probably would of been helpful to have a reference model. Anyhow looking up these tests on open benchmark, I am not sure if they use the same metric but kaveri scores ~ 3 in test 1 the i3's/pentiums are above 20 and the i5 ~45. That looks broke. The other tests are as equally skewed? Could really use a reference system for clarification on what's up with this.
Edit: I can't use the OBM site (impossible to navigable intelligibly for me), so maybe I am doing it wrong... who knows.
Edit 2: yeah your results are wrong on openbench sit the kaveri scores >20 in test 1 blah blah you got the wrong values for the test names.
Last edited by nightmarex; 01-17-2014 at 07:52 AM.
but kaveri scores ~ 3 in test 1 the i3's/pentiums are above 20 and the i5 ~45. That looks broke. The other tests are as equally skewed?
Yes, this result is incorrect (or correct, but have nothing to do with reality). Compare with the result of E1-1200 APU. They have approximately the same value for the G-HPL. See http://openbenchmarking.org/result/1...PL-HPCCHALLE83.
I have a similar trouble with defaut settings of ACML (64bit, fma; OMP_NUM_THREADS=6) for FX-6300 and LU-decomposition (85-90% CPU-time of whole work):