[SYMBOL] [A] [B] [C] [D] [E] [F] [G] [H] [I] [J] [K] [L] [M] [N] [O] [P] [Q] [R] [S] [T] [U] [V] [W] [X] [Z] P6 processor (Intel) packet-switched protocols packets 2nd collisions flow control fragmentation MSS (maximum segment size ) payload-free size distribution time-to-live page scanner paging compared to swapping disk problems and free list page coloring swap space anonymous memory sizing virtual memory parallel member write policies (disk mirroring) parameters caches, changing Ethernet, changing maxcontig maxusers kernel tables minfree pt_cnt rotational delay parity context (disk striping) parity-protected striping parity-protected striping with independent disks pattern analysis, networks data transfer patterns inverse request-response message passing patterns packet size distribution request-response payload-free packets PCI (Peripheral Component Interconnect) bus architecture CompactPCI PCI Special Interest Group web site PCI-to-PCI bridge perf file, sar command, enabling performance accounting sar command SE Toolkit software performance analysis tools performance instrumentation counters (PICs) performance tuning definition dynamic performance analysis environment, importance of Heisenbergs Principle of Uncertainty I/O subsystem, importance of latency problems with, overview of proxy servers resource utilization single- user workstations filesystems kernel swap space static factors swap space test design throughput and latency considerations trade-offs and web servers disks filesystems kernel memory network sizing swap space workgroup servers disks filesystems kernel memory NFS swap space Peripheral Component Interconnect bus [See PCI bus] peripherals interconnect performance monitoring tools PCI bus architecture SBus architecture persistent superblocks (Linux md package) physical address physical layer (OSI model) cables and connectors interfaces PICs (performance instrumentation counters) PIO (programmed I/O) pipes, sticky bit piplining conditional branches platters plugs pmap command, estimating memory requirements with point-to-point Fibre Channel topology political considerations, workload management PowerPC processors, memory page size preempting threads prefetching , application optimization and presentation layer (OSI model) prex command I/O tracing restarting priocntl command, process priorities, adjusting priority paging priority, processes adjusting 2nd determining 2nd private IP addresses probe-based application analysis Solaris TNF problem solving algorithms hardware level architecture levels of transformation memory hierarchy and programming languages and /proc directory (Linux), kernel variable tuning process accounting enabling overhead records, reviewing process filtering, I/O tracing process monitoring prstat command top command process queues, viewing state process scheduling classes dispatch queues dispatch tables, tuning dynamic priority Linux priorities adjusting 2nd determining 2nd real-time scheduling classes, listing SMP systems Solaris Solaris threading model static priority processes blocked free list memory management memory estimating requirements monitoring consumption ps command uax command memory usage, displaying virtual address spaces processors [See also mulitprocessor systems; SMP] architecture, specifying with compiler options cache hierarchy cache mapping cache organization cache size anomalies cache-busters caching CISC design clock rates conditional branches CPI (cycles per instruction) critical path DEC Alpha memory page size design basics dispatch queues failure FISC design graphics support I/O wait time idle time improvements in instruction stages Intel 80x86, memory page size Intel P6 internal clock rate load average load, vmstat command locality metrics, MIPS microSPARC, memory page size MIPS, memory page size MMX extensions mpstat command optimum load considerations performance monitoring performance, importance of piplining PowerPC, memory page size process accounting process queues process scheduling psrinfo command RISC design, second generation sets, creating with psrset command SPARC LDSTUB instructions memory page size speed compared to memory speed spin locking state, changing with psradm command Sun E-series Sun I-series superpiplined superscalar 2nd SuperSPARC, memory page size system time system time compared to user time time command operation Transmeta Crusoe 2nd UltraSPARC, memory page size user time variable length instructions VIS extensions vmstat command von Neumann model profile feedback (compilers) profiler-based application analysis GNU gprof profiling application analysis compiling with executing with program counter programmed I/O (PIO) programming code [See also applications] algorithms Boyer-Moore searching Knuth-Morris-Pratt searching string searching analyzing ptime command time command timex command arithmetic optimization compiler optimization considerations compilers, profile feedback fast compiler option linked list cache-busting loop optimization microstate accounting, enabling non-unit stride cache-busting pitfalls poor, reasons for string optimization TICK register, enabling timing information, gethrtime system call xarch compiler option xchip compiler option xcrossfile compiler option xdepend compiler option xinlining compiler option xO compiler option xprefetch compiler option xsfpconst compiler option xvector compiler option programming languages, problem solving and promiscuous mode (network interfaces) protocol field (IP headers) protocols circuit-switched IP NLM (Network Lock Manager) packet-switched routing TCP three-way handshake UDP proxy servers, tuning prstat command, process monitoring prtconf command, synchronous transfer rate (SCSI) ps command process memory consumption process priorities, determining 2nd psradm command, processor state, changing psrinfo command processor failure monitoring processor information psrset command, processor sets, creating pt_cnt parameter ptime command application analysis disk write throttle |