This import seems to be required on some versions of numpy.

Update link

Typo in try_compile_tmp in op.py

Set Python exception when C code fails

Fixed test values with ``ifelse``

By using ``__call__`` instead of ``make_node``, test values are properly
set.

This makes it possible to obtain the Op's outputs as a list without
having to go through ``make_node(..).outputs``.

REBASED: C code for GPUAdvancedIncSubtensor1.Handles one case to make it faster

add c_code_cache_version to GpuAdvancedIncSubtensor1 as it now have c_co...

No warning if g++ does not support -march=native

Instead of detecting llvm, we try to use -march=native, which is more
general.

Fix C code of GpuDownsampleFactorMaxGrad.

Use safe_new(x) instead of x.type() in scan grad.

Simpler graphs for canonical slices

This copies the test_value, making possible to use a test value
for sequences. Add test.

Moving an assert out of for loop (perf optimization) and adding back a declartion of a CudaNdarray_inplace_add

Compute hash on whole source file

Fix insertion of -arch=... when compiling with nvcc

Minor fixes

Fixes gh-1306

It used to be inserted only for the first Op after GPU initialization.

It was causing errors when running with cuda-memcheck, or when running
on Mac OS X.