pat_view
scaling viewer for CrayPat data
- Author:
Hewlett Packard Enterprise Development LP.
- Copyright:
Copyright 2019,2021-2024 Hewlett Packard Enterprise Development LP.
- Manual section:
1
SYNOPSIS
pat_view [options] [[file 1] [file 2] …]
DESCRIPTION
The pat_view utility is used to view CrayPat data. pat_view takes as input several ap2 data sets and creates either a graph or the raw data of the scaling information for the input data. With the PDF output, columns represent the data sets sorted in ascending order based on # of PEs * # of Threads. The Y axis represents time, samples, or percentages based on options, with a stacked graph of the top routines (or regions) by usages. Note: The –pdf and –csv arguments are mutually exclusive.
OPTIONS
The pat_view utility supports the following arguments.
Generate a PDF formatted file. (default)
–csv
Generate a Comma Separated Values file.
-o file
Specify the name of the output file. By default, the output file is named SCALING.pdf for PDF files and SCALING.csv for CSV files.
–absolute
Use absolute time or sample data. (default)
–relative
Use percentages for each of the data types.
–imbalance
Use imbalance times or samples from the data.
–bottom
For PDF displays, put the routine legend at the bottom. Note: Not applicable to CSV mode.
-w num
Display page width in pixels for PDF displays, based upon 72 dpi. Note: Not applicable to CSV mode. (default: 1000 pixels)
-h num
Display page height in pixels for PDF displays, based upon 72 dpi. Note: Not applicable to CSV mode. (default: 770 pixels)
-n num
Limit the number of top functions to num. (default: 9)
–max num
Set the fixed upper bound for values on the Y axis to num. Note: Not applicable to CSV mode. (default: dynamic)
–pes num
Select only data files with PE count of num. (default: none)
–threads num
Select only data files with threads count of num. (default: none)
–ranks num
Select only data files with a rank count (calculated as PEs * Threads) of num. (default: none)
–skyline
Show display of the data of the top functions as a group of bars. Each bar represents one of the data files. Note: Not applicable to CSV mode.
–areagraph
Show display of the data in a stacked areagraph containing the top functions, with each data file along the X axis. Note: Not applicable to CSV mode.
–quiet
Reduce the amount of on-going status that pat_view outputs when running.
–nonames
Replace the filenames with just file sequence numbers (e.g. * File #1, File #2*, etc). This helps with very long filenames that may cause problems with the legend or with X axis labels.
Note: Using the –nonames option, file positions are as they are found in the command line, rearranged according to data criteria (e.g. PEs and Threads). Note: Not applicable to CSV mode.