I have access to a dual processor machine, so I thought I'd try out the
new pbuild stuff in 3.0.1. Unfortunately, exporting SAGE_BUILD_THREADS=3
and SAGE_PBUILD=yes resulted in a *slower* build:
real205m21.375s
user184m45.427s
sys 24m50.515s
was at the bottom after running `m
Hi John,
thanks for your answer. Maybe it's of no relevance for the described
problem but I noticed the following anyway:
In procs/interface.h, I have
...
#include
#define bigfloat RR
...
#include
typedef complex CC;
#define bigcomplex CC
...
inline RR to_bigfloat(const int& n) {return to_RR(
On May 6, 10:05 am, Dan Drake <[EMAIL PROTECTED]> wrote:
Hi Dan
> I have access to a dual processor machine, so I thought I'd try out the
> new pbuild stuff in 3.0.1. Unfortunately, exporting SAGE_BUILD_THREADS=3
> and SAGE_PBUILD=yes resulted in a *slower* build:
>
> real 205m21.375s
> u
On Tue, 06 May 2008 at 03:07AM -0700, mabshoff wrote:
> IIRC you also saw quiet odd things happening with ptest.
That was on a different machine. I still have an account at the
University of Minnesota and am ssh'ed into a computer there. (It happens
to be the computer in my old office. :)
> But e
Hi John (cc: sage-devel),
On Tue, May 6, 2008 at 12:58 AM, John Cremona <[EMAIL PROTECTED]> wrote:
> That certainly looks like a bug in my code. Please can you send me a
> complete example so I can track it down?
Please see
http://trac.sagemath.org/sage_trac/ticket/3111
> You may well be
That certainly merits a blog post somewhere - ?
On May 5, 2:02 pm, [EMAIL PROTECTED] wrote:
> My computation of bernoulli(10^7+4) using GP version 2.3.3 has completed in
> 217417011 miliseconds. That's about 2 days, 12 hours. Anybody know how I
> can print the thing to file?
>
> Machine:
>
On May 6, 3:41 pm, Dan Drake <[EMAIL PROTECTED]> wrote:
> On Tue, 06 May 2008 at 03:07AM -0700, mabshoff wrote:
> > IIRC you also saw quiet odd things happening with ptest.
Hi Dan,
> That was on a different machine. I still have an account at the
> University of Minnesota and am ssh'ed into a co
On May 6, 2008, at 12:53 PM, mhampton wrote:
>
> That certainly merits a blog post somewhere - ?
>
> On May 5, 2:02 pm, [EMAIL PROTECTED] wrote:
>> My computation of bernoulli(10^7+4) using GP version 2.3.3 has
>> completed in 217417011 miliseconds. That's about 2 days, 12
>> hours. Anybod
William has mentioned some congruence tests that we can perform -- I'd like to
make sure that I got the right answer before we pat ourselves on the back too
much.
On Tue, 6 May 2008, mhampton wrote:
>
> That certainly merits a blog post somewhere - ?
>
> On May 5, 2:02 pm, [EMAIL PROTECTED] w
On Tue, May 6, 2008 at 10:15 AM, <[EMAIL PROTECTED]> wrote:
>
> William has mentioned some congruence tests that we can perform -- I'd like
> to make sure that I got the right answer before we pat ourselves on the back
> too much.
>
>
David Harvey's congruence tests would be pretty good. Jus
Oskar,
I think you are probably right there. Of course that code is never
executed in practice since the way the code is organised this function
is only called when disc is negative, but of course that is not going
to keep the compiler happy. So the line
return to_bigfloat(0);
should be repla
Dan's original question was "why was it not doing a parallel build"?
I asked the same thing a few days ago, when the answer given was the
PBUILD does not mean "parallel build" at all. It does a normal build
that (perhaps) lets you do stuff in parallel later. But when, or
what, or why, I don't kn
On May 6, 2008, at 1:18 PM, William Stein wrote:
>
> On Tue, May 6, 2008 at 10:15 AM, <[EMAIL PROTECTED]> wrote:
>>
>> William has mentioned some congruence tests that we can perform
>> -- I'd like to make sure that I got the right answer before we pat
>> ourselves on the back too much.
>>
On Tue, May 6, 2008 at 10:57 AM, David Harvey <[EMAIL PROTECTED]> wrote:
>
>
>
> On May 6, 2008, at 1:18 PM, William Stein wrote:
>
> >
> > On Tue, May 6, 2008 at 10:15 AM, <[EMAIL PROTECTED]> wrote:
> >>
> >> William has mentioned some congruence tests that we can perform
> >> -- I'd like
I think a blog post with PARI timings and then timings for a modular
dsage approach would be cool.
--Mike
On Tue, May 6, 2008 at 11:08 AM, William Stein <[EMAIL PROTECTED]> wrote:
>
>
> On Tue, May 6, 2008 at 10:57 AM, David Harvey <[EMAIL PROTECTED]> wrote:
> >
> >
> >
> > On May 6, 2008,
On May 6, 7:55 pm, "John Cremona" <[EMAIL PROTECTED]> wrote:
> Dan's original question was "why was it not doing a parallel build"?
Yes, pbuild builds only the Sage library in parallel. You asked the
same question for 3.0.1.alpha1 and I gave you the answer in
https://groups.google.com/group/sage
On May 6, 2008, at 2:12 PM, Mike Hansen wrote:
> I think a blog post with PARI timings and then timings for a modular
> dsage approach would be cool.
Probably not so cool, since it would be like 50 machines vs one machine.
david
--~--~-~--~~~---~--~~
To post t
> Probably not so cool, since it would be like 50 machines vs one machine.
Sure, but the Mathematica blog post is scalablity: "In Mathematica, a
core principle is that everything should be scalable. So in my job of
creating algorithms for Mathematica I have to make sure that
everything I produce
I agree, I think demonstrating a distributed algorithm would be very
cool. From what I can tell of processor trends, we won't see enormous
gains in speed but we might see an awful lot of processors (like
Intel's prototype 80-core chip).
On May 6, 12:19 pm, "Mike Hansen" <[EMAIL PROTECTED]> wrote
On May 6, 2008, at 2:19 PM, Mike Hansen wrote:
>> Probably not so cool, since it would be like 50 machines vs one
>> machine.
>
> Sure, but the Mathematica blog post is scalablity: "In Mathematica, a
> core principle is that everything should be scalable. So in my job of
> creating algorithms
Michael,
I hope you realised that I was only criticising my own lamentable lack
of understanding and not your explanations. I think that the number
of drinks I'll be buying you next time we meet must be into double
figures by now.
For most of us, even though we build all or most of the alphas a
Hi John,
*JUST FOR TESTING OUT MY PREVIOUS SUSPICION*, I changed to
bigcomplex cubic::hess_root() const
{
bigfloat discr = I2bigfloat(disc());
// if(!is_positive(disc()))
//{
// cout<<"Error: hess_root called with negative dicriminant!\n";
// return to_bigfloat(0);
//}
bi
On Tue, May 6, 2008 at 11:55 AM, David Harvey <[EMAIL PROTECTED]> wrote:
>
>
> On May 6, 2008, at 2:19 PM, Mike Hansen wrote:
>
> >> Probably not so cool, since it would be like 50 machines vs one
> >> machine.
> >
> > Sure, but the Mathematica blog post is scalablity: "In Mathematica, a
> >
Oskar,
I'm sure these are all related. I did not mean to imply that your
compiler was broken! Only that as I myself only have one, the only
testing I have done is with that one; and since Sage has started
shipping with eclib (which is about 6 months) I know that lots of
other compilers out the
Try building with SAGE_BUIILD_THREADS=2. If you are building with 3
threads on a 2 cpu system you could be seeing some type of
cache/scheduler issue as pbuild fully saturates the threads it
launches to 100% cpu usage.
On Tue, May 6, 2008 at 2:05 AM, Dan Drake <[EMAIL PROTECTED]> wrote:
> I have
On May 6, 11:48 pm, "Gary Furnish" <[EMAIL PROTECTED]> wrote:
> Try building with SAGE_BUIILD_THREADS=2. If you are building with 3
> threads on a 2 cpu system you could be seeing some type of
> cache/scheduler issue as pbuild fully saturates the threads it
> launches to 100% cpu usage.
I *ser
I have been trying out Cython in order to wrap a C++ library (Frobby)
for Sage, and I had a bit of trouble getting things set up. With help
from #sage-devel I have managed to get a minimal example running now,
and I promised to describe what I came up with, so I've packaged it up
as a very simple
On May 6, 11:59 pm, "Franco Saliola" <[EMAIL PROTECTED]> wrote:
> Not sure if this has been reported.
Hi Franco,
> Compilation failed with the message:
>
> sage: An error occurred while installing ntl-5.4.2.p2
Your gcc is way too old and know buggy:
G_LLL_QP.c: In function `long int NTL::G_B
On Tue, May 6, 2008 at 6:06 PM, mabshoff <[EMAIL PROTECTED]> wrote:
> Your gcc is way too old and know buggy:
Ah, the problem is that these machines haven't been updated in a
while. (I'm trying to compile on machines in a computer lab.)
> > I'm attaching the install.log.
>
> Please don't do
On May 7, 12:03 am, "Franco Saliola" <[EMAIL PROTECTED]> wrote:
> Me again. Different machine, same OS, different error.
>
> Compilation failed with the message:
>
> sage: An error occurred while installing lapack-20071123.p0
Hhhm:
gcc version 4.0.1 (Apple Computer, Inc. build 5341)
*
Thanks!
On May 6, 4:55 pm, Bjake Hammersholt Roune <[EMAIL PROTECTED]>
wrote:
> I have been trying out Cython in order to wrap a C++ library (Frobby)
> for Sage, and I had a bit of trouble getting things set up. With help
> from #sage-devel I have managed to get a minimal example running now,
> a
On Tue, 06 May 2008 at 03:07AM -0700, mabshoff wrote:
> IIRC you also saw quiet odd things happening with ptest.
As I said, that was on another machine...but now I'm getting strange
behavior again!
In an earlier email I said that 'make test' worked fine with the usual
build. I decided to try a pb
That's exactly what happened to me.
John
On 07/05/2008, Dan Drake <[EMAIL PROTECTED]> wrote:
> On Tue, 06 May 2008 at 03:07AM -0700, mabshoff wrote:
>
> > IIRC you also saw quiet odd things happening with ptest.
>
>
> As I said, that was on another machine...but now I'm getting strange
> behavi
On May 7, 8:48 am, "John Cremona" <[EMAIL PROTECTED]> wrote:
Dan, John,
> That's exactly what happened to me.
That is #3097. A patch and a work around exits, but it was too late in
the dev cycle to apply it since in its current form it will break on -
sdist. I have a fix for that, so it will be
34 matches
Mail list logo
