mmizzle9 / cudpp

Automatically exported from code.google.com/p/cudpp
Other
0 stars 0 forks source link

Add short/ushort support #112

Open GoogleCodeExporter opened 8 years ago

GoogleCodeExporter commented 8 years ago
Currently, cudpp supports char/uchar and int/uint, but not short/ushort.

The attached patch (vs. r285) adds short/ushort support for scan and reduce.
Tested only on Win7 x64, VS 2010, CUDA 4.0, on Fermi hardware. (I don't see why 
it wouldn't work everywhere else just as well, but who knows...)

Original issue reported on code.google.com by m0b...@gmail.com on 29 Feb 2012 at 4:00

Attachments:

GoogleCodeExporter commented 8 years ago
This is awesome, thanks for the patch. 

I would imagine this will work out of the box (I've got some compiling 
issues on my side, so I'll have to track that down). 

My concern, though, is compile time, which (as everyone on this list 
knows) is really long. Adding 2 (possibly 4, if we do char) new 
datatypes to a scan compile is expensive. 

Strawman for discussion: Add this support and wrap it with a #define 
that cmake could flag if a user wanted it. (I just don't see a lot of 
people wanting to scan shorts or chars.) 

Original comment by jow...@gmail.com on 29 Feb 2012 at 7:27

GoogleCodeExporter commented 8 years ago
I think all datatypes, operators, and algorithms should be wrapped somehow for 
compile-time enable/disable.  Adding CMAKE settings for them would be ideal.  A 
core set of datatypes should be enabled by default.

Original comment by harr...@gmail.com on 29 Feb 2012 at 11:47