Searched hist:152596 (Results 1 - 4 of 4) sorted by relevance
/freebsd-10.0-release/lib/msun/src/ | ||
H A D | s_tanf.c | diff 152596 Sat Nov 19 00:38:27 MST 2005 bde Moved all the optimizations for |x| <= 9pi/2 from __ieee754_rem_pio2f() to its 3 callers and manually inline them. On Athlons, with favourable compiler flags and optimizations and favourable pipeline conditions, this gives a speedup of 30-40 cycles for cosf(), sinf() and tanf() on the range pi/4 < |x| <= 9pi/4, so thes functions are now signifcantly faster than the hardware trig functions in many cases. E.g., in a benchmark with uniformly distributed x in [-2pi, 2pi], A64 hardware fcos took 72-129 cycles and cosf() took 37-55 cycles. Out-of-order execution is needed to get both of these times. The optimizations in this commit apparently work more by removing 1 serialization point than by reducing latency. |
H A D | s_cosf.c | diff 152596 Sat Nov 19 00:38:27 MST 2005 bde Moved all the optimizations for |x| <= 9pi/2 from __ieee754_rem_pio2f() to its 3 callers and manually inline them. On Athlons, with favourable compiler flags and optimizations and favourable pipeline conditions, this gives a speedup of 30-40 cycles for cosf(), sinf() and tanf() on the range pi/4 < |x| <= 9pi/4, so thes functions are now signifcantly faster than the hardware trig functions in many cases. E.g., in a benchmark with uniformly distributed x in [-2pi, 2pi], A64 hardware fcos took 72-129 cycles and cosf() took 37-55 cycles. Out-of-order execution is needed to get both of these times. The optimizations in this commit apparently work more by removing 1 serialization point than by reducing latency. |
H A D | s_sinf.c | diff 152596 Sat Nov 19 00:38:27 MST 2005 bde Moved all the optimizations for |x| <= 9pi/2 from __ieee754_rem_pio2f() to its 3 callers and manually inline them. On Athlons, with favourable compiler flags and optimizations and favourable pipeline conditions, this gives a speedup of 30-40 cycles for cosf(), sinf() and tanf() on the range pi/4 < |x| <= 9pi/4, so thes functions are now signifcantly faster than the hardware trig functions in many cases. E.g., in a benchmark with uniformly distributed x in [-2pi, 2pi], A64 hardware fcos took 72-129 cycles and cosf() took 37-55 cycles. Out-of-order execution is needed to get both of these times. The optimizations in this commit apparently work more by removing 1 serialization point than by reducing latency. |
H A D | e_rem_pio2f.c | diff 152596 Sat Nov 19 00:38:27 MST 2005 bde Moved all the optimizations for |x| <= 9pi/2 from __ieee754_rem_pio2f() to its 3 callers and manually inline them. On Athlons, with favourable compiler flags and optimizations and favourable pipeline conditions, this gives a speedup of 30-40 cycles for cosf(), sinf() and tanf() on the range pi/4 < |x| <= 9pi/4, so thes functions are now signifcantly faster than the hardware trig functions in many cases. E.g., in a benchmark with uniformly distributed x in [-2pi, 2pi], A64 hardware fcos took 72-129 cycles and cosf() took 37-55 cycles. Out-of-order execution is needed to get both of these times. The optimizations in this commit apparently work more by removing 1 serialization point than by reducing latency. |
Completed in 142 milliseconds