Mark
|
a99a214589
|
Fix #5849 - broken snapvector inverts player speed
Snapvector was converting floats to 16 bit integers instead of 32 bit integers.
|
2013-01-09 23:15:21 -06:00 |
|
Tim Angus
|
4ccd548512
|
* I zigged when I should have zagged
|
2011-10-28 18:51:31 +00:00 |
|
Tim Angus
|
675e7a641a
|
* clang support
|
2011-10-21 22:48:53 +00:00 |
|
Thilo Schulz
|
8a500d71da
|
Set default rounding mode to FE_NEAREST again. Thanks to Matthias Bentrup for providing some explanations.
|
2011-09-27 14:43:20 +00:00 |
|
Thilo Schulz
|
c927fab58f
|
Implement Mathias Benthrup's suggestion for x86 ASM snapvector implementation which reduces cache misses.
|
2011-09-19 18:30:24 +00:00 |
|
Thilo Schulz
|
664f8e578d
|
- align sse control word storage space to 16 byte boundary for snapvector
- replace some whitespace with tabs in snapvector.c
- Give gcc a bit more freedom in choice of registers
|
2011-06-22 14:36:11 +00:00 |
|
Thilo Schulz
|
206edd3419
|
- Still forgot qasm-inline.h
- Leave GCC more freedom in choice of registers
|
2011-06-13 11:07:13 +00:00 |
|
Thilo Schulz
|
04821a529d
|
Forgot the inline assembler parts, here they come! Using inline asm to get rid of all the boilerplate stuff
|
2011-06-13 10:54:37 +00:00 |
|