Skip to content

P
pytensor
提交 e4242a33 authored 作者: Ludwig Schmidt-Hackenberg's avatar Ludwig Schmidt-Hackenberg
浏览文件
操作

works now also for vector, tensor3 and tensor4 and for more than 2 inputs + corrected memory bugs

上级 403ab069
隐藏空白字符变更
内嵌 并排
正在显示 包含 76 行增加 和 141 行删除
theano/sandbox/cuda/basic_ops.py
浏览文件 @ e4242a33
......@@ -2942,181 +2942,116 @@ class GpuJoin(tensor.Join, GpuOp):
out[0] = rval
def c_code(self, node, name, inputs, out_, sub):
if node.inputs[0].data not in [0, 1]:
raise NotImplementedError()
# only works for the first two axis
if len(inputs) != 3:
# only works for two arrays
raise NotImplementedError()
if any([i.ndim != 2 for i in node.inputs[1:]]):
# only works for type T.matrix
raise NotImplementedError()
axis = inputs[0]
n_cndas = len(inputs[1:])
input_1 = inputs[1]
input_2 = inputs[2]
axis = inputs[0]
fail = sub['fail']
out = out_[0]
# getting the shapes of all the involved tensors (input[0]+out)
str = """
int axis = PyInt_AsLong((PyObject*)%(axis)s);
int axis = PyInt_AsLong((PyObject*)%(axis)s);
int nd = CudaNdarray_NDIM(%(input_1)s);
int shape_%(input_1)s[nd];
int shape_out[nd];
int dims_array1[nd];
int errorcode;
for(int i = 0; i<nd; i+=1){
dims_array1[i] = CudaNdarray_HOST_DIMS(%(input_1)s)[i];
}
nd = CudaNdarray_NDIM(%(input_2)s);
int dims_array2[nd];
for(int i = 0; i<nd; i+=1){
dims_array2[i] = CudaNdarray_HOST_DIMS(%(input_2)s)[i];
}
int dims_out[nd];
if(axis==0)
for(int i = 0; i<nd; i+=1)
{
dims_out[0] = dims_array1[0] + dims_array2[0];
dims_out[1] = dims_array1[1];
shape_%(input_1)s[i] = CudaNdarray_HOST_DIMS(%(input_1)s)[i];
shape_out[i] = shape_%(input_1)s[i];
}
if(axis==1)
""" % locals()
# getting the shapes of all the involved tensors (input[1:])
# + check: all input tensors have same shape as final out
# execept for "axis" dimension
for i, cdna in enumerate(inputs[2:]):
str += """
nd = CudaNdarray_NDIM(%(cdna)s);
int shape_%(cdna)s[nd];
for(int i = 0; i<nd; i+=1)
{
dims_out[0] = dims_array1[0];
dims_out[1] = dims_array1[1] + dims_array2[1];
shape_%(cdna)s[i] = CudaNdarray_HOST_DIMS(%(cdna)s)[i];
if((i!=axis) && (shape_%(cdna)s[i]!=shape_out[i]))
{
//(fail)s; //deactivated, because this causes segfault
}
}
if (CudaNdarray_prep_output(& %(out)s, 2, dims_out))
""" % locals()
# computing the new shape for the out tensors
str += """
int width_sum = 0;\n""" % locals()
for i, cdna in enumerate(inputs[1:]):
str += "\t\twidth_sum += CudaNdarray_HOST_DIMS(%(cdna)s)[axis];\n" % locals()
str += "\t\tshape_out[axis] = width_sum;\n"
str += """
if (CudaNdarray_prep_output(&%(out)s, nd, shape_out))
{
%(fail)s;
}
PyObject *slice;
PyObject *out_sub;
PyObject *start, *stop, *step;
step = NULL;
if(axis==0)
{
start = PyInt_FromLong(0);
stop = PyInt_FromLong(dims_array1[0]);
slice = PySlice_New(start, stop, step);
out_sub = CudaNdarray_Subscript((PyObject*)%(out)s, slice);
errorcode = CudaNdarray_CopyFromCudaNdarray((CudaNdarray*)out_sub, %(input_1)s);
if((slice == NULL) || (out_sub == NULL) || (errorcode != 0))
{
Py_XDECREF(slice);
Py_XDECREF(out_sub);
Py_XDECREF(start);
Py_XDECREF(stop);
Py_XDECREF(step);
Py_XDECREF(%(out)s);
%(fail)s;
}
Py_XDECREF(start);
Py_XDECREF(slice);
Py_XDECREF(out_sub);
int errorcode;
int sum;
sum =0;
start = stop;
stop = PyInt_FromLong(PyInt_AsLong(start) + dims_array2[0]);
slice = PySlice_New(start, stop, step);
out_sub = CudaNdarray_Subscript((PyObject*)%(out)s, slice);
errorcode = CudaNdarray_CopyFromCudaNdarray((CudaNdarray*)out_sub, %(input_2)s);
if((slice == NULL) || (out_sub == NULL) || (errorcode != 0))
{
Py_XDECREF(slice);
Py_XDECREF(out_sub);
Py_XDECREF(start);
Py_XDECREF(stop);
Py_XDECREF(step);
Py_XDECREF(%(out)s);
%(fail)s;
}
PyObject *slice_tuple;
PyObject *full_slice;
PyObject *section_slice;
Py_XDECREF(slice);
Py_XDECREF(out_sub);
Py_XDECREF(start);
Py_XDECREF(stop);
Py_XDECREF(step);
}
""" % locals()
if(axis==1)
# start copying the data into the new out tensors
for i, cdna in enumerate(inputs[1:]):
str += """
sum += shape_%(cdna)s[axis];
stop = PyInt_FromLong(sum);
slice_tuple = PyTuple_New(nd);
full_slice = PySlice_New(NULL, NULL, NULL);
section_slice = PySlice_New(start, stop, step);
for(int i=0; i<nd; i++)
{
PyObject *slice_tuple;
PyObject *full_slice;
PyObject *section_slice;
PyObject *start_axis2, *stop_axis2;
start = PyInt_FromLong(0);
stop = PyInt_FromLong(dims_out[0]);
stop_axis2 = PyInt_FromLong(dims_array1[1]);
slice_tuple = PyTuple_New(2);
full_slice = PySlice_New(start, stop, step);
section_slice = PySlice_New(start, stop_axis2, step);
PyTuple_SetItem(slice_tuple, 0, full_slice);
PyTuple_SetItem(slice_tuple, 1, section_slice);
out_sub = CudaNdarray_Subscript((PyObject*)%(out)s, slice_tuple);
errorcode = CudaNdarray_CopyFromCudaNdarray((CudaNdarray*)out_sub, %(input_1)s);
if((full_slice == NULL) || (section_slice == NULL) || (out_sub == NULL) || (errorcode != 0))
if(i!=axis)
{
Py_XDECREF(full_slice);
Py_XDECREF(section_slice);
Py_XDECREF(slice_tuple);
Py_XDECREF(out_sub);
Py_XDECREF(start);
Py_XDECREF(stop);
Py_XDECREF(step);
Py_XDECREF(start_axis2);
Py_XDECREF(stop_axis2);
Py_XDECREF(%(out)s);
%(fail)s;
Py_INCREF(full_slice);
PyTuple_SetItem(slice_tuple, i, full_slice);
}
Py_XDECREF(stop);
Py_XDECREF(full_slice);
Py_XDECREF(section_slice);
Py_XDECREF(out_sub);
Py_XDECREF(slice_tuple);
start_axis2 = stop_axis2;
stop = PyInt_FromLong(dims_out[0]);
stop_axis2 = PyInt_FromLong(dims_array2[1] + dims_array1[1]);
slice_tuple = PyTuple_New(2);
full_slice = PySlice_New(start, stop, step);
section_slice = PySlice_New(start_axis2, stop_axis2, step);
PyTuple_SetItem(slice_tuple, 0, full_slice);
PyTuple_SetItem(slice_tuple, 1, section_slice);
out_sub = CudaNdarray_Subscript((PyObject*)%(out)s, slice_tuple);
errorcode = CudaNdarray_CopyFromCudaNdarray((CudaNdarray*)out_sub, %(input_2)s);
if((full_slice == NULL) || (section_slice == NULL) || (out_sub == NULL) || (errorcode != 0))
else if(i==axis)
{
Py_XDECREF(full_slice);
Py_XDECREF(section_slice);
Py_XDECREF(slice_tuple);
Py_XDECREF(out_sub);
Py_XDECREF(start);
Py_XDECREF(stop);
Py_XDECREF(step);
Py_XDECREF(start_axis2);
Py_XDECREF(stop_axis2);
Py_XDECREF(%(out)s);
%(fail)s;
Py_INCREF(section_slice);
PyTuple_SetItem(slice_tuple, i, section_slice);
}
}
out_sub = CudaNdarray_Subscript((PyObject*)%(out)s, slice_tuple);
errorcode = CudaNdarray_CopyFromCudaNdarray((CudaNdarray*)out_sub, %(cdna)s);
if((full_slice == NULL) || (section_slice == NULL) || (out_sub == NULL) || (errorcode != 0))
{
Py_XDECREF(full_slice);
Py_XDECREF(section_slice);
Py_XDECREF(slice_tuple);
Py_XDECREF(out_sub);
Py_XDECREF(start);
Py_XDECREF(stop);
Py_XDECREF(step);
Py_XDECREF(start_axis2);
Py_XDECREF(stop_axis2);
Py_XDECREF(%(out)s);
%(fail)s;
}
"""% locals()
Py_XDECREF(full_slice);
Py_XDECREF(section_slice);
Py_XDECREF(out_sub);
Py_XDECREF(slice_tuple);
start = stop;
""" % locals()
str+="""
Py_XDECREF(start);
Py_XDECREF(stop);
Py_XDECREF(step);"""
return str
gpu_join = GpuJoin()
......
Markdown 格式
0% 或
您添加了 0 人 到此讨论。请谨慎行事。
请先完成此评论的编辑！
请 注册 或者 后发表评论