uima-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jaroslaw Cwiklik <uim...@gmail.com>
Subject Re: Multi Core Usage
Date Thu, 08 Nov 2012 15:04:12 GMT
Not sure I fully understand your problem. In the CPE, the CR runs in its
own thread supplying CASes to processing pipelines.
Cas Consumer(s) run in a dedicated thread(s) as well.
The processingUnitThreadCount property determines scaleout for your AEs
if you set it to say 3, you get 3 processing threads (pipelines) and 3
instances of each AE.

Take a look at this document:
http://uima.apache.org/d/uimaj-2.3.1/tutorials_and_users_guides.pdf
page 61.

Can you explain this better:
"...this does not work because it only adds the 3 "levels" reading,
analyzing, writing to differnt threads."

- Jerry


On Wed, Nov 7, 2012 at 11:09 AM, Andreas Niekler <
aniekler@informatik.uni-leipzig.de> wrote:

> Hi there,
>
> i have a uima defininition including a Reader, The Analysis chain and a
> consumer. I tried to scale out the whole application by adding the
> parameters <casProcessors casPoolSize="3" processingUnitThreadCount="1">
> and set them to higher values. This does not work because it only adds the
> 3 "levels" reading, analyzing, writing to differnt threads.
>
> How can i scale out the analysis part. One important thing to mention is
> that all the annotators in the chain must be linear. This means for one
> document they have to run one after another. So the solution would be to
> scale out the document processing and process each document concurrent but
> in each thread with the whole chain.
>
> Is that possible?
>
> Thank you
>
> --
> Andreas Niekler, Dipl. Ing. (FH)
> NLP Group | Department of Computer Science
> University of Leipzig
> Johannisgasse 26 | 04103 Leipzig
>
> mail: aniekler@informatik.uni-**leipzig.deg.de<aniekler@informatik.uni-leipzig.deg.de>
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message