Numpy performance differences depending on numerical values











up vote
5
down vote

favorite
3












I found a strange performance difference while evaluating an expression in Numpy.



I executed the following code:



import numpy as np

myarr = np.random.uniform(-1,1,[1100,1100])


and then 



%timeit np.exp( - 0.5 * (myarr / 0.001)**2 )

>> 184 ms ± 301 µs per loop (mean ± std. dev. of 7 runs, 1 loop each)


and



%timeit np.exp( - 0.5 * (myarr / 0.1)**2 )

>> 12.3 ms ± 34.3 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)


That's an almost 15x faster computation in the second case! Note that the only difference is the factor being 0.1 or 0.001.



What's the reason for this behaviour? Can I change something to make the first calculation as fast as the second?






python performance numpy







share|improve this question






















  • On OSX and NumPy 1.15.3, Python 3.7, I see 11ms vs 13ms.
    – Nils Werner
    3 hours ago












  • In Win7, NumPy 1.15.3, Python 3.6.2, I see 227 vs 22 ms
    – Brenlla
    3 hours ago












  • @jpp I'm using Ubuntu 18.04, Python 3.6.6, Numpy 1.15.4
    – Ethunxxx
    3 hours ago








  • 2




    In my system, exp of large (negative) numbers are slower: exp(-1) is faster than exp(-1000). So it probably comes down to some slower covergence of the exp algorithm with large numbers
    – Brenlla
    3 hours ago






  • 1




    @MattMessersmith Reasonable explanation, but nope. exp(1) is still much faster than exp(1000)
    – Brenlla
    3 hours ago















up vote
5
down vote

favorite
3












I found a strange performance difference while evaluating an expression in Numpy.



I executed the following code:



import numpy as np

myarr = np.random.uniform(-1,1,[1100,1100])


and then 



%timeit np.exp( - 0.5 * (myarr / 0.001)**2 )

>> 184 ms ± 301 µs per loop (mean ± std. dev. of 7 runs, 1 loop each)


and



%timeit np.exp( - 0.5 * (myarr / 0.1)**2 )

>> 12.3 ms ± 34.3 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)


That's an almost 15x faster computation in the second case! Note that the only difference is the factor being 0.1 or 0.001.



What's the reason for this behaviour? Can I change something to make the first calculation as fast as the second?






python performance numpy







share|improve this question






















  • On OSX and NumPy 1.15.3, Python 3.7, I see 11ms vs 13ms.
    – Nils Werner
    3 hours ago












  • In Win7, NumPy 1.15.3, Python 3.6.2, I see 227 vs 22 ms
    – Brenlla
    3 hours ago












  • @jpp I'm using Ubuntu 18.04, Python 3.6.6, Numpy 1.15.4
    – Ethunxxx
    3 hours ago








  • 2




    In my system, exp of large (negative) numbers are slower: exp(-1) is faster than exp(-1000). So it probably comes down to some slower covergence of the exp algorithm with large numbers
    – Brenlla
    3 hours ago






  • 1




    @MattMessersmith Reasonable explanation, but nope. exp(1) is still much faster than exp(1000)
    – Brenlla
    3 hours ago













up vote
5
down vote

favorite
3









up vote
5
down vote

favorite
3






3





I found a strange performance difference while evaluating an expression in Numpy.



I executed the following code:



import numpy as np

myarr = np.random.uniform(-1,1,[1100,1100])


and then 



%timeit np.exp( - 0.5 * (myarr / 0.001)**2 )

>> 184 ms ± 301 µs per loop (mean ± std. dev. of 7 runs, 1 loop each)


and



%timeit np.exp( - 0.5 * (myarr / 0.1)**2 )

>> 12.3 ms ± 34.3 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)


That's an almost 15x faster computation in the second case! Note that the only difference is the factor being 0.1 or 0.001.



What's the reason for this behaviour? Can I change something to make the first calculation as fast as the second?






python performance numpy







share|improve this question













I found a strange performance difference while evaluating an expression in Numpy.



I executed the following code:



import numpy as np

myarr = np.random.uniform(-1,1,[1100,1100])


and then 



%timeit np.exp( - 0.5 * (myarr / 0.001)**2 )

>> 184 ms ± 301 µs per loop (mean ± std. dev. of 7 runs, 1 loop each)


and



%timeit np.exp( - 0.5 * (myarr / 0.1)**2 )

>> 12.3 ms ± 34.3 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)


That's an almost 15x faster computation in the second case! Note that the only difference is the factor being 0.1 or 0.001.



What's the reason for this behaviour? Can I change something to make the first calculation as fast as the second?






python performance numpy




python performance numpy






share|improve this question













share|improve this question











share|improve this question




share|improve this question










asked 4 hours ago









Ethunxxx

359415




359415












  • On OSX and NumPy 1.15.3, Python 3.7, I see 11ms vs 13ms.
    – Nils Werner
    3 hours ago












  • In Win7, NumPy 1.15.3, Python 3.6.2, I see 227 vs 22 ms
    – Brenlla
    3 hours ago












  • @jpp I'm using Ubuntu 18.04, Python 3.6.6, Numpy 1.15.4
    – Ethunxxx
    3 hours ago








  • 2




    In my system, exp of large (negative) numbers are slower: exp(-1) is faster than exp(-1000). So it probably comes down to some slower covergence of the exp algorithm with large numbers
    – Brenlla
    3 hours ago






  • 1




    @MattMessersmith Reasonable explanation, but nope. exp(1) is still much faster than exp(1000)
    – Brenlla
    3 hours ago


















  • On OSX and NumPy 1.15.3, Python 3.7, I see 11ms vs 13ms.
    – Nils Werner
    3 hours ago












  • In Win7, NumPy 1.15.3, Python 3.6.2, I see 227 vs 22 ms
    – Brenlla
    3 hours ago












  • @jpp I'm using Ubuntu 18.04, Python 3.6.6, Numpy 1.15.4
    – Ethunxxx
    3 hours ago








  • 2




    In my system, exp of large (negative) numbers are slower: exp(-1) is faster than exp(-1000). So it probably comes down to some slower covergence of the exp algorithm with large numbers
    – Brenlla
    3 hours ago






  • 1




    @MattMessersmith Reasonable explanation, but nope. exp(1) is still much faster than exp(1000)
    – Brenlla
    3 hours ago
















On OSX and NumPy 1.15.3, Python 3.7, I see 11ms vs 13ms.
– Nils Werner
3 hours ago






On OSX and NumPy 1.15.3, Python 3.7, I see 11ms vs 13ms.
– Nils Werner
3 hours ago














In Win7, NumPy 1.15.3, Python 3.6.2, I see 227 vs 22 ms
– Brenlla
3 hours ago






In Win7, NumPy 1.15.3, Python 3.6.2, I see 227 vs 22 ms
– Brenlla
3 hours ago














@jpp I'm using Ubuntu 18.04, Python 3.6.6, Numpy 1.15.4
– Ethunxxx
3 hours ago






@jpp I'm using Ubuntu 18.04, Python 3.6.6, Numpy 1.15.4
– Ethunxxx
3 hours ago






2




2




In my system, exp of large (negative) numbers are slower: exp(-1) is faster than exp(-1000). So it probably comes down to some slower covergence of the exp algorithm with large numbers
– Brenlla
3 hours ago




In my system, exp of large (negative) numbers are slower: exp(-1) is faster than exp(-1000). So it probably comes down to some slower covergence of the exp algorithm with large numbers
– Brenlla
3 hours ago




1




1




@MattMessersmith Reasonable explanation, but nope. exp(1) is still much faster than exp(1000)
– Brenlla
3 hours ago




@MattMessersmith Reasonable explanation, but nope. exp(1) is still much faster than exp(1000)
– Brenlla
3 hours ago

















active

oldest

votes











Your Answer






StackExchange.ifUsing("editor", function () {
StackExchange.using("externalEditor", function () {
StackExchange.using("snippets", function () {
StackExchange.snippets.init();
});
});
}, "code-snippets");

StackExchange.ready(function() {
var channelOptions = {
tags: "".split(" "),
id: "1"
};
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',
convertImagesToLinks: true,
noModals: true,
showLowRepImageUploadWarning: true,
reputationToPostImages: 10,
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
});


}
});














 

draft saved


draft discarded


















StackExchange.ready(
function () {
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fstackoverflow.com%2fquestions%2f53415531%2fnumpy-performance-differences-depending-on-numerical-values%23new-answer', 'question_page');
}
);

Post as a guest















Required, but never shown






























active

oldest

votes













active

oldest

votes









active

oldest

votes






active

oldest

votes
















 

draft saved


draft discarded



















































 


draft saved


draft discarded














StackExchange.ready(
function () {
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fstackoverflow.com%2fquestions%2f53415531%2fnumpy-performance-differences-depending-on-numerical-values%23new-answer', 'question_page');
}
);

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







-

Popular posts from this blog

What visual should I use to simply compare current year value vs last year in Power BI desktop

Image
0 I have column SubToBind that shows current year ratio and column PY SubToBind which shows previous year ratio. How can I simply create a card that would be displaying them both in a nice way? Something like that: I cannot achieve that with regular card, I also tried Card with States by OKViz but also unable to see both values. In KPI visual - I dont like the word "Goal", it will confuse end users. Are there any cards or other visuals that would do the job? Thanks powerbi dax powerbi-embedded share | improve this question edited Jun 11 '18 at 20:11 Oleg ...
Read more

How to ignore python UserWarning in pytest?

Image
up vote 1 down vote favorite I am using openpyxl for parsing .xlsm files, and pytest for testing. When I open a file, I get the: OpenPyxl -> UserWarning: Data Validation extension is not supported and will be removed That is not really a problem bcs. program works and I can't change the .xlsm file to fix that. But... When I run pytest with something like: def test_wrong_file_format(): assert check_excel(open_excel('file.xlsm')) == True I will get the warning i mentioned altought check_excel(open_excel('file.xlsm')) returns True and the test should suceed... Is there a nice way to tell the pytest that "It's not a bug it's a feature" and tests should pass even when they get this warning? Is there other way than using something like: with pytest.warns(UserWarning): ...
Read more

Alexandru Averescu

Image
Alexandre Averescu Alexandru Averescu Naissance 3 avril 1859 Babele, Ismail  (ro) près de Izmaïl (principauté de Moldavie) Décès 2 octobre 1938 (à 79 ans) [ 1 ] Bucarest Origine   Roumanie Arme cavalerie Grade Maréchal Années de service 1901-1917 Commandement chef d'état-major général 2 e  armée 3 e  armée Groupe d'armée sud armée du Danube Conflits Guerre russo-turque de 1877-1878 Deuxième guerre balkanique Première Guerre mondiale Faits d'armes Bataille de Turtucaia/Tutrakan Bataille de Mărăști Distinctions Autres fonctions 3 fois président du Conseil des ministres ministre des Affaires étrangères ministre sans portefeuille conseiller de la Couronne modifier   Alexandre Averescu (en roumain : Alexandru Averescu .mw-parser-output .prononciation>a{background:url("//upload.wikimedia.org/wikipedia/commons/thumb/8/8a/Loudspeaker.svg/11px-Loudspeaker.svg.png")center ...
Read more