ptxas demoting "double" in reduce app to "float" on OS X

[GoogleCodeExporter]

CUDPP build on OS X:

nanaimo-bar 9381$ nvcc --version
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2009 NVIDIA Corporation
Built on Thu_Mar_11_12:08:00_PST_2010
Cuda compilation tools, release 3.0, V0.2.1221

CUDA Driver Version 4.0.17
GPU Driver Version 1.6.26.31 (256.00.35f05)

nanaimo-bar 9381$ make
ptxas /tmp/tmpxft_0001369b_00000000-5_reduce_app.compute_10.ptx, line 21877; 
warning : Double is not supported. Demoting to float

Original issue reported on code.google.com by jow...@gmail.com on 20 Jun 2011 at 10:45

[GoogleCodeExporter]

This is only reported when compiling the code for pre SM_13 GPUs.  It doesn't 
get demoted when compiling for SM_13 and higher.  We have to compile for SM_10 
and higher, so we are always going to get this error unless we change the code 
compiled for different GPUs, which is excessive, IMO.  If you want, you can try 
to make this warning go away, but I don't think it's very high priority.

Changing to enhancement and reducing priority.

Original comment by harr...@gmail.com on 20 Jun 2011 at 11:01

• Added labels: OpSys-All, Priority-Low, Type-Enhancement
• Removed labels: Priority-Medium, Type-Defect

[GoogleCodeExporter]

It would be tricky to get rid of these warnings, and they are not causing 
incorrect function, so closing.

Original comment by harr...@gmail.com on 6 Jul 2011 at 2:37

• Changed state: WontFix