Per core performance is a good thing...

However, you don't mention disk access.  I would spend more on faster 
disks/SSDs than cores.

Martin Morrison
Infrastructure Application Engineer/Systems Analyst
 Engineering Design Systems, Inc.
540.345.1410
[email protected]<mailto:[email protected]>
[cid:[email protected]]
[cid:[email protected]]<http://www.facebook.com/engineeringdesignsystemsinc>[cid:[email protected]]<http://www.twitter.com/engdesignsystem>
 [cid:[email protected]] 
<https://www.youtube.com/channel/UCkJ7-Sbji64Dwke89L1R8Lw>
www.edsi.com<http://www.edsi.com>

From: mapguide-users [mailto:[email protected]] On Behalf 
Of Tim Price
Sent: Friday, August 18, 2017 11:33 AM
To: '[email protected]' <[email protected]>
Subject: [mapguide-users] Researching Processors for MGOS

Happy Friday Everyone,

We're beginning the process of looking at a speedy Dell windows server for a 
MGOS3.1/Rest/custom viewer installation for a concurrent user base of 10-20 
staff. (total staff 100-150) We rely heavily on tile definitions, load 
procedures and MS SQL vector data sources with many 
filters/expressions/theming/schema overrides on the layers. FDO is looking at 
gdal rasters, shape and MS sql.

We currently run a Intel E5-2690 (8 core, 16 thread) processor with all MG 
services on that hardware server. It runs no other applications. What we can 
gleam form our setup is that the per-core performance seems to be more 
important than RAM or number of cores of a MGOS installation in our 
configuration. This is because RAM and # cores is generous on most servers 
these days. We are studying https://www.cpubenchmark.net/singleThread.html . 
Can anyone confirm that per-core performance is what we are really after or 
provide additional advice?

Thanks,

-Tim

_______________________________________________
mapguide-users mailing list
[email protected]
https://lists.osgeo.org/mailman/listinfo/mapguide-users

Reply via email to