Charm++ parallel programming system

[Xiang Ni <xiangni2 AT illinois.edu>]

Hi Kelly,

Checkpoint/restart and message logging are two kind of fault tolerance 
strategies supported by Charm++. In-memory checkpoint/restart is in 
production level and nightly tested.

If you want to try it out, I would suggest you build charm like ./build 
charm++ net-darwin-x86_64 syncft. The restart will only work for more than 2 
nodes due to the algorithm. We're updating our manual for a detailed 
explanation of how it works.

Thanks,

Xiang
On Aug 3, 2012, at 8:24 AM, Kelly Davis wrote:

>> This particular set of options is not expected to work. You've
>> requested that the runtime be built to support 2 different fault
>> tolerance protocols simultaneously, where the code is only written to
>> enable one at a time.
> 
> So, I have to choose if I want "fault tolerance support" xor if I want
> "message logging fault tolerance support"?
> 
> Also, could you elaborate a bit more on what...
> 
>> In their current form, it doesn't really make sense to combine them
> 
> means?
> 
>> ...they're both somewhat experimental, and not so much suited for
>> day-to-day use as yet.
> 
> Reading the documentation I was under the impression that fault tolerance
> was a supported feature. Is this not the case? (Actually, this was the main
> reason I choose to use charm++)
> 
>> Since you're building the OS X, I assume you're targeting a single node,
>> rather than a cluster
> 
> I am targeting a single node for development, but for deployment I am
> going to use a linux cluster.
> 
>> So, my overall suggestion: if you just want to write, compile, and run
>> Charm++ applications, drop the FT options.
> 
> 
> This seems reasonable, but it still leaves open the possibility that
> code paths in the development version are not exercised in the
> deployment version.
> 
> To minimize the slip 'twixt the cup and the lip I thought it wise to build
> and deploy on setups that are as similar as is possible. (Costs prevent
> developing on Linux) It's a pity one can not do so.
> 
>> We'll add a test in the code that will give a more explicit error 
>> on these incompatible options and stop the build cleanly.
> 
> Ok, thanks.
> 
> Kelly
> 
> _______________________________________________
> charm mailing list
> charm AT cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/charm
> 
> _______________________________________________
> ppl mailing list
> ppl AT cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/ppl