diff mbox

gnu: lapack: Build the LAPACKE library.

Message ID 1464184325-14056-1-git-send-email-dthompson2@worcester.edu
State New
Headers show

Commit Message

David Thompson May 25, 2016, 1:52 p.m. UTC
From: David Thompson <davet@gnu.org>

* gnu/packages/maths.scm (lapack)[arguments]: Add "-DLAPACKE=ON".
---
 gnu/packages/maths.scm | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

Comments

Leo Famulari May 27, 2016, 5:08 p.m. UTC | #1
On Wed, May 25, 2016 at 09:52:05AM -0400, David Thompson wrote:
> From: David Thompson <davet@gnu.org>
> 
> * gnu/packages/maths.scm (lapack)[arguments]: Add "-DLAPACKE=ON".

Builds for me on x86_64.
Eric Bavier May 27, 2016, 5:55 p.m. UTC | #2
On 2016-05-27 12:08, Leo Famulari wrote:
> On Wed, May 25, 2016 at 09:52:05AM -0400, David Thompson wrote:
>> From: David Thompson <davet@gnu.org>
>> 
>> * gnu/packages/maths.scm (lapack)[arguments]: Add "-DLAPACKE=ON".
> 
> Builds for me on x86_64.

How much does this add to the size of the lapack package?  Would we want 
to install the libraries and headers into another output?  IFAIK, not 
many codes make use of both the standard interface and LAPACKE at the 
same time.
David Thompson June 3, 2016, 4:39 p.m. UTC | #3
On Fri, May 27, 2016 at 1:55 PM, Eric Bavier <ericbavier@openmailbox.org> wrote:
> On 2016-05-27 12:08, Leo Famulari wrote:
>>
>> On Wed, May 25, 2016 at 09:52:05AM -0400, David Thompson wrote:
>>>
>>> From: David Thompson <davet@gnu.org>
>>>
>>> * gnu/packages/maths.scm (lapack)[arguments]: Add "-DLAPACKE=ON".
>>
>>
>> Builds for me on x86_64.
>
> How much does this add to the size of the lapack package?  Would we want to
> install the libraries and headers into another output?  IFAIK, not many
> codes make use of both the standard interface and LAPACKE at the same time.

Before:

store item                                                       total    self
/gnu/store/dkz7fips64c5h05hfg6lmvhmb7jlipb6-lapack-3.5.0
71.4     6.4   9.0%
/gnu/store/gaximh48qbqf4712bw224ghq919hqp64-gfortran-4.9.3-lib
64.9    27.1  38.0%
/gnu/store/8m00x5x8ykmar27s9248cmhnkdb2n54a-glibc-2.22
37.9    36.5  51.1%
/gnu/store/5y72ql4zbnbd50lf88k91k0y766kq75x-bash-static-4.3.42
1.4     1.4   2.0%
total: 71.4 MiB

After:

store item                                                       total    self
/gnu/store/1whq33kw6136h4jcy14cfk1klx5hn1s3-lapack-3.5.0
97.2     9.4   9.6%
/gnu/store/gaximh48qbqf4712bw224ghq919hqp64-gfortran-4.9.3-lib
64.9    27.1  27.9%
/gnu/store/v39bh3ln3ncnzhyw0kd12d46kww9747v-gcc-4.9.3-lib
60.7    22.9  23.5%
/gnu/store/8m00x5x8ykmar27s9248cmhnkdb2n54a-glibc-2.22
37.9    36.5  37.5%
/gnu/store/5y72ql4zbnbd50lf88k91k0y766kq75x-bash-static-4.3.42
1.4     1.4   1.4%
total: 97.2 MiB

Basically, closure size goes up by 20MiB, but the store item itself is
only 3MiB bigger.

WDYT?

- Dave
diff mbox

Patch

diff --git a/gnu/packages/maths.scm b/gnu/packages/maths.scm
index 19ed44f..c844db2 100644
--- a/gnu/packages/maths.scm
+++ b/gnu/packages/maths.scm
@@ -333,7 +333,8 @@  large scale eigenvalue problems.")
     (inputs `(("fortran" ,gfortran)
               ("python" ,python-2)))
     (arguments
-     `(#:configure-flags '("-DBUILD_SHARED_LIBS:BOOL=YES")
+     `(#:configure-flags '("-DBUILD_SHARED_LIBS:BOOL=YES"
+                           "-DLAPACKE=ON")
        #:phases (alist-cons-before
                  'check 'patch-python
                  (lambda* (#:key inputs #:allow-other-keys)