Why is ParallelDo slower than Do?
$begingroup$
I have problems to write parallel code in mathematica.
Why is
candidates = {};
SetSharedVariable[candidates];
Do[
ParallelDo[
eq = RandomReal + RandomReal;
AppendTo[candidates, eq]
, {j, 1, 1000}]
, {i, 1, 10}]
slower than the non parallel version
candidates = {};
Do[
Do[
eq = RandomReal + RandomReal;
AppendTo[candidates, eq]
, {j, 1, 1000}]
, {i, 1, 10}]
?
parallelization
New contributor
$endgroup$
add a comment |
$begingroup$
I have problems to write parallel code in mathematica.
Why is
candidates = {};
SetSharedVariable[candidates];
Do[
ParallelDo[
eq = RandomReal + RandomReal;
AppendTo[candidates, eq]
, {j, 1, 1000}]
, {i, 1, 10}]
slower than the non parallel version
candidates = {};
Do[
Do[
eq = RandomReal + RandomReal;
AppendTo[candidates, eq]
, {j, 1, 1000}]
, {i, 1, 10}]
?
parallelization
New contributor
$endgroup$
$begingroup$
I reverted your post to before the edit because it looks like a different question (like Henrik said in his comment). Note, however, that if you ask it in precisely such form it will be likely closed due to not enough info: you need to provide the minimal working example, not through some undefined functions into a piece of code that no one will be able to run and test.
$endgroup$
– corey979
yesterday
1
$begingroup$
See here mathematica.stackexchange.com/a/48296/12 I suggest you don't use SetSharedVariable until you get quite fluent in using the parallel tools. It effectively "unparallelizes" your code.
$endgroup$
– Szabolcs
yesterday
add a comment |
$begingroup$
I have problems to write parallel code in mathematica.
Why is
candidates = {};
SetSharedVariable[candidates];
Do[
ParallelDo[
eq = RandomReal + RandomReal;
AppendTo[candidates, eq]
, {j, 1, 1000}]
, {i, 1, 10}]
slower than the non parallel version
candidates = {};
Do[
Do[
eq = RandomReal + RandomReal;
AppendTo[candidates, eq]
, {j, 1, 1000}]
, {i, 1, 10}]
?
parallelization
New contributor
$endgroup$
I have problems to write parallel code in mathematica.
Why is
candidates = {};
SetSharedVariable[candidates];
Do[
ParallelDo[
eq = RandomReal + RandomReal;
AppendTo[candidates, eq]
, {j, 1, 1000}]
, {i, 1, 10}]
slower than the non parallel version
candidates = {};
Do[
Do[
eq = RandomReal + RandomReal;
AppendTo[candidates, eq]
, {j, 1, 1000}]
, {i, 1, 10}]
?
parallelization
parallelization
New contributor
New contributor
edited yesterday
corey979
20.9k64282
20.9k64282
New contributor
asked yesterday
Matthias HellerMatthias Heller
283
283
New contributor
New contributor
$begingroup$
I reverted your post to before the edit because it looks like a different question (like Henrik said in his comment). Note, however, that if you ask it in precisely such form it will be likely closed due to not enough info: you need to provide the minimal working example, not through some undefined functions into a piece of code that no one will be able to run and test.
$endgroup$
– corey979
yesterday
1
$begingroup$
See here mathematica.stackexchange.com/a/48296/12 I suggest you don't use SetSharedVariable until you get quite fluent in using the parallel tools. It effectively "unparallelizes" your code.
$endgroup$
– Szabolcs
yesterday
add a comment |
$begingroup$
I reverted your post to before the edit because it looks like a different question (like Henrik said in his comment). Note, however, that if you ask it in precisely such form it will be likely closed due to not enough info: you need to provide the minimal working example, not through some undefined functions into a piece of code that no one will be able to run and test.
$endgroup$
– corey979
yesterday
1
$begingroup$
See here mathematica.stackexchange.com/a/48296/12 I suggest you don't use SetSharedVariable until you get quite fluent in using the parallel tools. It effectively "unparallelizes" your code.
$endgroup$
– Szabolcs
yesterday
$begingroup$
I reverted your post to before the edit because it looks like a different question (like Henrik said in his comment). Note, however, that if you ask it in precisely such form it will be likely closed due to not enough info: you need to provide the minimal working example, not through some undefined functions into a piece of code that no one will be able to run and test.
$endgroup$
– corey979
yesterday
$begingroup$
I reverted your post to before the edit because it looks like a different question (like Henrik said in his comment). Note, however, that if you ask it in precisely such form it will be likely closed due to not enough info: you need to provide the minimal working example, not through some undefined functions into a piece of code that no one will be able to run and test.
$endgroup$
– corey979
yesterday
1
1
$begingroup$
See here mathematica.stackexchange.com/a/48296/12 I suggest you don't use SetSharedVariable until you get quite fluent in using the parallel tools. It effectively "unparallelizes" your code.
$endgroup$
– Szabolcs
yesterday
$begingroup$
See here mathematica.stackexchange.com/a/48296/12 I suggest you don't use SetSharedVariable until you get quite fluent in using the parallel tools. It effectively "unparallelizes" your code.
$endgroup$
– Szabolcs
yesterday
add a comment |
1 Answer
1
active
oldest
votes
$begingroup$
Because managing write access to shared memory is expensive: Subprocesses have to wait until they are granted write access (because another process uses that ressource).
Moreover, it is in general more efficient to use Parallel
only upon the most outer loop construct.
By the way: Using Append
and AppendTo
are the worst methods to build a list, because they involve a copy of the full list each time another element is appended. Instead of complexity $O(n)$ for a list of $n$ elements, you get an implementation of complexity $O(n^2)$. Better use Table
or, if you don't know how long the list is about to get, use Sow
and Reap
. Internal`Bag
is a further option, and it is even compilable.
$endgroup$
$begingroup$
Thanks, that actually helped a lot. I just dont understand how to use Sow and Reap to avoid Append To be more specific: instead of ParallelDo I use now ParallelTable: eq = ParallelTable[ FNumeric[ SetPrecision[N[monlistnumeric[[i]] + monlistnumeric[[j]], 20], 10]] , {j, jj}]; FNumeric is a function, that returns either 0 or a value I want to store. I then do eq = DeleteCases[eq, 0]; candidates = Join[candidates, eq]; Is there a more efficient way to do this?
$endgroup$
– Matthias Heller
yesterday
2
$begingroup$
@MatthiasHeller, you're welcome. How is this new code related to your post? You should consider a new post with your real problem and all relevant data. I may have a look. In general, depending on the details, there are various ways to perform the computation efficiently; these way might not useParallel
at all, but ratherCompile
d code.
$endgroup$
– Henrik Schumacher
yesterday
add a comment |
Your Answer
StackExchange.ready(function() {
var channelOptions = {
tags: "".split(" "),
id: "387"
};
initTagRenderer("".split(" "), "".split(" "), channelOptions);
StackExchange.using("externalEditor", function() {
// Have to fire editor after snippets, if snippets enabled
if (StackExchange.settings.snippets.snippetsEnabled) {
StackExchange.using("snippets", function() {
createEditor();
});
}
else {
createEditor();
}
});
function createEditor() {
StackExchange.prepareEditor({
heartbeatType: 'answer',
autoActivateHeartbeat: false,
convertImagesToLinks: false,
noModals: true,
showLowRepImageUploadWarning: true,
reputationToPostImages: null,
bindNavPrevention: true,
postfix: "",
imageUploader: {
brandingHtml: "Powered by u003ca class="icon-imgur-white" href="https://imgur.com/"u003eu003c/au003e",
contentPolicyHtml: "User contributions licensed under u003ca href="https://creativecommons.org/licenses/by-sa/3.0/"u003ecc by-sa 3.0 with attribution requiredu003c/au003e u003ca href="https://stackoverflow.com/legal/content-policy"u003e(content policy)u003c/au003e",
allowUrls: true
},
onDemand: true,
discardSelector: ".discard-answer"
,immediatelyShowMarkdownHelp:true
});
}
});
Matthias Heller is a new contributor. Be nice, and check out our Code of Conduct.
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
StackExchange.ready(
function () {
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fmathematica.stackexchange.com%2fquestions%2f195006%2fwhy-is-paralleldo-slower-than-do%23new-answer', 'question_page');
}
);
Post as a guest
Required, but never shown
1 Answer
1
active
oldest
votes
1 Answer
1
active
oldest
votes
active
oldest
votes
active
oldest
votes
$begingroup$
Because managing write access to shared memory is expensive: Subprocesses have to wait until they are granted write access (because another process uses that ressource).
Moreover, it is in general more efficient to use Parallel
only upon the most outer loop construct.
By the way: Using Append
and AppendTo
are the worst methods to build a list, because they involve a copy of the full list each time another element is appended. Instead of complexity $O(n)$ for a list of $n$ elements, you get an implementation of complexity $O(n^2)$. Better use Table
or, if you don't know how long the list is about to get, use Sow
and Reap
. Internal`Bag
is a further option, and it is even compilable.
$endgroup$
$begingroup$
Thanks, that actually helped a lot. I just dont understand how to use Sow and Reap to avoid Append To be more specific: instead of ParallelDo I use now ParallelTable: eq = ParallelTable[ FNumeric[ SetPrecision[N[monlistnumeric[[i]] + monlistnumeric[[j]], 20], 10]] , {j, jj}]; FNumeric is a function, that returns either 0 or a value I want to store. I then do eq = DeleteCases[eq, 0]; candidates = Join[candidates, eq]; Is there a more efficient way to do this?
$endgroup$
– Matthias Heller
yesterday
2
$begingroup$
@MatthiasHeller, you're welcome. How is this new code related to your post? You should consider a new post with your real problem and all relevant data. I may have a look. In general, depending on the details, there are various ways to perform the computation efficiently; these way might not useParallel
at all, but ratherCompile
d code.
$endgroup$
– Henrik Schumacher
yesterday
add a comment |
$begingroup$
Because managing write access to shared memory is expensive: Subprocesses have to wait until they are granted write access (because another process uses that ressource).
Moreover, it is in general more efficient to use Parallel
only upon the most outer loop construct.
By the way: Using Append
and AppendTo
are the worst methods to build a list, because they involve a copy of the full list each time another element is appended. Instead of complexity $O(n)$ for a list of $n$ elements, you get an implementation of complexity $O(n^2)$. Better use Table
or, if you don't know how long the list is about to get, use Sow
and Reap
. Internal`Bag
is a further option, and it is even compilable.
$endgroup$
$begingroup$
Thanks, that actually helped a lot. I just dont understand how to use Sow and Reap to avoid Append To be more specific: instead of ParallelDo I use now ParallelTable: eq = ParallelTable[ FNumeric[ SetPrecision[N[monlistnumeric[[i]] + monlistnumeric[[j]], 20], 10]] , {j, jj}]; FNumeric is a function, that returns either 0 or a value I want to store. I then do eq = DeleteCases[eq, 0]; candidates = Join[candidates, eq]; Is there a more efficient way to do this?
$endgroup$
– Matthias Heller
yesterday
2
$begingroup$
@MatthiasHeller, you're welcome. How is this new code related to your post? You should consider a new post with your real problem and all relevant data. I may have a look. In general, depending on the details, there are various ways to perform the computation efficiently; these way might not useParallel
at all, but ratherCompile
d code.
$endgroup$
– Henrik Schumacher
yesterday
add a comment |
$begingroup$
Because managing write access to shared memory is expensive: Subprocesses have to wait until they are granted write access (because another process uses that ressource).
Moreover, it is in general more efficient to use Parallel
only upon the most outer loop construct.
By the way: Using Append
and AppendTo
are the worst methods to build a list, because they involve a copy of the full list each time another element is appended. Instead of complexity $O(n)$ for a list of $n$ elements, you get an implementation of complexity $O(n^2)$. Better use Table
or, if you don't know how long the list is about to get, use Sow
and Reap
. Internal`Bag
is a further option, and it is even compilable.
$endgroup$
Because managing write access to shared memory is expensive: Subprocesses have to wait until they are granted write access (because another process uses that ressource).
Moreover, it is in general more efficient to use Parallel
only upon the most outer loop construct.
By the way: Using Append
and AppendTo
are the worst methods to build a list, because they involve a copy of the full list each time another element is appended. Instead of complexity $O(n)$ for a list of $n$ elements, you get an implementation of complexity $O(n^2)$. Better use Table
or, if you don't know how long the list is about to get, use Sow
and Reap
. Internal`Bag
is a further option, and it is even compilable.
edited yesterday
answered yesterday
Henrik SchumacherHenrik Schumacher
60k582168
60k582168
$begingroup$
Thanks, that actually helped a lot. I just dont understand how to use Sow and Reap to avoid Append To be more specific: instead of ParallelDo I use now ParallelTable: eq = ParallelTable[ FNumeric[ SetPrecision[N[monlistnumeric[[i]] + monlistnumeric[[j]], 20], 10]] , {j, jj}]; FNumeric is a function, that returns either 0 or a value I want to store. I then do eq = DeleteCases[eq, 0]; candidates = Join[candidates, eq]; Is there a more efficient way to do this?
$endgroup$
– Matthias Heller
yesterday
2
$begingroup$
@MatthiasHeller, you're welcome. How is this new code related to your post? You should consider a new post with your real problem and all relevant data. I may have a look. In general, depending on the details, there are various ways to perform the computation efficiently; these way might not useParallel
at all, but ratherCompile
d code.
$endgroup$
– Henrik Schumacher
yesterday
add a comment |
$begingroup$
Thanks, that actually helped a lot. I just dont understand how to use Sow and Reap to avoid Append To be more specific: instead of ParallelDo I use now ParallelTable: eq = ParallelTable[ FNumeric[ SetPrecision[N[monlistnumeric[[i]] + monlistnumeric[[j]], 20], 10]] , {j, jj}]; FNumeric is a function, that returns either 0 or a value I want to store. I then do eq = DeleteCases[eq, 0]; candidates = Join[candidates, eq]; Is there a more efficient way to do this?
$endgroup$
– Matthias Heller
yesterday
2
$begingroup$
@MatthiasHeller, you're welcome. How is this new code related to your post? You should consider a new post with your real problem and all relevant data. I may have a look. In general, depending on the details, there are various ways to perform the computation efficiently; these way might not useParallel
at all, but ratherCompile
d code.
$endgroup$
– Henrik Schumacher
yesterday
$begingroup$
Thanks, that actually helped a lot. I just dont understand how to use Sow and Reap to avoid Append To be more specific: instead of ParallelDo I use now ParallelTable: eq = ParallelTable[ FNumeric[ SetPrecision[N[monlistnumeric[[i]] + monlistnumeric[[j]], 20], 10]] , {j, jj}]; FNumeric is a function, that returns either 0 or a value I want to store. I then do eq = DeleteCases[eq, 0]; candidates = Join[candidates, eq]; Is there a more efficient way to do this?
$endgroup$
– Matthias Heller
yesterday
$begingroup$
Thanks, that actually helped a lot. I just dont understand how to use Sow and Reap to avoid Append To be more specific: instead of ParallelDo I use now ParallelTable: eq = ParallelTable[ FNumeric[ SetPrecision[N[monlistnumeric[[i]] + monlistnumeric[[j]], 20], 10]] , {j, jj}]; FNumeric is a function, that returns either 0 or a value I want to store. I then do eq = DeleteCases[eq, 0]; candidates = Join[candidates, eq]; Is there a more efficient way to do this?
$endgroup$
– Matthias Heller
yesterday
2
2
$begingroup$
@MatthiasHeller, you're welcome. How is this new code related to your post? You should consider a new post with your real problem and all relevant data. I may have a look. In general, depending on the details, there are various ways to perform the computation efficiently; these way might not use
Parallel
at all, but rather Compile
d code.$endgroup$
– Henrik Schumacher
yesterday
$begingroup$
@MatthiasHeller, you're welcome. How is this new code related to your post? You should consider a new post with your real problem and all relevant data. I may have a look. In general, depending on the details, there are various ways to perform the computation efficiently; these way might not use
Parallel
at all, but rather Compile
d code.$endgroup$
– Henrik Schumacher
yesterday
add a comment |
Matthias Heller is a new contributor. Be nice, and check out our Code of Conduct.
Matthias Heller is a new contributor. Be nice, and check out our Code of Conduct.
Matthias Heller is a new contributor. Be nice, and check out our Code of Conduct.
Matthias Heller is a new contributor. Be nice, and check out our Code of Conduct.
Thanks for contributing an answer to Mathematica Stack Exchange!
- Please be sure to answer the question. Provide details and share your research!
But avoid …
- Asking for help, clarification, or responding to other answers.
- Making statements based on opinion; back them up with references or personal experience.
Use MathJax to format equations. MathJax reference.
To learn more, see our tips on writing great answers.
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
StackExchange.ready(
function () {
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fmathematica.stackexchange.com%2fquestions%2f195006%2fwhy-is-paralleldo-slower-than-do%23new-answer', 'question_page');
}
);
Post as a guest
Required, but never shown
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
Sign up or log in
StackExchange.ready(function () {
StackExchange.helpers.onClickDraftSave('#login-link');
});
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
Required, but never shown
$begingroup$
I reverted your post to before the edit because it looks like a different question (like Henrik said in his comment). Note, however, that if you ask it in precisely such form it will be likely closed due to not enough info: you need to provide the minimal working example, not through some undefined functions into a piece of code that no one will be able to run and test.
$endgroup$
– corey979
yesterday
1
$begingroup$
See here mathematica.stackexchange.com/a/48296/12 I suggest you don't use SetSharedVariable until you get quite fluent in using the parallel tools. It effectively "unparallelizes" your code.
$endgroup$
– Szabolcs
yesterday