[Lazarus] Theoretical question about future of Lazarus

[Marc Santhoff]

On Di, 2015-10-13 at 18:07 +0300, Juha Manninen wrote:
> On Tue, Oct 13, 2015 at 5:02 PM, Marc Santhoff <M.Santhoff at web.de> wrote:
> > You only want a figure, the big picture?
> 
> Yes, basically.
> 
> > Making target "bigide" is not completely comparable, it depends on what
> > additional components are installed.
> 
> No, "bigide" has a fixed set of components.

Hmm, but it does build the components I have installed as an extra to
"naked" Lazarus. So it does more work than a vanilla installation?

> > And if the disk is a rotating one or an SSD. And the moons phase, me thinks ... ;)
> 
> SSD is a valid way to speed up a computer.
> Moon phase clearly was the cause for some nasty bugs which I could not
> solve, but "make bigide" timing can be reproduced quite accurately
> with the same compiler settings.

Compiler settings are unchanged from the installation.

> >  > dmesg|grep -i cpu
> > CPU: AMD Phenom(tm) II X4 975 Processor (3612.05-MHz K8-class CPU)
> > FreeBSD/SMP: Multiprocessor System Detected: 4 CPUs
> 
> 3,6 GHz. Is it the Black Edition (Deneb)?

I had no idea, but a search for the invoice said: yes, it is.

> > OK,not really fast and a bit dated, but enough for me. ;)
> > Since "gmake -j4" failed I'm posting only one test:
> >
> >> gmake clean
> >> time gmake bigide
> > 135.043u 10.460s 2:39.86 91.0%  2541+1181k 1059+11804io 74pf+0w
> >> gmake clean
> >> time gmake bigide
> > 135.264u 10.245s 2:37.18 92.5%  2542+1182k 0+11803io 0pf+0w
> >
> > Files are stored on a rotating disc.
> > Two and a half minutes is not enough for a coffee break.
> 
> Not bad. Still a lot faster than Beebox.

Need it for CAD, simulations and the like. But the Beebox looks good for
smaller servers and media boxes. And it's beautiful to some extent.

But the first run was Lazarus 1.4.2. Trying again using version 1.4.4
and including gmake clean:

> time gmake clean bigide
135.607u 11.301s 2:39.85 91.8%  2513+1182k 56+17961io 0pf+0w

> gmake clean
> time gmake bigide
135.717u 10.049s 2:37.20 92.7%  2537+1180k 37+11886io 0pf+0w

No big difference. For fun another run on a SSD:

> time gmake clean bigide
135.293u 9.968s 2:26.14 99.3%   2526+1184k 23+5787io 37pf+0w

Helps a bit.

I'd like to see the times for RPi and RPi2 ... but on those devices the
storage media playes a role even bigger, some sd-cards are awfully slow.

> Yet, the processing power / electric power ratio is quite amazing in
> Celeron N3000. Its TDP is 4W.

That's the only thing not so nice, but using power management functions
it's ok, having enough power when needed.

-- 
Marc Santhoff <M.Santhoff at web.de>