Power Exploration: MLB World Series, Bumgarner, And Box Scores

Exhaustive “power statistics” now play a significant role in the ability to optimize a design for power.

popularity

The San Francisco Giants are fresh off their third Major League Baseball (MLB) World Series win in the last 5 years. That’s notable in itself, but then consider Madison Bumgarner, the starting pitcher for the Giants who was named the 2014 World Series MVP. Bumgarner finished this year’s series with a 2-0 record, one five-inning save (game seven) and 0.43 ERA in three appearances, highlighting the most dominant World Series performance in MLB history.

No one can disagree that Bumgarner was “hot” during the Giant’s post-season run. One of the most talked about topics heading into game seven was how many innings would Bumgarner be able to pitch in relief. This debate focused on the heavily scrutinized “pitch count” statistic. The “pitch count” tracks, as the name implies, the number of pitches thrown by a pitcher during the game. MLB is big business with high team salaries, so maximizing ROI on a multi-million dollar pitching arm is understandable. Those of you who followed game seven know that statistics went out the window as Bumgarner defied odds in delivering a masterful performance in the Giants victory.

MLB has long been associated with capturing statistics on every aspect of the game, especially exhaustive details on individual and team performance. In fact, even baseball player salaries have been represented as a treemap!

Early_Edition_baseballsalaries
Source: Infosthetics

The statistical aspect of MLB often appeals to engineers. This makes perfect sense as engineers are data driven, wired for analyzing trends and identifying patterns in information. The opportunity to review the previous day’s box scores is often an enjoyable distraction.

The intense focus to optimize design power is now approaching a similar paradigm as to how statistics are used in baseball. Power information is being calculated, tracked and reported for every aspect of the design. Optimization decisions are based on reviewing data, analyzing trends, and identifying patterns. This exploration can span across the chip, IP, module, or even component level of the design.

Optimizing a design for power requires a box score, actually several box scores over time (activity). These key “power statistics,” as reported in the box score, are utilized for performing power exploration on a design.

Statistics and Views Required for Effective Power Exploration

  • Power contributors (IP, Module, Micro architecture, components)
  • Information on average activity
  • Average clock gating
  • Register views
  • Memory views
  • Clock details
  • Power domains
  • Opportunities for power reduction
  • Treemap views

“Power statistics” now play a significant role in the ability to optimize a design for power. Designers will continue to need new and innovative ways to capture, analyze and report power information. Albeit, as Madison Bumgarner proved during the recent World Series, even with all available data one cannot cool down a “hot” pitcher.