-
Notifications
You must be signed in to change notification settings - Fork 41
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
OpenSHMEM 1.6rc1 Draft #540
Conversation
Add atomic fetch/swap to the list, make it clear CAS is conditional, and unify some language ("a PE" -> "the PE", "returns with" -> "returns", "that symmetric data" -> "the symmetric data")
Add scan, avoid active set language, don't say broadcast avoids copying to self since the teams based version does do that.
This makes the collective nature more explicit to match language in the allocation section.
…w list Update put signal to match the language of a regular put more closely and add signal set/add/fetch.
Switch example 3 from `shmem_int_sum_to_all` to `shmem_int_sum_reduce` since the former is deprecated.
Add `SHMEM_CTX_SESSION_TOTAL_OPS` and `SHMEM_CTX_SESSION_BATCH` to the library constant table (text based `SHMEM_TEAM_NUM_CONTEXTS` and `SHMEM_CTX_NOSTORE` respectively since they have similar uses)
[ChangeLog] Add inclusive and exlcusive scan
Co-authored-by: James Dinan <[email protected]>
Signed-off-by: James Dinan <[email protected]>
\item Bob Cernohous, \ac{HPE} | ||
\item James Dinan\footnotemark[1], NVIDIA | ||
\item Megan Grodowitz, Arm Inc. | ||
\item Max Grossman, Georgia Tech |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@agrippa Is this the right institution for you?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I believe @agrippa is currently affiliated with Cruise (https://www.getcruise.com/), but he represented Georgia Tech when he served on the spec committee.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks! Max, let us know if you want to update your affiliation.
For example, the list of \acp{PE} in the parent team should not start at a high | ||
number and then continue to include \acp{PE} in the lower end of the parent | ||
team's \ac{PE} range. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@wokuno To change to "Given a parent team with XYZ PEs, a call to shmem_team_split strided with the following arguments would be invalid."
@@ -13,8 +13,7 @@ | |||
\begin{apiarguments} | |||
\apiargument{IN}{parent\_team}{An \openshmem team.} | |||
|
|||
\apiargument{IN}{start}{The lowest \ac{PE} number of the subset of \acp{PE} from | |||
the parent team that will form the new team.} | |||
\apiargument{IN}{start}{The first \acs{PE} number of the subset of \acp{PE} from the parent team that will form the new team. If the stride is less than zero, the first \acs{PE} number is the highest \acs{PE} of the parent team; if it is greater than zero, it is the lowest; if the stride is zero, it is the starting \acs{PE}.} |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@kwaters4 To look into improving the clarity of this sentence.
All proposed edits merged into #546 |
This PR contains all proposed section committee changes for the OpenSHMEM 1.6rc1 draft, including the following:
PDF Draft: OpenSHMEM-1.6rc1.pdf