Multiprocessor processing element 77 Table 5-3 Protocol Processor Utilization Packet length 1024 256 64 Processor utilization 0.519 0.921 0.968 cost is equal to 30 instructions [98] for the ARM-based adapter (8.5% of the protocol processing delay). Considering all the previous architectural enhancements of the adapter, the PE is left only with the task of protocol processing, enabling it to be renamed as protocol processor. Peak utilization of the protocol processor for the analyzed packet lengths is shown in Table 5-3. Utilizations in the table, higher than those of the previous adapter configurations, show that for short packets, the protocol processor's utiliza- tion reaches almost 100%. However, utilization is relatively low for the case of long packets. This is due to several parameters, including the contention of the protocol processor with the MMU and, most importantly, with the intelligent DMA