ELECTRONIC T-NOTES


CHESSBASE USA'S WEEKLY ON-LINE NEWSLETTER


FOR THE WEEK OF MAY 13, 2001


THE FRITZMARK

by Steve Lopez

benchmark (n.) A standard of comparison.

FritzMark (n.) A standard of comparison between two or more computers running the Fritz6 interface.

There's nothing terribly mysterious about that -- the FritzMark is just a number that's used as a relative rating of computer performance, so that you can compare the expected performance of different machines. You run the FritzMark feature on two different computers and can expect stronger chess play from the machine that scores the higher FritzMark.

Heck, you can even use the feature to compare performances on the same machine at different times. For example, as your computer's processor runs it heats up over time. As it heats up it becomes less efficient. You can run a FritzMark on your computer right after you switch it on in the morning and make a note of the figure you receive. Then run FritzMark again late at night after you've been using your computer all day. Chances are that you'll see a somewhat lower FritzMark figure in the evening than you did in the morning.

But a FritzMark figure by itself means nothing, nada, zilch, bupkis. It's a method of comparison. And, for those nodding off at the back of the class, the word "comparison" here strongly connotes that two or more numbers are involved.

To check the FritzMark (which is called "TigerMark" in Chess Tiger, "JuniorMark" in Junior6, etc.) on your machine, go to the Tools menu in your playing program. You'll see the menu command for this function clearly displayed.

To get the best possible play and strength from Fritz (or the other chessplaying programs), you don't want to have any unneccesary programs running on your machine. The same thing applies to gauging a FritzMark. Exit out of all non-required applications before running FritzMark to get the best possible results.

Start the FritzMark function and you'll see the following window appear:


To begin the benchmarking process, click the "Start" button. If you're using Fritz6, the program will load Fritz5.32 as the default benchmarking engine (since it's the fastest engine that comes on the Fritz6 CD). Other programs (Chess Tiger, Junior6, Nimzo8) will load other engines, of course. After anywhere from several seconds to several minutes (depending on the speed of your computer), you'll begin to see information displayed in this window:

And, of course, the "Close" button ends the process.

The program uses the kilonodes figure in conjunction with the hash table size to assign a benchmark figure (the FritzMark) evaluating the efficiency and performance of the computer in analyzing positions. You can then use this figure to compare that machine to another machine on which you've run the FritzMark, or (as noted above) compare the performance of the same machine under different circumstances (such as length of time the computer's been running).

As noted previously, a single FritzMark figure is meaningless (so the question "What does a FritzMark of x indicate?" is kind of a dopey question); you need two or more FritzMarks to be able to compare performances. As a guideline, here are FritzMarks for different machines. Some are from the Fritz Help files and some are from tests that I've run on different computers:

You can run the FritzMark on your machine and use these figures as a basis to compare Fritz' performance on your hardware to its performance on other machines. Obviously, the faster the processor, the higher the FritzMark and the better the performance.

Until next week, have fun!

You can e-mail me with your comments, suggestions, and analysis for Electronic T-Notes.