On Wed, 1 Jun 2011, Cody Permann wrote:

> We (the INL) have the need to grow the number of subdomains in our
> simulations past the current storage capacity of subdomain_id_type.
> This type is currently set as an unsigned char allowing only 256
> subdomains.  We are already starting to run and debug libMesh using
> a locally patched version allowing a storage size of 16 bits.  I
> wanted to see what thoughts the other developers have to growing
> this type permanently to 16 bits.  Should we make a new configure
> option to allow us to change the default size?  Should we just make
> the change to the library?

My personal preference would be to have it default to 8 bits, but
configurable to 16 bits with --enable-16bit-subdomain-ids and/or
--enable-everything.  Default to 16 bits but configurable to
--enable-8bit-subdomain-ids would be fine if enough people disagree
with me.

If you're futzing around with configuring data sizes and want to make
processor_id_type configure-time expandable that might be nice too.

> Are there other issues/concerns with changing this type?

I'm pretty sure we'll break if it goes above 32 bits.  (but it
actually might work at 32 bits too, if you want to add a non-default
option for that)

Struct padding in elem.h is such that adding 1 byte to
sizeof(subdomain_id_type) won't necessarily add 1 byte to
sizeof(Elem).  At first glance it looks like it would add 4 bytes on
32-bit systems or 0 bytes on 64-bit systems, but I wouldn't swear to
any of that.
---
Roy

------------------------------------------------------------------------------
Simplify data backup and recovery for your virtual environment with vRanger. 
Installation's a snap, and flexible recovery options mean your data is safe,
secure and there when you need it. Data protection magic?
Nope - It's vRanger. Get your free trial download today. 
http://p.sf.net/sfu/quest-sfdev2dev
_______________________________________________
Libmesh-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/libmesh-devel

Reply via email to