> adjusted tolerance for the test from 1e-10 to 1e-9 The opinions herein are my own, not necessarily SPEC's. I contributed to the development of benchmark 549.fotonik3d_r. When using -Ofast and vectors, which is a known limitation if I'm correct. So there's probably nothing we can do right now? I'm removing wrong-code as it's about floating point precision
90:319:0: note: Profitability threshold is 4 loop iterations. 90:319:0: optimized: loop vectorized using 32 byte vectors
***dbgcnt: upper limit 5 reached for vect_loop.*** 90:319:0: note: Static estimate profitability threshold = 9 90:319:0: note: Runtime profitability threshold = 4 I was able to track that to a single vectorization that happens here:ģ13 freqstep = (tmppower%freqlast-tmppower%freqfirst) &ģ14 / real(tmppower%nofreq - 1, kind=rfp)ģ19 do ifreq = 1, tmppower%nofreq tmppower%next fdbg-cnt=vect_loop:0 -fdbg-cnt=vect_loop:4:5:power.fppized
0 kommentar(er)
