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

Groups > comp.lang.forth > #44871 > unrolled thread

New Gforth snapshot

Back to article view | Back to comp.lang.forth

Contents

  New Gforth snapshot Bernd Paysan <bernd.paysan@gmx.de> - 2016-01-14 15:49 +0100
    Re: New Gforth snapshot anton@mips.complang.tuwien.ac.at (Anton Ertl) - 2016-01-14 15:07 +0000
      Re: New Gforth snapshot Bernd Paysan <bernd.paysan@gmx.de> - 2016-01-15 00:45 +0100
    Re: New Gforth snapshot Bernd Paysan <bernd.paysan@gmx.de> - 2016-01-16 03:45 +0100

#44871 — New Gforth snapshot

Version: 0.7.9-20160113, due to a change of the Debian version name, it's no 
longer in alphabetical order...

https://www.complang.tuwien.ac.at/forth/gforth/Snapshots/

If you use Debian, you can also get it through my Debian repository, set up 
to make net2o easier to install, see here:

https://fossil.net2o.de/net2o/doc/trunk/wiki/get-it.md

The new stuff is mostly build process changes, and makes it more straight-
forward to have a multi-arch installation.  You can then call gforth-386 for 
the 32 bit version, and gforth-amd64 for the 64 bit version (or gforth-arm 
and gforth-arm64 on an ARM64 system); the last install defines which the 
generic non-architecture symlink is used.  Library build is now enabled by 
default, as it now builds on all plattforms I'm currently generating 
snapshots for.  According to the benchmarks, the library is as fast as the 
non-library executable, so the next change might be to remove the normal, 
non-FPIC build, and just install the small wrapper application that calls 
gforth_main from the library.

Debian armhf builds will probably follow today (the slower the machine, the 
later the build ;-).

-- 
Bernd Paysan
"If you want it done right, you have to do it yourself"
net2o ID: kQusJzA;7*?t=uy@X}1GWr!+0qqp_Cn176t4(dQ*
http://bernd-paysan.de/

[toc] | [next] | [standalone]

#44872

Bernd Paysan <bernd.paysan@gmx.de> writes:
>According to the benchmarks, the library is as fast as the 
>non-library executable

Even on 386?  It used to cost one extra register, which was pretty
expensive on the 386.

- anton
-- 
M. Anton Ertl  http://www.complang.tuwien.ac.at/anton/home.html
comp.lang.forth FAQs: http://www.complang.tuwien.ac.at/forth/faq/toc.html
     New standard: http://www.forth200x.org/forth200x.html
   EuroForth 2015: http://www.rigwit.co.uk/EuroForth2015/

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

#44883

Anton Ertl wrote:

> Bernd Paysan <bernd.paysan@gmx.de> writes:
>>According to the benchmarks, the library is as fast as the
>>non-library executable
> 
> Even on 386?  It used to cost one extra register, which was pretty
> expensive on the 386.

Looks like we have only a very generic 2-register setting for GCC 5.x, so 
that's why.  Adding a third register (treating GCC 5.x like GCC >4.8) gives 
a considerable improvement, and that's then only available to the non-lib 
version.

So maybe there should be a switch for "build both" on THE register-starved 
architecture, while for the others we can just build the library version.

-- 
Bernd Paysan
"If you want it done right, you have to do it yourself"
net2o ID: kQusJzA;7*?t=uy@X}1GWr!+0qqp_Cn176t4(dQ*
http://bernd-paysan.de/

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

#44903

Bernd Paysan wrote:
> Debian armhf builds will probably follow today (the slower the machine,
> the later the build ;-).

Ok, armhf is out, needed assistance by explicit register declaration to 
improve performance on the non-fast engines, especially in libraries (one 
library didn't even work with --dynamic without that).

-- 
Bernd Paysan
"If you want it done right, you have to do it yourself"
net2o ID: kQusJzA;7*?t=uy@X}1GWr!+0qqp_Cn176t4(dQ*
http://bernd-paysan.de/

[toc] | [prev] | [standalone]

Back to top | Article view | comp.lang.forth

csiph-web