Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > comp.soft-sys.math.mathematica > #10056 > unrolled thread

Is Mathematica v8 slower than v7 ?

Started byperplexed <yudumbo@gmail.com>
First post2012-05-13 07:02 +0000
Last post2012-05-19 09:50 +0000
Articles 4 — 3 participants

Back to article view | Back to comp.soft-sys.math.mathematica


Contents

  Is Mathematica v8 slower than v7 ? perplexed <yudumbo@gmail.com> - 2012-05-13 07:02 +0000
    Re: Is Mathematica v8 slower than v7 ? David Bailey <dave@removedbailey.co.uk> - 2012-05-16 08:27 +0000
      Re: Is Mathematica v8 slower than v7 ? perplexed <yudumbo@gmail.com> - 2012-05-18 09:22 +0000
    Re: Is Mathematica v8 slower than v7 ? Rodrigo AP <digorna@gmail.com> - 2012-05-19 09:50 +0000

#10056 — Is Mathematica v8 slower than v7 ?

Fromperplexed <yudumbo@gmail.com>
Date2012-05-13 07:02 +0000
SubjectIs Mathematica v8 slower than v7 ?
Message-ID<jonm9a$gdk$1@smc.vnet.net>
I have tried Mathematica 8.0.4 for Windows on a laptop
running Vista Ultimate 64-bit with a T9400 (at 80% speed).
There was a previous version (7.x.x) that I have uninstalled
(so I cannot do more tests...).

Anyway, before uninstalling, since I was curious about the
relative performance of the two versions, I have saved a notebook
with a few line of code (each repeated at least 3 times).

Here are two examples:
Timing[Select[Range[5 10^5], EulerPhi[#] + # == DivisorSigma[1, #] &]]
(* v7 about 12 seconds,  v8  13.4 seconds, i.e. about 10% slower *)

Timing[Select[Range[12], Reduce[x^2 - # y^2 == 1 + x y && x > 0 && y > 0, 
{x, y}, Integers] === False &]]
(* v7 about 10.1 seconds, v8  14 seconds,  about 35% slower *)

In another case v8 was much faster:
Timing[Sum[N[1/x] Log[x], {x, 2, 10^6}]]
(* v7 about 7.6 seconds, v8 about 0.2 seconds i.e.
about 35 times faster *)

To be honest, my interest in Windows version is marginal (the laptop
is just for emergencies) since our official Mathematica 
licence is for linux, but I'm still wondering if upgrading to v8
can be useful even if one is not interested in all the new
features.

thanks

[toc] | [next] | [standalone]


#10093

FromDavid Bailey <dave@removedbailey.co.uk>
Date2012-05-16 08:27 +0000
Message-ID<jovoeb$mjg$1@smc.vnet.net>
In reply to#10056
On 13/05/2012 08:02, perplexed wrote:
> I have tried Mathematica 8.0.4 for Windows on a laptop
> running Vista Ultimate 64-bit with a T9400 (at 80% speed).
> There was a previous version (7.x.x) that I have uninstalled
> (so I cannot do more tests...).
>
> Anyway, before uninstalling, since I was curious about the
> relative performance of the two versions, I have saved a notebook
> with a few line of code (each repeated at least 3 times).
>
> Here are two examples:
> Timing[Select[Range[5 10^5], EulerPhi[#] + # == DivisorSigma[1, #]&]]
> (* v7 about 12 seconds,  v8  13.4 seconds, i.e. about 10% slower *)
>
> Timing[Select[Range[12], Reduce[x^2 - # y^2 == 1 + x y&&  x>  0&&  y>  0,
> {x, y}, Integers] === False&]]
> (* v7 about 10.1 seconds, v8  14 seconds,  about 35% slower *)
>
> In another case v8 was much faster:
> Timing[Sum[N[1/x] Log[x], {x, 2, 10^6}]]
> (* v7 about 7.6 seconds, v8 about 0.2 seconds i.e.
> about 35 times faster *)
>
> To be honest, my interest in Windows version is marginal (the laptop
> is just for emergencies) since our official Mathematica
> licence is for linux, but I'm still wondering if upgrading to v8
> can be useful even if one is not interested in all the new
> features.
>
> thanks
>
I don't think there is much point comparing the speed of symbolic 
calculations from version to version, because later versions may contain 
extra rigorous checks - this is certainly the case with certain symbolic 
definite integrals.

Changes in speed will obviously vary depending on what exactly you test.

In general, I would imagine you would see the same relative speed 
changes under Linux as you do under Windows (assuming both systems have 
adequate memory) because almost all the code will be the same.

David Bailey
http://www.dbaileyconsultancy.co.uk

[toc] | [prev] | [next] | [standalone]


#10119

Fromperplexed <yudumbo@gmail.com>
Date2012-05-18 09:22 +0000
Message-ID<jp54d6$gks$1@smc.vnet.net>
In reply to#10093
On May 16, 10:27 am, David Bailey <d...@removedbailey.co.uk> wrote:

> I don't think there is much point comparing the speed of symbolic
> calculations from version to version, because later versions may contain
> extra rigorous checks - this is certainly the case with certain symbolic
> definite integrals.

On the contrary, I believe there is a very strong point, at least for
me.

Mathematica, of which I'm a great fan, has a lot of functions
in different fields.

Assume that I'm mainly interested in, say, elementary number theory,
while I never use statistics or differential equations (or Wolfram
Alpha...).

Then, if none of the new features of v8 have an impact on what I do,
and on the contrary the functions I use more often are slightly
slower,
then I should think twice about upgrading from v7 to v8.

[toc] | [prev] | [next] | [standalone]


#10143

FromRodrigo AP <digorna@gmail.com>
Date2012-05-19 09:50 +0000
Message-ID<jp7qch$qig$1@smc.vnet.net>
In reply to#10056
I have a personal experience: the command Factor[] is much more slow
in Mathematica 7.0 than in Mathematica 5.2. The difference reaches
10x!

2012/5/18, perplexed <yudumbo@gmail.com>:
> On May 16, 10:27 am, David Bailey <d...@removedbailey.co.uk> wrote:
>
>> I don't think there is much point comparing the speed of symbolic
>> calculations from version to version, because later versions may contain
>> extra rigorous checks - this is certainly the case with certain symbolic
>> definite integrals.
>
> On the contrary, I believe there is a very strong point, at least for
> me.
>
> Mathematica, of which I'm a great fan, has a lot of functions
> in different fields.
>
> Assume that I'm mainly interested in, say, elementary number theory,
> while I never use statistics or differential equations (or Wolfram
> Alpha...).
>
> Then, if none of the new features of v8 have an impact on what I do,
> and on the contrary the functions I use more often are slightly
> slower,
> then I should think twice about upgrading from v7 to v8.
>
>


-- 
Rodrigo

[toc] | [prev] | [standalone]


Back to top | Article view | comp.soft-sys.math.mathematica


csiph-web