Merge pull request #1845 from abergeron/gpua_rng_opt

theano/sandbox/gpuarray/basic_ops.py

@@ -78,7 +78,7 @@ class Kernel(object):
             binvar = 'kbin_' + name
         self.binvar = binvar
         if objvar is None:
-            self.objvar = 'k_' + name
+            objvar = 'k_' + name
         self.objvar = objvar
 
     @staticmethod
@@ -893,7 +893,7 @@ KERNEL void k(GLOBAL_MEM %(ctype)s *a, ga_size n, ga_size m) {
         fail = sub['fail']
         typecode = pygpu.gpuarray.dtype_to_typecode(self.dtype)
         sync = bool(config.gpuarray.sync)
-        kname = self.gpu_kernels()[0].objvar
+        kname = self.gpu_kernels(node, name)[0].objvar
         s = """
         size_t dims[2] = {0, 0};
         void *args[3];

theano/sandbox/rng_mrg.py

@@ -952,17 +952,25 @@ class GPUA_mrg_uniform(GpuKernelBase, mrg_uniform_base):
             }
         }
 
-        if (PyGpuArray_NDIM(%(o_rstate)s) != 1)
+        if (PyGpuArray_NDIM(%(o_rstate)s) != 2)
         {
-            PyErr_SetString(PyExc_ValueError, "rstate must be vector");
-            %(fail)s;
+            PyErr_SetString(PyExc_ValueError, "rstate must be a matrix");
+            %(fail)s
         }
-        if (PyGpuArray_DIMS(%(o_rstate)s)[0] %% 6)
+        if (PyGpuArray_DIMS(%(o_rstate)s)[1] != 6)
         {
-            PyErr_Format(PyExc_ValueError, "rstate len must be multiple of 6");
-            %(fail)s;
+            PyErr_Format(PyExc_ValueError, "rstate must have 6 columns");
+            %(fail)s
+        }
+        if (%(o_rstate)s->ga.typecode != GA_INT) {
+            PyErr_Format(PyExc_ValueError, "rstate must be int32");
+            %(fail)s
+        }
+        if (!GpuArray_CHKFLAGS(&%(o_rstate)s->ga, GA_C_CONTIGUOUS)) {
+            PyErr_Format(PyExc_ValueError, "rstate must be C contiguous");
+            %(fail)s
         }
-        n_streams = PyGpuArray_DIMS(%(o_rstate)s)[0]/6;
+        n_streams = PyGpuArray_DIMS(%(o_rstate)s)[0];
         if (n_streams > n_elements)
           n_streams = n_elements;
 
@@ -984,7 +992,7 @@ class GPUA_mrg_uniform(GpuKernelBase, mrg_uniform_base):
         """ % locals()
 
     def c_code_cache_version(self):
-        return (2, self.GpuKernelBase_version)
+        return (3, self.GpuKernelBase_version)
 
 
 def guess_n_streams(size, warn=True):
@@ -1341,11 +1349,26 @@ class MRG_RandomStreams(object):
         assert final_samples.dtype == dtype
         return final_samples
 
+from theano.sandbox.gpuarray.opt import (register_opt as register_gpua,
+                                         host_from_gpu as host_from_gpua)
 
+@register_gpua()
 @local_optimizer([mrg_uniform])
+def local_gpua_mrg(node):
+    if (type(node.op) == mrg_uniform and
+        isinstance(node.inputs[0].type, GpuArrayType)):
+        outs = GPUA_mrg_uniform.new(node.inputs[0],
+                                    node.op.output_type.ndim,
+                                    node.op.output_type.dtype,
+                                    node.inputs[1])
+        return [outs[0], host_from_gpua(outs[1])]
+
+
+MRG_RNGs = (mrg_uniform, GPU_mrg_uniform, GPUA_mrg_uniform)
+@local_optimizer(MRG_RNGs)
 def mrg_random_make_inplace(node):
     op = node.op
-    if isinstance(op, mrg_uniform) and not op.inplace:
+    if isinstance(op, MRG_RNGs) and not op.inplace:
         # op might be gpu version
         new_op = op.__class__(op.output_type, inplace=True)
         return new_op.make_node(*node.inputs).outputs

theano/sandbox/test_rng_mrg.py

@@ -325,7 +325,8 @@ def test_consistency_GPUA_serial():
     for i in range(n_streams):
         stream_rstate = curr_rstate.copy()
         for j in range(n_substreams):
-            substream_rstate = numpy.array(stream_rstate.copy(), dtype='int32')
+            substream_rstate = numpy.array([stream_rstate.copy()],
+                                           dtype='int32')
             # Transfer to device
             rstate = gpuarray_shared_constructor(substream_rstate)
 
@@ -380,7 +381,7 @@ def test_consistency_GPUA_parallel():
         rstate = [curr_rstate.copy()]
         for j in range(1, n_substreams):
             rstate.append(rng_mrg.ff_2p72(rstate[-1]))
-        rstate = numpy.asarray(rstate).flatten()
+        rstate = numpy.asarray(rstate)
         rstate = gpuarray_shared_constructor(rstate)
 
         new_rstate, sample = rng_mrg.GPUA_mrg_uniform.new(rstate, ndim=None,