The Intel® Concurrency Checker observes how many threads are present in your system at regular intervals. Here is how the captured data is used to calculate your Computed Scaling score:
The Application Concurrency graph shows the concurrency levels that were detected during your assessment. For example, a concurrency level of 2 for 10% of the time means that exactly 2 threads were present for 10% of the time.
The Core Utilization graph shows how the cores in your system could process the threads detected during your assessment. Whenever there are less threads than cores, the number of threads determines how many cores can be utilized. For example, if there was exactly 1 thread for 30% of the time, then your application could use only 1 core for 30% of the time.
Your application will only be able to use all of the cores when the number of threads equals or exceeds the number of cores. To find the percentage of time that your application could use all cores, we add up the percentages of time from the Application Concurrency graph when the number of threads equals or exceeds the number of cores.
Computed Scaling is the average number of threads that could have run during your assessment. It predicts the performance improvement factor when your application is run on a multi-core system, compared to a single-core system. For example, a Computed Scaling value of 1.4 means your application could do 1.4 times as much work on the multi-core system compared to a hypothetical single-core system, during the same period of time.
To calculate your Computed Scaling value, we find the average Core Utilization, which represents the amount of processing that could be accomplished during your assessment.