Re: ,i>I guess it depends on which benchmark you choose to highlight.
I should have specified the - basic? Traditional? Original? Int and Float SPEC scores as the ones I was referring to.
Opteron generally does very well in apples to apples database tests, but database spec tests are also tests of the disk subsystem, etc. So a test comparing, say, an Opteron equipped with a 400 drive fiber channel disk array to, say, an Itanium wth a 12 drive SCSI array would not be fair to the Itanium and would not provide an accurate comparison of the two platforms.
TPC-C results have to be looked at carefully. For example, comparing two systems from the same company, both running windows and SQL Server Enterprise, an HP DL-585 (Opteron 4-way) does 187,296 in TPC-C at $2.04 per unit of performance. http://www.tpc.org/tpcc/results/tpcc_result_detail.asp?id=105042002 while an Integrity 5670 (Itanium 4-way) does 121,065 in TPC-C at $4.49 per unit of performance. http://www.tpc.org/tpcc/results/tpcc_result_detail.asp?id=103090501 And that's with the Opteron running 32-bit Windows and Database, while the Itanium is running 64-bit. Supposedly, running the Opteron as a 64-bit will speed it up. We'll have to wait and see. But that's also a test of the latest Opteron vs. Itanium's with only 3 times as much cache.
So here's a test of the very latest Itanium, with 4 1/2 times as much cache, running a OS and database that always gives better scores, all else being equal.