RFR: 8344306: RISC-V: Add zicond

[Hamlin Li]

On Tue, 26 Nov 2024 10:21:38 GMT, Hamlin Li <mli at openjdk.org> wrote:

>> Hi, please consider.
>> 
>> In cpu models we save ~1 cycle per removed branch.
>> This patch removes ~0.1% of branches in generic C2 generated code.
>> We should probably investigate if we can improve/add peephole optimization to remove more branches.
>> 
>> As the C1 cmov code is a bit tricky I left that as a followup.
>> 
>> I added gtests for the cmovs.
>> (we should add coverage for more of masm in this gtest suit)
>> Pro tip, invoke the gtestLauncher directly (you only need to build exploded):
>> `gtestLauncher -jdk build/linux-riscv64-server-fastdebug/jdk/ --gtest_break_on_failure --gtest_filter="*RiscV*" -XX:+UnlockDiagnosticVMOptions -XX:+UseZicond -XX:ParallelGCThreads=1 -XX:ConcGCThreads=1 -XX:CICompilerCount=2`
>> 
>> Tested on Spacemit X60, gtests and tier1.
>> 
>> Thanks, Robbin
>
> src/hotspot/cpu/riscv/macroAssembler_riscv.cpp line 1143:
> 
>> 1141:   bne(cmp1, cmp2, no_set);
>> 1142:   mv(dst, src);
>> 1143:   bind(no_set);
> 
> Do we have performance data comparison between `-UseZicond ` and `+ UseZicond `?

Seems the critical path when `+UseZicond` is longer than `-UseZicond`, I'm not sure how much impact the branch code will bring when `-UseZicond`.

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

PR Review Comment: https://git.openjdk.org/jdk/pull/22386#discussion_r1858216997