> ## Documentation Index
> Fetch the complete documentation index at: https://docs.abbyy.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Testing Processing Stations with different types of CPUs at the ABBYY bench (Small)

> Compare ABBYY FlexiCapture Processing Station throughput across Intel Xeon CPU types and core counts, with guidance on estimating cores needed for target loads.

This test compared the performance of the Processing Stations using different types of CPUs and estimated the number of CPU cores required to achieve target loads.

Thousands of grayscale pages processed in 24 hours, by CPU type and core count:

| CPU cores | Intel® Xeon® E5-2680 v4 (2.4 GHz) | Intel® Xeon® E5-2660 v2 (2.2 GHz) | Intel® Xeon® E5-2697A v4 (2.6 GHz) | Intel® Xeon® E5520 (2.27 GHz) | Intel® Xeon® E5-2640 v4 (2.4 GHz) | Intel® Xeon® E3-1230 v6 (3.5 GHz) |
| --------- | --------------------------------- | --------------------------------- | ---------------------------------- | ----------------------------- | --------------------------------- | --------------------------------- |
| 4         | 75                                | 83.5                              | 90                                 | 50                            | 95                                | 135                               |
| 8         | 142                               | 124                               | 150                                | 120                           | 170                               | 207                               |
| 12        | 205                               | 140                               | 211                                | n/a                           | n/a                               | n/a                               |

It should be noted that performance was measured using only one Processing Station, which means that the final result was not significantly affected by network, file storage, database, and FlexiCapture server loads. As more Processing Stations are added, the picture will change, showing a non-linear increase in performance.

For this reason, the data in the table above should not be used directly to estimate the number of CPU cores required to achieve specific performance.

Consider the following recommendations:

* If your estimates show that more than 50 CPU cores is needed, test the fully assembled System to make sure there are no bottlenecks.
* If your estimated number of CPU cores is between 50 and 100, increase this number by 25%.
* If your estimates show that more than 100 CPU cores is needed, load-testing is mandatory.

Additionally note that the performance of your Processing Stations will be affected by the performance of your disk subsystem.