<div> </div><div>Hi All,</div><div> </div><div>I am facing miscompare error when running povray (and few other C/C++ benchmarks) from spec cpu2006 suite enabling FMA4 (and disabling FMA3). I have used -ffp-contract=fast to turn on this option. (Compilation options and targets pasted below).</div>
<div> </div><div>>>>>>>>></div><div>clang version 3.2 (trunk 163295:163308) (llvm/trunk 163295)<br>Target: x86_64-unknown-linux-gnu<br>Thread model: posix</div><div> </div><div> </div><div>(Options to clang)</div>
<div> </div><div> -O3 -march=bdver2 -mavx -mno-fma -mfma4 -ffp-contract=fast -save-temps </div><div> </div><div><<<<<<<</div><div> </div><div>Note that BDVER2 supports both FMA3 and FMA4. Also the benchmark was run *successfully* when FMA3 was enabled. Reducing the testcase might take more time but has anyone noticed this issue?</div>
<div> </div><div>For those interested, miscompare message is as below</div><div> </div><div>*** Miscompare of SPEC-benchmark.log; for details see<br> CPU2006/peak_ref_llvm.0037/SPEC-benchmark.log.mis</div><div> </div>
<div>
0050: Pixels: 1280 Samples: 8960<br> Pixels: 1280 Samples: 47360<br> ^<br>0051: Rays: 8960 Saved: 0 Max Level: 1/5<br>
Rays: 47360 Saved: 0 Max Level: 1/5<br> ^<br>0055: CSG Intersection 232960 0<br> Box 47360 0<br>
^<br>0056: Plane 537600 510304<br> Cone/Cylinder 1515520 0<br> ^<br>...</div><div> </div><div> </div><div> </div><div> </div>
<div> </div><div><br clear="all"><br>-- <br><i style="font-family:times new roman,serif"><b> Anitha</b></i><br><br>
</div>