Browse Prior Art Database

Virtual Processor Index (VPI)

IP.com Disclosure Number: IPCOM000147802D
Original Publication Date: 2007-Mar-27
Included in the Prior Art Database: 2007-Mar-27
Document File: 1 page(s) / 27K

Publishing Venue

IBM

Abstract

Equating unalike Processor Technologies utilizing a single performance index metric. Presently most people use MHz and Processor Utilization, neither of which provides proper sizing information. In general, all comparable processor performance and sizing analysis is done on best-guess or processor utilization analysis and is critically flawed in that one processor architecture can perform significantly better or worse than another while both showing 100%, 50%, or any other non-relative utilization metric. The Virtual Processor Index allows for a simple method of comparing approximate performance of integrated, commercial Server Applications (database, mail, web, messaging, etc.) as well as Business-oriented Desktop Applications (MS Office, Lotus, Web browsing, Web applications, etc.)

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 66% of the total text.

Page 1 of 1

Virtual Processor Index (VPI)

The base method of the VPI is presently a simple equation. (80% of integer performance as expressed by SpecCPU2000_Int + 20% of floating point performance as expressed by SpecCPU2000_FP) multiplied by the number of processor cores. SpecInt and SpecFP are ideal as they are single-threaded and will express the processor's integer and FP units in directly comparable numbers to other processor core architectures. The benchmarks are affected by memory and chipset differences, so system+processor comparison is sometimes more accurate than merely processor to processor comparison. This can be accounted for during the expression of the relative performance at the time of system sizing and comparison. SpecCPU_Flow rates are not applicable as they are multi-threaded and will not transcribe in a one-to-one scenario their data from one system processor architecture to another. The point of the VPI is to be able to compare the equivalent linear scaling of one system processor architecture to another. The greater purpose is to be able to evaluate a system workload for system consolidation assuming the use of a 95%+ linear scaling consolidation solution, e.g. VMware, Xen, and to some degree, MS Virtual Server.

VPI came about as an answer to the need to be able to size Desktop Application workloads for consolidation on to servers for our Virtual Client Solution (VCS). It has become apparent that VPI is applic...