TPC-H is not a good benchmark

>Its funny how when IBM shows a significantly higher benchmark result, like on SPEC
>CPU2006 or even SAP SD 2-Tier, certain folks here say *wow*, this is an amazing
>CPU

Yes, because SPECcpu2006 is the standard for CPU performance and SAP SD 2-tier is probably the best server benchmark out there (real application, not under control of a HW vendor, many results, etc.).

>and dont even question the fact that maybe IBM has cracked the benchmark yet
>here, SPARC T4 does incredibly well on TPC-H, an >independently audited benchmark,

Independent auditing is meaningless for TPC-H. Some components of SPECcpu2006 are 'cracked', but you can just ignore those and look at GCC if you want. Or you can assume that everyone will crack them reasonably well and that it all nets out at the end.

>much more real world benchmark (than spec cpu will ever >be), and where IBM and HP
>have recently published results so therefore Oracle must >have cracked the benchmark?

TPC-H is about as 'real world' as the illegitimate love child of the Easter Bunny, Santa Claus and the Wicked Witch of the West. The queries are 20 years out of date, the data set is ridiculous.

For data warehousing, TPC-H is utterly unreflective of modern workloads.

David