so ->
build/bdist.linux-x86_64/egg/pycuda
byte-compiling build/bdist.linux-x86_64/egg/pycuda/autoinit.py to
autoinit.pyc
byte-compiling build/bdist.linux-x86_64/egg/pycuda/compiler.py to
compiler.pyc
byte-compiling build/bdist.linux-x86_64/egg/pycuda/tools.py to tools.pyc
byte-compiling build/bdist.linux-x86_64/egg/pycuda/__init__.py to
__init__.pyc
byte-compiling build/bdist.linux-x86_64/egg/pycuda/compyte/__init__.py to
__init__.pyc
byte-compiling build/bdist.linux-x86_64/egg/pycuda/compyte/array.py to
array.pyc
byte-compiling build/bdist.linux-x86_64/egg/pycuda/compyte/dtypes.py to
dtypes.pyc
byte-compiling build/bdist.linux-x86_64/egg/pycuda/gpuarray.py to
gpuarray.pyc
byte-compiling build/bdist.linux-x86_64/egg/pycuda/_mymako.py to _mymako.pyc
byte-compiling build/bdist.linux-x86_64/egg/pycuda/debug.py to debug.pyc
byte-compiling build/bdist.linux-x86_64/egg/pycuda/sparse/__init__.py to
__init__.pyc
byte-compiling build/bdist.linux-x86_64/egg/pycuda/sparse/pkt_build.py to
pkt_build.pyc
byte-compiling build/bdist.linux-x86_64/egg/pycuda/sparse/cg.py to cg.pyc
byte-compiling build/bdist.linux-x86_64/egg/pycuda/sparse/packeted.py to
packeted.pyc
byte-compiling build/bdist.linux-x86_64/egg/pycuda/sparse/coordinate.py to
coordinate.pyc
byte-compiling build/bdist.linux-x86_64/egg/pycuda/sparse/operator.py to
operator.pyc
byte-compiling build/bdist.linux-x86_64/egg/pycuda/sparse/inner.py to
inner.pyc
byte-compiling build/bdist.linux-x86_64/egg/pycuda/cumath.py to cumath.pyc
byte-compiling build/bdist.linux-x86_64/egg/pycuda/_cluda.py to _cluda.pyc
byte-compiling build/bdist.linux-x86_64/egg/pycuda/characterize.py to
characterize.pyc
byte-compiling build/bdist.linux-x86_64/egg/pycuda/reduction.py to
reduction.pyc
byte-compiling build/bdist.linux-x86_64/egg/pycuda/gl/autoinit.py to
autoinit.pyc
byte-compiling build/bdist.linux-x86_64/egg/pycuda/gl/__init__.py to
__init__.pyc
byte-compiling build/bdist.linux-x86_64/egg/pycuda/driver.py to driver.pyc
byte-compiling build/bdist.linux-x86_64/egg/pycuda/curandom.py to
curandom.pyc
byte-compiling build/bdist.linux-x86_64/egg/pycuda/elementwise.py to
elementwise.pyc
byte-compiling build/bdist.linux-x86_64/egg/pycuda/scan.py to scan.pyc
creating stub loader for pycuda/_driver.so
creating stub loader for pycuda/_pvt_struct.so
byte-compiling build/bdist.linux-x86_64/egg/pycuda/_driver.py to _driver.pyc
byte-compiling build/bdist.linux-x86_64/egg/pycuda/_pvt_struct.py to
_pvt_struct.pyc
creating build/bdist.linux-x86_64/egg/EGG-INFO
copying pycuda.egg-info/PKG-INFO -> build/bdist.linux-x86_64/egg/EGG-INFO
copying pycuda.egg-info/SOURCES.txt -> build/bdist.linux-x86_64/egg/EGG-INFO
copying pycuda.egg-info/dependency_links.txt ->
build/bdist.linux-x86_64/egg/EGG-INFO
copying pycuda.egg-info/not-zip-safe ->
build/bdist.linux-x86_64/egg/EGG-INFO
copying pycuda.egg-info/requires.txt ->
build/bdist.linux-x86_64/egg/EGG-INFO
copying pycuda.egg-info/top_level.txt ->
build/bdist.linux-x86_64/egg/EGG-INFO
writing build/bdist.linux-x86_64/egg/EGG-INFO/native_libs.txt
creating 'dist/pycuda-2014.1-py2.6-linux-x86_64.egg' and adding
'build/bdist.linux-x86_64/egg' to it
removing 'build/bdist.linux-x86_64/egg' (and everything under it)
Processing pycuda-2014.1-py2.6-linux-x86_64.egg
removing
'/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg'
(and everything under it)
creating
/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg
Extracting pycuda-2014.1-py2.6-linux-x86_64.egg to
/usr/lib64/python2.6/site-packages
pycuda 2014.1 is already the active version in easy-install.pth
Installed
/usr/lib64/python2.6/site-packages/pycuda-2014.1-py2.6-linux-x86_64.egg
Processing dependencies for pycuda==2014.1
Searching for decorator==3.4.0
Best match: decorator 3.4.0
Processing decorator-3.4.0-py2.6.egg
decorator 3.4.0 is already the active version in easy-install.pth
Using /usr/lib64/python2.6/site-packages/decorator-3.4.0-py2.6.egg
Searching for pytest==2.3.5
Best match: pytest 2.3.5
Adding pytest 2.3.5 to easy-install.pth file
Installing py.test script to /usr/bin
Installing py.test-2.6 script to /usr/bin
Using /usr/lib/python2.6/site-packages
Searching for pytools==2014.3.4
Best match: pytools 2014.3.4
Processing pytools-2014.3.4-py2.6.egg
pytools 2014.3.4 is already the active version in easy-install.pth
Using /usr/lib64/python2.6/site-packages/pytools-2014.3.4-py2.6.egg
Searching for py==1.4.18
Best match: py 1.4.18
Adding py 1.4.18 to easy-install.pth file
Using /usr/lib/python2.6/site-packages
Searching for appdirs==1.4.0
Best match: appdirs 1.4.0
Processing appdirs-1.4.0-py2.6.egg
appdirs 1.4.0 is already the active version in easy-install.pth
Using /usr/lib64/python2.6/site-packages/appdirs-1.4.0-py2.6.egg
Finished processing dependencies for pycuda==2014.1
******************************************************************************************
*Mike McFarlane*
mike.mcfarlane(a)iproov.com
+44 7557 780175
******************************************************************************************
4 years, 7 months
Re: [PyCUDA] Windows installation instructions have been deleted from the wiki
by Andreas Kloeckner
Daniel Pineo <daniel(a)pineo.net> writes:
> The instructions to build and install PyCUDA on Windows have been removed
> from the wiki. There was a very detailed set of instructions that I had
> previously used to successfully install. Can these still be found
> somewhere? Is there a way to restore them to the wiki?
I try to catch vandalism of this nature and undo it right away,
but this seems to have slipped through the cracks. Sorry about that, it
should be back up.
Andreas
4 years, 8 months
Windows installation instructions have been deleted from the wiki
by Daniel Pineo
The instructions to build and install PyCUDA on Windows have been removed
from the wiki. There was a very detailed set of instructions that I had
previously used to successfully install. Can these still be found
somewhere? Is there a way to restore them to the wiki?
4 years, 8 months
Re: [PyCUDA] Question How to Safely Use pycuda with mpi4py
by kjs
Eric Larson wrote:
> Hey Kevin,
>
> Not sure about the CUDA limitations, I'll let others speak to that...
>
> But in developing the mne-python CUDA filtering code, IIRC the primary
> limitation was (by far) transferring the data to and from the GPU. The FFT
> computations themselves were a fraction of the total time. I suspect using
> multiple jobs won't help CUDA filtering very much since the jobs would
> presumably compete for the same memory bandwidth, but I would love to be
> wrong about this. If it works better, it would be great to open an
> mne-python issue for it, as we are always looking for speedups :)
>
> Cheers,
> Eric
> On Nov 1, 2014 7:21 PM, "kjs" <bfb(a)riseup.net> wrote:
>
>> Hello,
>>
>> I have written an MPI routine in Python that sends jobs to N worker
>> processes. The root process handles file IO and the workers do
>> computation. In the worker processes calls are made to the cuda enabled
>> GPU to do FFTs.
>>
>> Is it safe to have N processes potentially making calls to the same GPU
>> at the same time? I have not made any amendments to the cuda code[0],
>> and have little knowledge of what could possibly go wrong.
>>
>> Thanks much,
>> Kevin
>>
>> [0] I am using python-mne with cuda enabled to call scikits.cuda.fft
>> https://github.com/mne-tools/mne-python/blob/master/mne/cuda.py
>>
>> _______________________________________________
>> PyCUDA mailing list
>> PyCUDA(a)tiker.net
>> http://lists.tiker.net/listinfo/pycuda
>>
>>
>
Thanks Andreas, this is good to know. I noticed that even though pycuda
is currently only using one of two GPUs, that GPU is only ever at ~35%
memory and ~22% processing utilization. This could be related to Eric's
observation that the PCI-e 16x bus bandwidth reaches capacity while the
GPU is pushing out fast FFT'ed arrays. Thus allowing for only one or two
arrays in the GPU at the same time.
From what I have seen, using cuda speeds up my FFTs ~2x. Though, the
workers do many other computations on the CPU. It's a worst case
scenario that all N workers are trying to send data to the GPU at the
same time.
-Kevin
4 years, 8 months
Re: [PyCUDA] Problem with PyCUDA on Ubuntu 14.10
by Tomasz Rybak
Dnia 2014-11-01, sob o godzinie 17:01 -0700, Eric Larson pisze:
> Before I left work on Friday I checked to ensure that the packages in
> Synaptic were all NVIDIA version 331 (I think). Even if the numbers
> are right in Synaptic, I suppose there could still be a version
> mismatch somewhere, so I'll check "dmesg" specifically on Monday.
>
>
> FWIW I am pretty sure that the CUDA version jumped up from 5.5 (or
> 5.0?) in 14.04 up to CUDA 6.0 in 14.10. But I assume that PyCUDA is
> designed (and tested) to be forward-compatible, so I doubt that's the
> problem.
It depends on how Ubuntu builds PyCUDA.
Check pycuda package dependencies:
$ apt-cache show python-pycuda
whether there is any dependency on package
with 5.5 in its name, or only 6.0 packages.
Regards
--
Tomasz Rybak GPG/PGP key ID: 2AD5 9860
Fingerprint A481 824E 7DD3 9C0E C40A 488E C654 FB33 2AD5 9860
http://member.acm.org/~tomaszrybak
4 years, 8 months
Question How to Safely Use pycuda with mpi4py
by kjs
Hello,
I have written an MPI routine in Python that sends jobs to N worker
processes. The root process handles file IO and the workers do
computation. In the worker processes calls are made to the cuda enabled
GPU to do FFTs.
Is it safe to have N processes potentially making calls to the same GPU
at the same time? I have not made any amendments to the cuda code[0],
and have little knowledge of what could possibly go wrong.
Thanks much,
Kevin
[0] I am using python-mne with cuda enabled to call scikits.cuda.fft
https://github.com/mne-tools/mne-python/blob/master/mne/cuda.py
4 years, 8 months
Re: [PyCUDA] Problem with PyCUDA on Ubuntu 14.10
by Andreas Kloeckner
Eric Larson <larson.eric.d(a)gmail.com> writes:
> PyCUDA worked perfectly on Ubuntu 14.04, but after upgrade to 14.10 I get
> the following in both Python 2 and Python 3:
>
>>>> import pycuda.autoinit
> Traceback (most recent call last):
> File "<stdin>", line 1, in <module>
> File
> "/home/larsoner/.local/lib/python2.7/site-packages/pycuda-2014.1-py2.7-linux-x86_64.egg/pycuda/autoinit.py",
> line 4, in <module>
> cuda.init()
> pycuda._driver.Error: cuInit failed: unknown
>
> This is the same on latest `master` (shown above) and using the version in
> the Ubuntu repos. I can manually compile and run at least this example
> (devicequery):
>
> http://www.cac.cornell.edu/vw/gpu/example_submit.aspx
>
> and my system is using the proprietary NVIDIA drivers, so CUDA appears to
> be configured properly. Has anyone else experienced this? I've reported the
> bug here as well:
>
> https://bugs.launchpad.net/ubuntu/+source/pycuda/+bug/1388217
Check if 'dmesg' reveals anything. You might have a mismatch between
libcuda.so and you driver. (If so, good job, Ubuntu.)
Andreas
4 years, 8 months
Problem with PyCUDA on Ubuntu 14.10
by Eric Larson
PyCUDA worked perfectly on Ubuntu 14.04, but after upgrade to 14.10 I get
the following in both Python 2 and Python 3:
>>> import pycuda.autoinit
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File
"/home/larsoner/.local/lib/python2.7/site-packages/pycuda-2014.1-py2.7-linux-x86_64.egg/pycuda/autoinit.py",
line 4, in <module>
cuda.init()
pycuda._driver.Error: cuInit failed: unknown
This is the same on latest `master` (shown above) and using the version in
the Ubuntu repos. I can manually compile and run at least this example
(devicequery):
http://www.cac.cornell.edu/vw/gpu/example_submit.aspx
and my system is using the proprietary NVIDIA drivers, so CUDA appears to
be configured properly. Has anyone else experienced this? I've reported the
bug here as well:
https://bugs.launchpad.net/ubuntu/+source/pycuda/+bug/1388217
Cheers,
Eric
4 years, 8 months
