14 DAY TRIAL // SPEC (Standard Performance Evaluation Corporation) has introduced a new standard for power-efficiency measurements regarding CPU architectures. This comes as a response to AMD's K10 TDP inconsistencies that have been identified in a set of whitepapers on the producer's website.
The SPEC corporation issued a new set of benchmarks based on independent criteria, that can safely be used to compare CPU efficiency. This new and unbiased standard should offer CPU producers the opportunity of testing their products' quality suing the same reference.
The test has been named SPECpower_ssj2008 and runs on a server. The application would take the processor through all the workload stages, starting from full load, and then decreasing ten percent per timing interval, until the idle stage. Each loading stage is then ranked, the results are summed, then divided by the sum of the average power consumed for each loading state. The end result is called "ssj_opps/Watt."
The new test brings up an interesting approach. Many times, the workload variation is omitted in other benchmarks. The SPEC test calculates the final scores using an averaging approach. The actual performance measuring stage is divided between AMD's new ACP standard and Intel's modified TDP, which allows for no common reference in benchmarking. SPEC's standard promises to deliver accurate and comprehensive comparative tests regarding the CPU performance. This would also shed some light regarding an unbiased power performance test between the AMD and Intel processors and show who the boss really is.
The cooling solution pick can be done via the older TDP procedure, since it is assumed that the processor is likely to be working at full load for large amounts of time. But when it comes to power-efficiency figures, the SPEC test measures it by applying a wide pattern of usage; rather unsuitable for cooling solution evaluation. The new specifications are alleged to bring up honest figures to boost the competition on the CPU stage.