Pentium 4 vs Pentium 4E

P4E vs P4 Results

The graph below compares the performance of optimized versions of MILC asqtad code as run on the following sytems:


The Pentium 4 has an L2 cache size of 512 KB, and the Pentium 4E has an L2 cache size of 1024 KB.

The addition of software prefetch hints to codes run on the Pentium 4 processor does not improve performance. Prefetch hints on the Pentium 4E, on the other hand, do increase performance. The curve on the P4 above was taken w/out prefetching, all P4E curves were taken with prefetching. MILC V5 code with the Gottlieb/Foster field major changes, and with inline SSE optimization was used.