[tex-live] strange discrepancy in running time of etex between TL2015 and TL2017

[jfbu]

Hi,

Le 27 juil. 2017 at 17:26, George N. White III <gnwiii at gmail.com> wrote :

> On 27 July 2017 at 09:44, jfbu <jfbu at free.fr> wrote:
> [...] 
> Hence now no \numexpr, only Knuth tex
> 
> When I do
> 
> time tex testspeed.tex
> 
> I consistently get on my iMac 10.9.5:
> 
> - with TL2015: about 1.62s (last digit non-significant)
> 
> - with TL2017: about 2.18s (idem)
> 
> Thus, it appears, but somehow I had overlooked it so far,
> that darwinlegacy tex (et al.) is at least 33% slower on TL2017
> than formerly
> 
> 
> I have iMac 10.11.6 with 4 TeX Live systems (current macports, TL2015--TL2017) and 
> checkX.sh scripts that format the same set of basic files on each.
> 
> After several iterations, times are reasonably stable.  The last run gave:
> 
>               macports               TL2015             TL2016              TL2017
> 
> real	 0m2.628s    0m3.901s    0m4.610s    0m4.587s
> 
> user 	0m2.156s    0m3.352s    0m4.085s    0m3.996s 
> 
> sys	  0m0.313s    0m0.343s    0m0.368s    0m0.378s
> 
> 
> 
> There is a big gain to had from compiling locally over the generic TL builds, and not much change from TL2016 to TL2017.
> 

This is very interesting thanks.

It dispels somewhat my obsession about the specificity of darwinlegacy.

The difference with the macports binaries is impressive and unexpected to me.

And the change from TL2015 to TL2017 is at least 10% which again
surprises me. However it could be inside the margin of fluctuation
from run to run perhaps.

I am very much interested now into compiling locally, but I will be off internet
for a while so I will have to think by myself how to achieve that after having
rsync'ed the TeXLive sources, and thinking by myself might prove hard.

Best,

Jean-Francois

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://tug.org/pipermail/tex-live/attachments/20170727/2afef17f/attachment.html>