RFR: 8255553: [PPC64] Introduce and use setbc and setnbc P10 instructions [v4]

[Martin Doerr]

On Wed, 4 Nov 2020 10:58:39 GMT, Martin Doerr <mdoerr at openjdk.org> wrote:

>> Ziviani has updated the pull request with a new target base due to a merge or a rebase. The incremental webrev excludes the unrelated changes brought in by the merge/rebase. The pull request contains one additional commit since the last revision:
>> 
>>   8255553: [PPC64] Introduce and use setbc and setnbc P10 instructions
>>   
>>   - setbc RT,BI: sets RT to 1 if CR(BI) is 1, otherwise 0.
>>   - setnbc RT,BI: sets RT to -1 if CR(BI) is 1, otherwise 0.
>>   Ref: PowerISA 3.1, page 129.
>>   
>>   These instructions are particularly interesting to improve the following
>>   pattern `(src1<src2)? -1: ((src1>src2)? 1: 0)`, which can be found in
>>   `instruct cmpL3_reg_reg_ExEx()@ppc.ad`, by removing its branches.
>>   
>>   Long.toString, that generate such pattern in getChars, has showed a
>>   good performance gain by using these new instructions.
>>   
>>   Example:
>>   for (int i = 0; i < 200_000; i++)
>>     res = Long.toString((long)i);
>>   
>>   java -Xcomp -XX:CompileThreshold=1 -XX:-TieredCompilation TestToString
>>   
>>   Without setbc (average): 0.1178 seconds
>>   With setbc (average): 0.0396 seconds
>
> src/hotspot/cpu/ppc/macroAssembler_ppc.inline.hpp line 255:
> 
>> 253: 
>> 254: // set dst to -1, 0, +1
>> 255: inline void MacroAssembler::set_cmpu3(Register dst) {
> 
> Shorter possiblity with only 1 additional instruction on any Power version:
> cror(CCR0, Assembler::less, CCR0, Assembler::summary_overflow); // treat unordered like less
> set_cmp3(dst);

Or even better with parameter:
inline void MacroAssembler::set_cmpu3(Register dst, bool treat_unordered_like_less) {
  if (treat_unordered_like_less) {
    cror(CCR0, Assembler::less, CCR0, Assembler::summary_overflow); // treat unordered like less
  } else {
    cror(CCR0, Assembler::greater, CCR0, Assembler::summary_overflow); // treat unordered like greater
  }
  set_cmp3(dst);
}
This allows more cleanup in interpreter and C1. (unordered_result is only +1 or -1 in TemplateTable::float_cmp which we can assert.)

-------------

PR: https://git.openjdk.java.net/jdk/pull/907