No, not about TPC-C. Hardware firms neither. The last hurrah for it was the fight between Itanium and PowerPC and then everybody stopped caring. It devolved int

No, not about TPC-C. Hardware firms neither. The last hurrah for it was the fight between Itanium and PowerPC and then everybody stopped caring. It devolved into a benchmark where you'd get another % performance for every cache miss you saved from the SCSI / block IO request stack. It was totally IOPs bound, so they loaded up these huge machines with literally tens of thousands of hard disks that cost millions to run, and come out with some number that really wasn't very relevant. Actually I heard toward the end even the smaller 4 socket machines were driving tens of thousands of HDDs, so who knows what the 16/32 socket ones required. I think PCI attached NAND promised to (and did) vastly reduce the cost of the hardware, because the test is pretty well IOPs-bound, but they were still very expensive, and interest had already been waning, so it didn't turn that around.