This site uses different types of cookies, including analytics and functional cookies (its own and from other sites). To change your cookie settings or find out more, click here. If you continue browsing our website, you accept these cookies.
When to use AMP Engine Designer 2020.2 and Alteryx Analytics Hub introduce the new AMP Engine. It is most times faster than the e1 engine. This article will give some guidelines on how to decide which engine to use. AMP stands for Alteryx Multi-threaded Processing. As implied by its name it enables processing of multiple packets in parallel. Contrasts this with e1 Engine that processes data record by record in sequence.
The AMP engine must have at least 400 MB to process one thread of a workflow. For example, with 8 threads, there must be at least 3.2 GB of memory available to AMP at runtime. If there are only 2.5 GB available at runtime, AMP will use 6 threads to satisfy the 400 MB minimum.
Procedure Experience shows that for modern multi core machines have best results. Adding additional cores can lead to almost linear performance increase. Please note that the below are merely guidelines, however your mileage might vary. Note also that not all tools have yet been converted to use AMPas AMP is actively being developed.
Testing AMP: Run an existing workflow a few times in the e1 Engine, then run the workflow a few times with the AMP Engine. Compare the Results Pane (messages, warnings, errors, duration of the run) as well as any resulting data to check for differences.
To identify whether a workflow ran with the AMP Engine, check the results pan for a message stating "This is AMP; running [N] worker threads".
AMP performs better than e1 for typical prep / blend / analysis use cases within Alteryx.
You should get an advantage even when running workflows with a mixture of converted and non-converted tools.
Processes beyond AMP (R, Python etc) or workflows with mostly non-converted tools - use e1.
In some cases (Multi Row-Formula tools) and certain Spatial workflows AMP will be slower and might require some tweaking to work as expected.
The benefit of AMP increases with increasing data size.
For smaller hardware and small data, use e1.
Server Support For the multi worker node environment that is typically used by many enterprise customers use in Production AMP Engine has not been sufficiently tested. However, with the release of Server 2020.3 the expectation is to have updated guidelines. Please note that this will depend on the progress and outcome of the tests and is outside Support.