Hey,

Wow, that sounds neat! I was hoping to bum some free hardware from intel
so I could continue optimizing the 1.4 tree... :)

Any way, what version of 1.6 have you written this against? One of the
released beta's or the source tree?

https://github.com/memcached/memcached/tree/engine-pu

Our main code tree is over at github.com/memcached/memcached - the link I
posted above is the "1.6" tree, which is called "engine-pu" on our end.

The best way for you to contribute is to grab the very latest engine-pu
branch, and make sure your code fully works against that. Then push your
tree with the new engine into it somewhere, ideally also github. We can
then track changes and exchange feedback or easily merge it (without
losing any history you feel like sharing with us).

Keep in midn that if you want us to distribute your engine along with 1.6,
it'll need to have a compatible license. We also try to not ship code
which is patent encumbered, so please disclose any which may be in use in
your changes.

We really appreciate you taking the time to do this, at any rate. I would
love to see the code!

have fun,
-Dormando

On Wed, 20 Jun 2012, rajiv wrote:

> Hi,
>  
> I work at Intel and we have added a new optimized "engine" to Memcached 1.6.  
> This engine uses parallel hash table accesses and an LRU implementation that 
> gives us
> lock free GET operations.  We have tested this version doing GET 
> operations on 1 to 16 cores and seems to scale pretty well.
>  
> We would like to contribute these changes back to the 1.6 branch.  Can 
> someone tell me or point me to info on the process to do that?  Is it is 
> multi step process
> where I first need to upload the entire 1.6 version that we have with the new 
> code and then later after some review we integrate our changes into the 1.6
> mainline?  Or do I have to download lthe latest 1.6 branch, port our changes 
> to it and post that?  Also where (URL) is it that I would upload the 
> version/changes
> that I have?
>  
> Sorry for these basic questions - I have never actually contributed to Open 
> Source code in the past.
>  
> thanks,
> \rajiv
>
>

Reply via email to