Scribd
Upload
52 views2 pages

Mini Report

The document summarizes improvements made to code for a Massimiliano Gentile's MPHS Project. Key changes include: 1) Implementation of interaction analysis between vector data types and changes in interaction evaluation. 2) Code optimization through reorganization and parallelization, reducing lines of code by 12% and function times by up to 42%. 3) While CPU usage improved and function times decreased for most tests, system/user time ratios and voluntary context switches increased, indicating less efficient use of CPU resources.

Uploaded by

nFo0000
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
52 views2 pages

Mini Report

The document summarizes improvements made to code for a Massimiliano Gentile's MPHS Project. Key changes include: 1) Implementation of interaction analysis between vector data types and changes in interaction evaluation. 2) Code optimization through reorganization and parallelization, reducing lines of code by 12% and function times by up to 42%. 3) While CPU usage improved and function times decreased for most tests, system/user time ratios and voluntary context switches increased, indicating less efficient use of CPU resources.

Uploaded by

nFo0000
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

Report MPHS Project

Massimiliano Gentile

Implementation
 Implementation of interaction analysis between vector data types

 Changes in the evaluation of interaction

Fixes
 Added links to the summary page
 Minor changes to remove all the GCC warnings (but the behavior of similar() has changed)

Optimizations
 Reorganization of the code to improve readability and maintainability
o db_generate_html() from 671 loc to 100 loc (-571)
o main_effects_plot_to_file() from 903 loc to 211 loc (-692)
 The code related to the latex report generation can be entirely excluded if CREATE_LATEX_PDF is
not defined
 Parallelization of the surface plot images creation with threads
Statistics
Old code New code Diff
Lines of code 3753 3294 -459 (-12%) Good

Core 2 Duo
Database 256 points
Function time 37.9 sec 29.75 sec -8.15 (-21%) Good
Ratio System/User Time 43.4% 55.2% +11.8 Bad
% CPU this job got 108% 158% +50 (+25%) Good
Voluntary context switches 52201 150466 +98265 (+188%) Bad

Database 27648 points


Function time 11:56 7:50 -246 (-34.4%) Good
Ratio System/User Time 3.6% 8.6% +5 Bad
% CPU this job got 121% 186% +65 (+32.5%) Good
Voluntary context switches 303303 926972 +623669 (+206%) Bad

HBomb (8core)
Database 256 points
Function time 33.5 sec 19.36 sec -14.14 (-42%) Good
Ratio System/User Time 140.6% 220% +79.4 Bad
% CPU this job got 200% 501% +301 (+37.6%) Good
Voluntary context switches 1266759 1974929 +708170 (+56%) Bad

Database 27648 points


Function time 6:16 6:11 -5 (-1.3%) Good
Ratio System/User Time 104% 262% +158 Bad
% CPU this job got 268% 707% +439 (+55%) Good
Voluntary context switches 17394555 50980220 +33585665 (+193%) Bad

NB: In the 27648 points simulation, with the new code, 191 threads are spawned.

You might also like