You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
BEGIN_OP(payload) in a branch with retries (usually in nonblocking structures) would cause the same payload to be persisted for more than once, i.e., to be added into to_be_persisted list several times. This redundancy would affect the performance and increase the average size of to_be_persisted list.
One way to avoid that is to adjust existing API so that users can only set the birth epoch of payload during BEGIN_OP and add it to to_be_persisted list only when the linearization point has been passed (has succeeded). An example can be found in this thread in PR #4.
The text was updated successfully, but these errors were encountered:
Well we may have to add payload to to-be-persisted bucket multiple times, since reset epoch in payload needs to be explicitly persisted after ABORT_OP in order to overwrite potential automatic writes-back after tagging but before resetting epoch. (relevant to Issue #7)
BEGIN_OP(payload)
in a branch with retries (usually in nonblocking structures) would cause the same payload to be persisted for more than once, i.e., to be added intoto_be_persisted
list several times. This redundancy would affect the performance and increase the average size ofto_be_persisted
list.One way to avoid that is to adjust existing API so that users can only set the birth epoch of payload during
BEGIN_OP
and add it toto_be_persisted
list only when the linearization point has been passed (has succeeded). An example can be found in this thread in PR #4.The text was updated successfully, but these errors were encountered: