Lines Matching refs:results
50 …bandwidth. Different cores may be more closely connected to different memory units. This results in
80 …values may be crucial to high performance; the defaults may produce poor results. The best binding…
127 The Fortran routine `cpu_time()` can sometimes produce misleading results when run with multiple th…
160 Requesting the `spread` binding produces better results for small core counts but poorer ones for l…
197 …c we reran it with the highest optimization level that produced the same results as gcc and icx: `…
198 The results are displayed in Fig. {any}`fig_icc_O1_streams`; sure enough, the results now match tha…
207 Next we display the STREAMS results using gcc with parallel efficiency instead of speedup in {any}`…
218 - For MPI, the default binding and mapping on this system produces results that are as good as prov…
225 Fig. {any}`fig_m2_gcc_streams` provides the results. Based on the plateau in the middle of the plot…
235 OpenMPI (installed via Homebrew) produced similar results.
246 compilers so we only display results for gcc.
263 and results from the amount of MPI communication required from the communication pattern which resu…
267 Fig. {any}`fig_m2_gamg` provides the results. The performance is better than predicted by the STREA…
281 All compilers deliver largely the same performance so we only present results with gcc.
323 For the Apple M2, we present the results using Unix shared-memory communication of the matrix and v…