< prev index next >

src/cpu/ppc/vm/macroAssembler_ppc.hpp

Print this page
rev 13389 : PPC: Implement MulAdd and SquareToLen intrinsics

This implementation is based on the algorithm implemented in java. It
yields a performance speedup of:
JDK8 - 23%
JDK9 - 5%
JDK10 - 5%

*** 810,819 **** --- 810,821 ---- void multiply_128_x_128_loop(Register x_xstart, Register y, Register z, Register yz_idx, Register idx, Register carry, Register product_high, Register product, Register carry2, Register tmp); + void muladd(Register out, Register in, Register offset, Register len, Register k, + Register tmp1, Register tmp2, Register carry); void multiply_to_len(Register x, Register xlen, Register y, Register ylen, Register z, Register zlen, Register tmp1, Register tmp2, Register tmp3, Register tmp4, Register tmp5, Register tmp6, Register tmp7, Register tmp8, Register tmp9, Register tmp10,
< prev index next >