Computer: CPU Burden


The following can lighten the CPU burden which would enable slower computers to keep up in fast markets.

  • Have the Optimize Speed chart property box checked.  Press CTRL-P to show the form.
  • Have fewer charts open simultaneously.
  • Use a smaller Max Bars setting for the chart so less memory is used and studies have fewer bars to calculate for.
  • Dress each chart with fewer studies, alert objects and draw tools.
  • The PriceFinder™ feature makes extensive computations.  Use it sparingly.
  • Remove from the chart objects list obsolete draw tools no longer being used or seen in the chart view.
  • Keep the SetUp | Manager symbol watch list trimmed down by removing unused or unneeded symbols.
  • On a regular basis, run the Windows system Defrag and ScanDisk programs.

Optimize Speed
On the chart property form that shows by pressing CTRL-P have the Optimize Speed box checked.   When this box is checked the program will not recalculate studies when an equal price tick is received.  Most studies are based on price and their calculation would not change because the price did not change.  If a new bar starts the calculation will be made and the studies updated.   If studies that are based on volume are used, you may want to recalculate the study.  For the majority of users, having the Optimize Speed box checked will be acceptable.   For many symbols, the equal price ticks constitute 2/3rds of the ticks sent, and so it is a huge CPU savings to skip recalculating the studies on an equal price tick.

Memory Used
Click menu Help | Memory Report to show the following form which estimates the memory used by the number of open windows, the objects (studies) that dress the charts, and the Max Bars setting for each chart.   In the example, the 43 open windows use 153.9 megabytes of memory.

Chart Objects
The following have been ranked from greatest CPU burden to least burden.

PriceFinder™ calculations can be massive.  Use the feature sparingly.   PriceFinder evaluates all studies that precede it on the Chart Objects list.   Therefore, have just the studies needed by PriceFinder at the top of the Chart Objects list, immediately followed by the DYO which has the PriceFinder.   Those studies, alerts, and draw tools that follow the DYO with PriceFinder are ignored by PriceFinder.

Studies typically recalculate with each new tick, and benefit from the Optimize Speed box being checked.  A Study on Study is a heavier calculation by having to look up the study value to be used in the calculation.

Alerts and DYOs also recalculate with each new tick.  The exception would be if the Close Only box is checked on the alert property form.  In general, alerts and DYOs have a heavy burden with lots of code to implement obtaining the values for the tests and performing the evaluation.  For the marker visual, the BGD, ZON, and HLT selections impose the greatest CPU demands.

The visual of the spread for the study is not an issue, but rather the recalculation frequency.   Being a study, it recalculates on every tick.  There is less burden for studies than for alerts because the prices are easier to fetch.  The CPU burden is greatest for Simple and Weighted, and least for Exponential and Smoothed average formulas.

The Pesavento Patterns and Pyrapoint tools recalculate on bar completion rather than with every new tick. Thus the overall burden is light.  The Ensign Map study is a big CPU calculation, but that is done only when the study is applied or when it is scheduled to recalculate.  The Map does not recalculate with each new tick or each new bar.  So while its initial calculation burden is great, the saving grace is the Map calculation is rarely done.  Thus, the overall burden is light.

Draw tools occupy some memory to hold their parameters.  There is some calculation and drawing burden when the chart draws or redraws.   But, draw tools do not recalculate with each new tick.  Therefore, overall burden for all draw tools and notes is light.  It makes no difference whether the line style is thick or thin.

Minimized
The memory needed to hold chart data and study objects is the same whether a chart is seen or unseen. The calculation burden for studies is the same for displayed, minimized, and hidden charts.  There is a savings for the graphical update if a chart is minimized or hidden.

read more » CPU Speed
read more » Tune-up Ensign


Last modified 11/19/08 8:40 PM