Day2-03_DevelopingcodeforCell-DynamicProfiling-SDK30

Day2-03_DevelopingcodeforCell-DynamicProfiling-SDK30 - IBM...

Info iconThis preview shows pages 1–8. Sign up to view the full content.

View Full Document Right Arrow Icon
Cell Programming Workshop 3/2/2008 © 2007 IBM Corporation 1 SPU Dynamic Profiling Cell Programming Workshop Cell/Quasar Ecosystem & Solutions Enablement
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
© 2007 IBM Corporation Cell Programming Workshop 3/2/2008 2 Course Objectives ! To familiarize with the SPU dynamic profiling capability provided by the IBM full system simulator Trademarks - Cell Broadband Engine and Cell Broadband Engine Architecture are trademarks of Sony Computer Entertainment, Inc.
Background image of page 2
© 2007 IBM Corporation Cell Programming Workshop 3/2/2008 3 Course Agenda ! SPU performance model ! SPU pipeline statistics ! SPE visualizer ! Local Store stats
Background image of page 3

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
© 2007 IBM Corporation Cell Programming Workshop 3/2/2008 4 SPU Dynamic Performance Profile Checkpoints #include “profile.h” prof_clear(); // clear performance info prof_start(); // start recording performance info < something interesting > prof_stop(); // stop recording performance info Generates SPU0: CP31, 83(82), 4565 Checkpoint # Instruction counter Total (excl. no-op) Cycle counter SPU # C header to enable profiling
Background image of page 4
© 2007 IBM Corporation Cell Programming Workshop 3/2/2008 5 SPU Performance Model ! Referred to in the simulator as “pipeline mode” ! Models salient SPU microarchitectural behavior In-order issue No register renaming Software-controlled branch hints (no HW predict) LS arbitration rules Destructive ILB prefetch load Instruction run-out
Background image of page 5

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
© 2007 IBM Corporation Cell Programming Workshop 3/2/2008 6 SPU Instruction Mappings by Class Channel operations, move to/from SPR 0 6 odd (1) SPR Shuffle bytes, quadword rotate/shift, estimate, gather, form select mask, generate insertion control 0 4 odd (1) SHUF No-op 0 0 even (0) NOP Loads/stores, branch hints 0 6 odd (1) LS No-op 0 0 odd (1) LNOP Special byte operations 0 4 even (0) FXB Element rotate/shift 0 4 even (0) FX3 Load immediate, logical operations, integer add/subtract, sign extend, count leading zeros, select bits, carry/borrow generate 0 2 even (0) FX2 Double precision floating point 6 7 even (0) FPD Integer multiply, integer/float conversion, interpolate 0 7 even (0) FP7 Single precision floating point 0 6 even (0) FP6 Branch 0 4 odd (1) BR Instruction Types Issue Stall Cycles Exec Cycles Exec Pipe Inst Class
Background image of page 6
© 2007 IBM Corporation Cell Programming Workshop 3/2/2008 7 Obtaining SPU Pipeline Statistics ! Via TCL Commands $sim spu n stats print array set s [$sim spu n stats export] $sim spu n display statistics … ! Via GUI Controls
Background image of page 7

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Image of page 8
This is the end of the preview. Sign up to access the rest of the document.

This note was uploaded on 10/03/2011 for the course CDA 6938 taught by Professor Zou,c during the Spring '08 term at University of Central Florida.

Page1 / 27

Day2-03_DevelopingcodeforCell-DynamicProfiling-SDK30 - IBM...

This preview shows document pages 1 - 8. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online