Graham Markall
Graham Markall
gpuci run tests
gpuci run tests
The issue with the current revision of this branch can be seen by running: ``` NUMBA_DUMP_ASSEMBLY=1 python runtests.py numba.cuda.tests.cudapy.test_compiler.TestCompileToPTX.test_global_kernel ``` where PTX looks like: ```assembly mov.u32 %r2, %tid.x; mov.u32 %r3,...
The latest commit seems to help with avoiding float64 generation. Plans for forward progress are: - Dump the type annotations for the entire test suite with and without this change,...
Summary of a little more analysis: * Mostly this has no effect on the register count, or makes it go down. * There are some functions for which the register...
I still plan to work on this.
I'm still going to get round to this one day...
I am still interested in finishing this off, when everything aligns to give me a chance of fixing it.
Still on the to-do list for one day...