diff options
| author | Rich Felker <dalias@aerifal.cx> | 2012-03-15 19:56:36 -0400 |
|---|---|---|
| committer | Rich Felker <dalias@aerifal.cx> | 2012-03-15 19:56:36 -0400 |
| commit | 9d507419db1289033c59da6976f8675f24ab873f (patch) | |
| tree | 622fe4942ed4255c860c8190c9a445b60f53febd | |
| parent | simplify nan check in sqrt (x86 asm); result of sqrt is never negative (diff) | |
| download | grovel-9d507419db1289033c59da6976f8675f24ab873f.tar.gz grovel-9d507419db1289033c59da6976f8675f24ab873f.tar.xz | |
remove special nan handling from x86 sqrt asm
a double precision nan, when converted to extended (80-bit) precision,
will never end in 0x400, since the corresponding bits do not exist in
the original double precision value. thus there's no need to waste
time and code size on this check.
| -rw-r--r-- | src/math/i386/sqrt.s | 3 |
1 files changed, 0 insertions, 3 deletions
diff --git a/src/math/i386/sqrt.s b/src/math/i386/sqrt.s index 1bd3b05e..8289d094 100644 --- a/src/math/i386/sqrt.s +++ b/src/math/i386/sqrt.s @@ -10,9 +10,6 @@ sqrt: fldl 4(%esp) and $0x7ff,%ecx cmp $0x400,%ecx jnz 1f - mov 8(%esp),%cx - inc %cx - jo 1f and $0x200,%eax sub $0x100,%eax sub %eax,(%esp) |