Skip to content
GitLab
Explore
Sign in
Primary navigation
Search or go to…
Project
SWIFTsim
Manage
Activity
Members
Labels
Plan
Issues
Issue boards
Milestones
Wiki
Code
Merge requests
Repository
Branches
Commits
Tags
Repository graph
Compare revisions
Snippets
Deploy
Releases
Model registry
Monitor
Incidents
Analyze
Value stream analytics
Contributor analytics
Repository analytics
Model experiments
Help
Help
Support
GitLab documentation
Compare GitLab plans
Community forum
Contribute to GitLab
Provide feedback
Keyboard shortcuts
?
Snippets
Groups
Projects
Show more breadcrumbs
SWIFT
SWIFTsim
Commits
f8655faa
Commit
f8655faa
authored
9 years ago
by
Peter W. Draper
Browse files
Options
Downloads
Patches
Plain Diff
Correct some spelling mistakes
parent
7790c02d
Branches
Branches containing commit
Tags
Tags containing commit
2 merge requests
!136
Master
,
!103
PASC paper
Changes
1
Hide whitespace changes
Inline
Side-by-side
Showing
1 changed file
theory/paper_pasc/pasc_paper.tex
+3
-3
3 additions, 3 deletions
theory/paper_pasc/pasc_paper.tex
with
3 additions
and
3 deletions
theory/paper_pasc/pasc_paper.tex
+
3
−
3
View file @
f8655faa
...
...
@@ -284,7 +284,7 @@ particle interactions are then computed by traversing the list of
particles and searching for their neighbours in the tree. In current
state-of-the-art simulations (e.g.
\cite
{
Schaye2015
}
), the gravity
calculation corresponds to roughly one quarter of the calculation time
whilst the hydrodynamics scheme takes approximat
iv
ely half of the
whilst the hydrodynamics scheme takes approximately half of the
total time. The remainder is spent in the astrophysics modules, which
contain interactions of the same kind as the SPH sums presented
here. Gravity could, however, be vastly sped-up compared to commonly
...
...
@@ -460,7 +460,7 @@ e.g.~METIS \cite{ref:Karypis1998} and Zoltan \cite{devine2002zoltan}, exist.
In
\swift
, the graph partitioning is computed using the METIS library.
The cost of each task is initially approximated via the
asymptotic cost of the task type and the number of particles involved.
After a task has been exec
t
ued, it's effective computational cost
After a task has been execu
t
ed, it's effective computational cost
is computed and used.
Note that this approach does not explicitly consider any geometric
...
...
@@ -836,7 +836,7 @@ using synchronous MPI and OpenMP.
Although fully asynchronous methods, due to their somewhat anarchic nature,
may seem more difficult to control, they are conceptually simple and easy
to implement
\footnote
{
The
\swift
source code consists of less than 21
\,
K lines
of code, of which rougly only one tenth are needed to implement the task-based
of code, of which roug
h
ly only one tenth are needed to implement the task-based
parallelism and asynchronous communication.
}
.
The real cost of using a task-based approach comes from having
to rethink the entire computation to fit the task-based setting. This may,
...
...
This diff is collapsed.
Click to expand it.
Preview
0%
Loading
Try again
or
attach a new file
.
Cancel
You are about to add
0
people
to the discussion. Proceed with caution.
Finish editing this message first!
Save comment
Cancel
Please
register
or
sign in
to comment