Shantanu Gupta wrote:
I have two related questions:
1) Is the creation of a benchmark's warm checkpoint a requirement for the
correct execution inside ruby ? Or it's only applied for a better
simulation performance.
The warm checkpoint is not required. If you dont' care about warming up
the caches, or are willing to warm them up at the onset of every
simulation, you can get away with only "cold" checkpoints.
2) Is it even possible to create a cold checkpoint for a transactional
memory benchmark ? Because it seems to me that without loading ruby
module, transactional memory benchmark will fail to execute correctly.
The Tourmaline module will perform functional simulation of
transactional memory, though it is out-of-date with respect to the LogTM
latest release. It is best if you stop your "cold" checkpoint before any
begin_xact's are encountered... usually with a "magic" instruction of
some sort.
Regards,
Dan
Shantanu
_______________________________________________
Gems-users mailing list
Gems-users@xxxxxxxxxxx
https://lists.cs.wisc.edu/mailman/listinfo/gems-users
Use Google to search the GEMS Users mailing list by adding "site:https://lists.cs.wisc.edu/archive/gems-users/"; to your search.
--
http://www.cs.wisc.edu/~gibson [esc]:wq!
|
![[Computer Systems Lab]](https://www-auth.cs.wisc.edu/pics/csl_logo.gif){kind=logo}