8:05 a.m.
We plan to release gretl 2022c before long, and today's snapshot can
be considered a "release candidate". We'd be grateful if people could
beat on the snapshots (or the git master code) and help us shake out
any remaining bugs before release.
This would also be a good time for translators to make any updates.
Allin Cottrell
8:37 a.m.
On Fri, 30 Sep 2022, Allin Cottrell wrote:
We plan to release gretl 2022c before long, and today's snapshot
can
be considered a "release candidate". We'd be grateful if people could
beat on the snapshots (or the git master code) and help us shake out
any remaining bugs before release.
If I may:
this version (the 2022c candidate) contains vast improvements in the way
user-written functions are handled internally. This should translate into
a sizeable speedup of certain feaures. For example, addons such as gig or
SVAR should be much faster, and the same should happen to function
packages like CUB, DFM, MGARCH or StrucTiSM[*].
We (the development team) tested the new features very extensively for
some time, and we're confident none of your old code should break, but
just in case: if you have gretl code lying around that you can use for
testing, please do so asap and tell us if you see anything strange.
Thenk you.
[*] disclaimer: these are all packages that I co-authored. I'm not trying
to promote my own work here, I'm just mentioning them because I know their
internals well and I feel confident in anticipating that their performance
should be much better.
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------
9:37 a.m.
Am 30.09.2022 um 14:37 schrieb Riccardo (Jack) Lucchetti:
On Fri, 30 Sep 2022, Allin Cottrell wrote:
> We plan to release gretl 2022c before long, and today's snapshot can
> be considered a "release candidate". We'd be grateful if people could
> beat on the snapshots (or the git master code) and help us shake out
> any remaining bugs before release.
If I may:
this version (the 2022c candidate) contains vast improvements in the
way user-written functions are handled internally. This should
translate into a sizeable speedup of certain feaures. For example,
addons such as gig or SVAR should be much faster, and the same should
happen to function packages like CUB, DFM, MGARCH or StrucTiSM[*].
I tested my own MSVAR package on gretl 2022b against today's snapshot,
but for a certain specification I did not see any performance gain, I
must say. If I understand correctly, user-written functions that are
called in a nested way (and/or within a loop) would benefit the most.
Perhaps that's not what the MSVAR hansl code does, I haven't re-checked.
Jack, maybe you could give a concrete example (with our StrucTiSM,
perhaps?) of what kind of gains we're talking about.
thanks
sven
10:31 a.m.
On Fri, 30 Sep 2022, Sven Schreiber wrote:
Jack, maybe you could give a concrete example (with our StrucTiSM,
perhaps?)
of what kind of gains we're talking about.
It really depends on the kind of problem you have and on your hw/sw
combination. For example, I'm attaching a zip file with 3 scripts: one
uses StrucTiSM to estimate a BSM with some exogenous variables, one is a
slight modification of the "Uhlig" example in the SVAR package and the
third one is a Jesús Fernández-Villaverde-style RBC script. On my work PC
(linux with 6 Intel i7-8700 cores) I'm getting the following results:
bikes Uhlig RBC
2022b 1.73981 11.0784 189.335
2022c-git 1.61362 9.97556 56.1038
speed gain 7.8% 11.1% 237.5%
So you can see how different the time savings can be. Numerical results
are, of course, identical.
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------
12:35 p.m.
On Fri, Sep 30, 2022 at 9:37 AM Sven Schreiber
<sven.schreiber(a)fu-berlin.de> wrote:
Am 30.09.2022 um 14:37 schrieb Riccardo (Jack) Lucchetti:
>
> this version (the 2022c candidate) contains vast improvements in the
> way user-written functions are handled internally. This should
> translate into a sizeable speedup of certain feaures. For example,
> addons such as gig or SVAR should be much faster, and the same should
> happen to function packages like CUB, DFM, MGARCH or StrucTiSM[*].
I tested my own MSVAR package on gretl 2022b against today's snapshot,
but for a certain specification I did not see any performance gain,
I've tested the MSVAR sample script, and found a time-saving of about
30 percent with the current code versus 2022b: not super-dramatic, but
noticeable. This was on two i7 machines, one old and one reasonably
current. The proportional speed differential was very similar in the
two cases.
Allin
12:43 p.m.
Am 30.09.2022 um 18:35 schrieb Cottrell, Allin:
On Fri, Sep 30, 2022 at 9:37 AM Sven Schreiber
<sven.schreiber(a)fu-berlin.de> wrote:
> I tested my own MSVAR package on gretl 2022b against today's snapshot,
> but for a certain specification I did not see any performance gain,
I've tested the MSVAR sample script, and found a time-saving of about
30 percent with the current code versus 2022b: not super-dramatic, but
noticeable. This was on two i7 machines, one old and one reasonably
current. The proportional speed differential was very similar in the
two cases.
Hm, thanks Allin, I didn't do very systematic testing, but 30% would
certainly be noticeable. I wonder what happened here on my setup. I will
try again on a different Windows machine and perhaps also pay closer
attention to the results.
cheers
sven
2:40 p.m.
On 30.09.2022 14:37, Riccardo (Jack) Lucchetti wrote:
On Fri, 30 Sep 2022, Allin Cottrell wrote:
> We plan to release gretl 2022c before long, and today's snapshot can
> be considered a "release candidate". We'd be grateful if people could
> beat on the snapshots (or the git master code) and help us shake out
> any remaining bugs before release.
If I may:
this version (the 2022c candidate) contains vast improvements in the
way user-written functions are handled internally. This should
translate into a sizeable speedup of certain feaures. For example,
addons such as gig or SVAR should be much faster, and the same should
happen to function packages like CUB, DFM, MGARCH or StrucTiSM[*].
We (the development team) tested the new features very extensively for
some time, and we're confident none of your old code should break, but
just in case: if you have gretl code lying around that you can use for
testing, please do so asap and tell us if you see anything strange.
Thenk you.
[*] disclaimer: these are all packages that I co-authored. I'm not
trying to promote my own work here, I'm just mentioning them because I
know their internals well and I feel confident in anticipating that
their performance should be much better.
I can confirm a big speedup in case of BMA package: +/- 50%.
Marcin
--
Marcin Błażejowski
740
days inactive
740
days old
6 comments
5 participants
participants (5)
-
Allin Cottrell -
Cottrell, Allin
-
Marcin Błażejowski -
Riccardo (Jack) Lucchetti -
Sven Schreiber