Store GL_FLOAT FlatInputs as fp32, not fp16.
authorSteinar H. Gunderson <sgunderson@bigfoot.com>
Sun, 19 Jan 2014 19:42:22 +0000 (20:42 +0100)
committerSteinar H. Gunderson <sgunderson@bigfoot.com>
Sun, 19 Jan 2014 19:42:22 +0000 (20:42 +0100)
commit68fd0c4ec98ed00d3ddeb87cfd809372a8b01e4b
treed0eee6e2ed1e66747fe1cde4debf793ee5c51d9d
parenta9bd8dc1e30b82bc99a445f80df09488b4f9d1f4
Store GL_FLOAT FlatInputs as fp32, not fp16.

There are two primary reasons for this:

 - GL_FLOAT FlatInput is primarily used for tests, and even more importantly,
   mostly accuracy tests. ATI's drivers appear to round off fp32 -> fp16
   wrong (truncate instead of round), which breaks some of these tests.
 - In case someone _would_ use GL_FLOAT inputs, they'd probably be updated
   every frame anyway, so the fp32 -> fp16 conversion step (probably on CPU)
   will negate any performance benefits by fp16 sampling anyway.
flat_input.cpp