World: r4wp

Sorry, just tried to make a point :-)

You replied the same second Kaj Luke!

No, that's another way to prove the point. You can just edit the 
time of your message. Do you want me to answer before the previous 
post? :-)

Maybe you can get the columns the same height again? No I just wrote 
it to point this out in case someone doesn't notice.

No, I can't edit previous messages. Maybe the troll can

I was doubting my previous postings ;)

We have no proof the second fake Carl is the same as the first fake 
Carl :)

To add to the idiocy, there's now a third, real Carl. He's stating 
he agrees with the fake one

Guess it's the real one now as the fake comments have gone

now to guess which of the other 184 comments were faked

Hmmmm.....We now know a way to encourage the real Carl to post :)

Thanks to the fake Carl .. please stand up and take a bow :)

Fake a bow?

just bend over ...

Carl has removed a few more tags to stop spoofing ...

in yellow

soo.. is this reply 10-Sep-2012 14:39:12 the real Carl or just more 
delicate fake Carl? The message still seem too simple about the future 
for the real Carl.

That's the real Carl - only the real Carl can remove comments, and 
comments were removed.

I think that the real hacker, has some surprise under the hood :-)

Note that my Etsy API script does patch HTTP (for purposes of retrieving 
API error messages).  You can view those changes here: http://reb4.me/r/etsy-http-hack.r

And with thanks to Nick for instigating, supporting and helping test!

Thanks Chris and Nick. I'll be using that code!

James, let me know if you encounter any problems. It's always a pain 
getting to grips with an OAuth app to start.

Thanks Chris I will. My wife has been itching to sell some stuff 
on etsy. Thanks for all you do Chris.

No one here can test my new %opengl-glu-glut-h.r on X11?

Cool kaj.

Thanks Kaj.

Hey, you are in the video too this time!

Kaj: how do Fibonaci.ruby and Fibonaci.r compare in speed on your 
demo machine?

Ah, found a verbose: 1 left in natives.reds thanks to your presentation...and 
the code is missing a macro....funny way to debug/improve my own 
code. ;-)

:-)

I published the complete benchmarks here early this year, I think. 
However, my Ruby results are considered unfair, because Syllable 
is using Ruby 1.8, while Ruby 1.9 is much faster. For the record, 
as far as I remember, REBOL is roughly twice as fast on such things 
as Ruby 1.8. Not sure if that applied to Fibonacci, but I remember 
it about Mandelbrot

Android binary is twice the size of Linux one because ARMv5 architecture 
is bad at dealing with 32-bit literal values, so it takes much more 
space than for IA-32.

Yeah, that's more or less what I explained

Even with my experimental literal pools allocator, it decreases the 
final size by 10KB only.

I like the part: "Where will Red be deployed when ready?...Everywhere!" 
;-)

Thank you for your presentation Kaj! I've enjoyed it, but I guess 
I'm too involved to be objective. ;-)

My pleasure

(I think the microbenchmark results were posted back in the REBOL3 
world.)

Oh, right

BTW, wrt to ARM big size binaries, it's also caused by ARM using 
32-bit words for every instruction, while IA-32 has still a lot of 
8 or 16-bit ones. For example, there's a lot of PUSH 0 instructions 
emitted for the datatype registration block (the unimplemented action 
pointers), that's 16bit on IA-32 and 32-bit on ARM.

However, ARM has also a lot of nice features and advantages over 
Intel CPUs.

I was reading the instruction set in 1987 :-)

[[Just dug out Kaj's numbers for the Mandelbrot benchmark posted 
in the REBOL3 world in Feb 2012:

Red/System: 15
REBOL2: 440
Ruby(1.8): 480


(All relative slowdown compared to the fastest implementation in 
Kaj's tests.)]]

You mean in the video? That's a faster machine

That was pre-faster-floats branch merge for Red/System, after that, 
Red/System score went down to 7 IIRC. ;-)

No, that's just the numbers you posted back in 2012-02. Slowdown 
compared to the performance of your C results back from 2012-02.

Oh, that way

So back then, the Red/System binary took 15x the time of the C binary 
to run to completion. And yes, that was before the float optimisations 
:)

(Not run to completion, but finish the mandelbrot computation. AFAIR, 
you timed from within the programs themselves, not via an external 
tool.)