Check it out. It's small and the report itself is funny.
http://swox.com/doc/x86-timing.pdf


There are several reasons for presenting this report:
1. Intel's published data is incomplete and full of errors.
2. Intel does not publish any data for 64-bit operations.
3. To allow straightforward comparison of AMD and Intel pipelines.
The here presented data is the result of extensive timing tests performed at Swox. While we have made an e ort to make sure the data is accurate, the reader is cautioned that some errors might have crept in. The data is surely less riddled by errors than Intel's manuals, though.
But isn't that their usual policy?

The Pentium 4 performance for 64-bit right shifts is really poor. 64-bit left shift as well as all 32-bit shift have acceptable performance. Right shift is an important operation, which makes Intel's decision to let it run slow, surprising.
Maybe their chip designers are left-handed?