Fix Elemwise for float16 inputs/outputs on GPU

theano/sandbox/gpuarray/elemwise.py

@@ -102,31 +102,55 @@ class GpuElemwise(HideC, Elemwise):
     def generate_kernel(self, node, nodename):
         inps = [make_argument(i, 'i%d' % (n,)) for n, i in
                 enumerate(node.inputs)]
-        scal_ins = [scalar.get_scalar_type(i.dtype) for i in node.inputs]
+        scal_v_ins = [scalar.get_scalar_type(i.dtype) for i in node.inputs]
 
         outs = [make_argument(o, 'o%d' % (n,)) for n, o in
                 enumerate(node.outputs) if not n in self.inplace_pattern]
-        scal_out = [scalar.get_scalar_type(o.dtype) for o in node.outputs]
+        scal_v_outs = [scalar.get_scalar_type(o.dtype) for o in node.outputs]
 
-        fake_node = Apply(self.scalar_op, [i() for i in scal_ins],
-                          [o() for o in scal_out])
+        fake_node = Apply(self.scalar_op, [i() for i in scal_v_ins],
+                          [o() for o in scal_v_outs])
+
+        scal_in = [i.name + '[i]' if i.dtype != 'float16' else
+                   '__half2float(' + i.name + '[i])' for i in inps]
 
         scal_out = []
         oi = 0
+        scal_f16 = []
         for n in range(len(node.outputs)):
             if n in self.inplace_pattern:
-                scal_out.append(inps[self.inplace_pattern[n]].name+'[i]')
+                arg = inps[self.inplace_pattern[n]]
             else:
-                scal_out.append(outs[oi].name+'[i]')
+                arg = outs[oi]
                 oi += 1
+            if arg.dtype == 'float16':
+                scal_f16.append(('tmpf16%i' % (len(scal_f16),), arg))
+                scal_out.append(scal_f16[-1][0])
+            else:
+                scal_out.append(arg.name + '[i]')
 
         kop = self.scalar_op.c_code(fake_node, nodename+'_scalar',
-                                    [i.name+'[i]' for i in inps],
-                                    scal_out,
+                                    scal_in, scal_out,
                                     dict(fail='return;'))
 
+        if scal_f16:
+            # if we have float16 scalars on output we have to wrap
+            # them and insert a stand-in float32 variable since
+            # float16 arithemtic is not available
+            code = ["{"]
+            for f in scal_f16:
+                code.append('ga_float %s;' % (f[0],))
+            # XXX: The replace is an ugly hack to make sure temp
+            # variables inthe middle are float32
+            code.append(kop.replace('npy_uint16', 'ga_float'))
+            for f in scal_f16:
+                code.append('%s[i] = __float2half_rn(%s);' % (f[1].name, f[0]))
+            code.append('}')
+            kop = '\n'.join(code)
+
         # Translate types for scalar composite ops (except complex).
-        # NB: OpenCL implicitly has 'stdint' defs at the kernel compilation stage
+        # NB: OpenCL implicitly has 'stdint' defs at the kernel
+        # compilation stage
         support_code = "" if pygpu.get_default_context().kind == 'opencl' else """
 #ifdef _MSC_VER
 #define signed __int8 int8_t