chapter5-m3-ziavras

9 commercial workload oltp online transaction

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: ntention) – Cache-to-cache transfer: 125 cycles These miss penalties are higher in today’s systems (superpipeling, etc.) 9 Commercial Workload • OLTP: OnLine Transaction Processing • Decision Support Systems(Database) • Search Engine (www.AltaVista.com) 10 • Resource-based pipeline stalls dominate • Combined data for data for appl. & kernel accesses • Only OLTP has major kernel accesses (they behave better than appl.) th 11 Relative performance for OLTP * True sharing misses dominate with or MB L3 cache with 4 or 8 MB L3 cache •True/false sharing misses unchanged L3 12 MP Performance 4 Processor Commercial Workload: OLTP, Decision Support (Database) Search Engine Support (Database), Search Engine • Uniprocessor cache misses improve with cache size increase (Instruction, Capacity/Conflict, Compulsory) 3.25 3 Instruction Capacity/Conflict Cold False Sharing True Sharing Sharing 2.75 (Memory) Cycles per Instruction • True sharing and false sharing unchanged going from MB...
View Full Document

This document was uploaded on 02/09/2014.

Ask a homework question - tutors are online