Observed a run time overflow, after getting modified old code to compile with new 0.31.0.
I saw in the blogpost, that
If the new default behavior is blocking you, use -Ddisable_overflow to
keep the same semantics as in 0.30 or -Dskip_crystal_compiler_rt to
not include out compiler-rt port and link against another build of it. The
disable_overflow flag is temporal and it will go away. Read more at #8170.
I tried that, and all I get is unknown command error.
Can you give examples of how to use the commands correctly to turn off new overflow check.
Not only did that eliminate the run time overflows, all my benchmarks run faster now than before compiling without overflow checks. The 0.31.0 version is now generally all around faster than with 0.30.x.
When I get a chance I’ll play with using multi-threading. I have a perfect numerical algorithm for that.