ignite-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Alexei Scherbakov <alexey.scherbak...@gmail.com>
Subject Re: MPI / OpenMP compared to Ignite
Date Sat, 07 May 2016 11:58:03 GMT
Hi,

OpenMP is like Ignite's Compute[1] subsystem running on the single JVM
(because it uses shared memory).
MPI is like Ignite's discovery SPI( and very limited, because MPI does not
allow adding nodes dynamically after cluster start ) + communication SPI
for message passing between nodes.
Often OpenMP is used together with MPI forming hybrid model, which provides
richer programming capabilities.

Ignite provides out of the box things like computation load balancing and
failover, distributed memory, job checkpoints, dynamic cluster membership,
and much more compared to OpenMP/MPI.
If you need such advanced capabilities in your OpenMP/MPI program, you must
implement it yourself.

Conclusion: Ignite provides much greater functionality compared to
OpenMP/MPI, is written i Java, so you get safety, performance and coding
speed.
I see no point in using something instead of Ignite for distributed
workloads.

Disclaimer: I'am not an expert on OpenMP/MPI, so my opinion may be biased.

[1] https://apacheignite.readme.io/docs/compute-grid


2016-05-06 8:35 GMT+03:00 arthi <Arthi.Kasturirangan.ap@nielsen.com>:

> Hi Team,
>
> How do you compare Ignite data & compute grid components to OpenMP / MPI?
> can you please call out the pros and cons of the two frameworks?
>
> Thanks,
> Arthi
>
>
>
> --
> View this message in context:
> http://apache-ignite-users.70518.x6.nabble.com/MPI-OpenMP-compared-to-Ignite-tp4804.html
> Sent from the Apache Ignite Users mailing list archive at Nabble.com.
>



-- 

Best regards,
Alexei Scherbakov

Mime
View raw message