Figures

FIGURE 2-1 Source Tab Showing Annotated Source Code for Function cputime

FIGURE 2-2 Source Tab Showing Annotated Source Code for Function icputime

FIGURE 2-3 Disassembly Tab Showing Instructions for the Line in Which x Is Incremented in Function cputime

FIGURE 2-4 Callers-Callees Tab With gpf_work as the Selected Function

FIGURE 2-5 Source Tab Showing Annotated Source Code for Functions gpf_a and gpf_b

FIGURE 2-6 Source Tab Showing Annotated Source Code for Function gpf_work

FIGURE 2-7 Callers-Callees Tab With real_recurse as the Selected Function

FIGURE 2-8 Callers-Callees Tab With bounce_a as the Selected Function

FIGURE 2-9 Callers-Callees Tab With bounce_b as the Selected Function

FIGURE 2-10 Functions Tab Showing Functions so_burncpu and sx_burncpu

FIGURE 2-11 Timeline Tab Showing the Seven Experiments Recorded for the Parent Process and its Descendant Processes

FIGURE 2-12 Timeline Tab at High Zoom Showing Event Markers and Gaps Between Them

FIGURE 2-13 Experiments Tab Showing Seven Experiments, Three of Which Are Marked as "Bad"

FIGURE 2-14 Timeline Tab at High Zoom, Showing Short Sample for Experiment 5

FIGURE 2-15 Event Tab Showing Very Short Duration Sample

FIGURE 2-16 The Functions Tab, Showing Several jsynprog Experiment Methods

FIGURE 2-17 The Summary Panel

FIGURE 2-18 The Metrics Tab of the Set Data Presentation Dialog

FIGURE 2-19 The Callers-Callees Tab, With jsynprog.main Selected

FIGURE 2-20 Source Tab, Listing jsynprog.java

FIGURE 2-21 The Disassembly Tab, Showing Annotated Bytecode

FIGURE 2-22 The Timeline Tab in the Java Representation

FIGURE 2-23 The Timeline Tab in the Expert-Java Representation

FIGURE 2-24 The Functions Tab, Showing Interpreted and Dynamically-compiled Versions of Routine.sys_op

FIGURE 2-25 Summary Tabs for Function psec_ From the Four-CPU Run (Left) and the Two-CPU Run (Right)

FIGURE 2-26 Summary Tabs for Function pdo_ From the Four-CPU Run (Left) and the Two-CPU Run (Right)

FIGURE 2-27 Functions Tab Showing Entries for critsum_ and redsum_

FIGURE 2-28 Functions Tab for the Four-CPU Experiment Showing Data for lock_local and lock_global

FIGURE 2-29 Source Tab for the Four-CPU Experiment for Function lock_global

FIGURE 2-30 Source Tab for the Four-CPU Experiment for Function lock_local

FIGURE 2-31 Functions Tab for the One-CPU Experiment Showing Data for lock_local and lock_global

FIGURE 2-32 Functions Tab for the One-CPU Experiment Showing Data for Functions computeA and computeB

FIGURE 2-33 Functions Tab for the Four-CPU Experiment Showing Data for Functions computeA and computeB

FIGURE 2-34 Source Tab for the Four-CPU Experiment Showing Annotated Source Code for computeA and computeB

FIGURE 2-35 Source tab for the Four-CPU Experiment Showing Annotated Source Code for cache_trash

FIGURE 2-36 Functions Tab Showing User CPU, FP Adds and FP Muls for the Six Variants of dgemv

FIGURE 2-37 Functions Tab Showing User CPU Time, CPU Cycles, Instructions Executed and D- and E-Cache Stall Cycles for the Six Variants of dgemv

FIGURE 2-38 Source Tab Showing Annotated Source Code for dgemv_g1 and dgemv_g2

FIGURE 2-39 Source Tab for dgemv_hi1 Showing Compiler Commentary That Includes Loop Interchange Messages

FIGURE 2-40 Source Tab for dgemv_hi2 Showing Compiler Commentary

FIGURE 3-1 Call Tree Illustrating Exclusive, Inclusive, and Attributed Metrics

FIGURE 5-1 The Performance Analyzer Window

FIGURE 5-2 The Functions Tab

FIGURE 5-3 The Callers-Callees Tab

FIGURE 5-4 The Source Tab

FIGURE 5-5 The Lines Tab

FIGURE 5-6 The Disassembly Tab

FIGURE 5-7 The PCs Tab

FIGURE 5-8 Data Objects Tab

FIGURE 5-9 The Timeline Tab

FIGURE 5-10 The LeakList Tab

FIGURE 5-11 The Statistics Tab

FIGURE 5-12 The Experiments Tab

FIGURE 5-13 The Summary Tab

FIGURE 5-14 Data Objects Summary

FIGURE 5-15 The Event Tab, Showing Event Data

FIGURE 5-16 The Legend Tab

FIGURE 5-17 The Leak Tab

FIGURE 5-18 The Formats Tab

FIGURE 7-1 Schematic Call Tree for a Multithreaded Program That Contains a Parallel Do or Parallel For Construct

FIGURE 7-2 Schematic Call Tree for a Parallel Region With a Worksharing Do or Worksharing For Construct