"So Ali, what's up with that? Go ahead, we're waiting..."
There is nothing up. I only can repeat my argument, which you continue to ignore and continue to substitute wrong metrics. When I want to compile a big project, be it software or firmware or FPGA, I need my [single] job done as fast as possible, to continue to debug it and move on adding more design features. Running four copies of the same compile does not help me nor anyone else. That's why people smarter than you have developed several benchmarks, for different kind of activities. You continue to use wrong benchmark, and seem to be very proud of this.
"the 2.8GHz Opteron 2220 is FASTER, clock for clock, than the 3GHz Xeon 5160 in SPECint2006-rate, and just 10% slower, clock for clock, than the Xeon in SPECint2006-base"
What the heck your "clock for clock" supposed to mean? Aha, now we are stretching the reality by prorating systems by clock speeds, to please yourself. Tell me please now that performance per watt is really important too, and companies cannot afford to pay for extra 10-20 Watts of electricity per workstation when a multi-million project depends on speed of compile turnarounds...
BTW, for the sake of accuracy, Dell 690 is a Workstation, not a server, as your cited tabloid refers to it. So, you better consider more appropriate metrics than spec_rate.
Cheers,
- Ali |
