Skip to content

Score (algorithm)

Anthony J edited this page Mar 15, 2019 · 4 revisions

How does the Score algorithm work?

The score algorithm works by looking at all the tests for a certain component and doing the following. The tests are running various common data structure functions. It's a good test for the simple fact that in every common application, they use internally one of these data structures. For example, computing the digits of pi is another good test because its universally the same algorithm for any language and also a CPU intensive task.

Note: The score will vary immensely on which type of test you will do in the CPU because you will have the option to run a single threaded test, and a multi threaded test.

Factors for determining the SCORE

How long it took the average test to perform that test given (n) iterations.

Example: The CPU successorship test took 00:00:00:00.0000454 seconds.

> How long each iteration takes to perform ().

Example: 00:00:00:00.00000021 per iteration.

Note: depending on the test, it's either per second or tick.

> Total execution time of all tests.

The total amount of time, all the tests took to perform. This is used so there isn't a bias for a particular test. They should all do well or be important together.

What to do with this information.

Each one of those factors has a weight and based on the weight and comparing the the average specs.

avg = The average duration for an individual test.

num = The number of iterations for an individual test.

timePerIteration = avg / num;

timePerIteration has a 35% weight.

totalDur = Total duration of the test for a given component with base number of iterations (*10,000*).

totalDur has a 35% weight

Lastly, comparing against the base computer specifications has a 30% weight


Average specifications and bench marking for the base computer.

Score: 50 CPU: N/A GPU: N/A RAM: N/A SSD: N/A HDD: N/A OS: N/A