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


Groups > comp.sys.apple2.programmer > #680 > unrolled thread

New high speed compression - lz4

Started byDagen Brock <DagenBrock@gmail.com>
First post2013-03-18 09:08 -0700
Last post2013-03-28 16:37 -0600
Articles 6 — 3 participants

Back to article view | Back to comp.sys.apple2.programmer


Contents

  New high speed compression - lz4 Dagen Brock <DagenBrock@gmail.com> - 2013-03-18 09:08 -0700
    Re: New high speed compression - lz4 "Bill Buckels" <bbuckels@mts.net> - 2013-03-18 14:57 -0500
    Re: New high speed compression - lz4 Egan Ford <datajerk@gmail.com> - 2013-03-18 16:20 -0600
      Re: New high speed compression - lz4 Dagen Brock <DagenBrock@gmail.com> - 2013-03-28 15:09 -0700
        Re: New high speed compression - lz4 Dagen Brock <DagenBrock@gmail.com> - 2013-03-28 15:15 -0700
        Re: New high speed compression - lz4 Egan Ford <datajerk@gmail.com> - 2013-03-28 16:37 -0600

#680 — New high speed compression - lz4

FromDagen Brock <DagenBrock@gmail.com>
Date2013-03-18 09:08 -0700
SubjectNew high speed compression - lz4
Message-ID<144e4322-945e-4ab7-a8f4-a05b449a57d8@googlegroups.com>
I was reading about a new, very fast compression algorithm and wondering if it maintained its speed for older CPU architectures like 65xx.  

Fortunately, it's already been ported to 65xx assembly.  I may have to try it out. 

https://code.google.com/p/lz4/

The 65xx decompressor is 166 bytes!  You can grab it here: http://pferrie.host22.com/misc/appleii.htm

[toc] | [next] | [standalone]


#681

From"Bill Buckels" <bbuckels@mts.net>
Date2013-03-18 14:57 -0500
Message-ID<ki7rjo$fpj$1@speranza.aioe.org>
In reply to#680
"Dagen Brock" <DagenBrock@gmail.com> wrote:

 > https://code.google.com/p/lz4/

Thanks Dagen.

That acme assembler looks like it might be fun. It would be very cool if the 
converter could accept 6502 object code directly and convert it to acme 
source. I need to watch these folks. Some really dedicated people, looks 
like.

I put myself on the acme mailing list to learn more.

Bill
 

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


#682

FromEgan Ford <datajerk@gmail.com>
Date2013-03-18 16:20 -0600
Message-ID<ki83v1$bg3$1@news.xmission.com>
In reply to#680
On 3/18/13 10:08 AM, Dagen Brock wrote:
> I was reading about a new, very fast compression algorithm and wondering if it maintained its speed for older CPU architectures like 65xx.
>
> Fortunately, it's already been ported to 65xx assembly.  I may have to try it out.
>
> https://code.google.com/p/lz4/
>
> The 65xx decompressor is 166 bytes!  You can grab it here: http://pferrie.host22.com/misc/appleii.htm
>

Hi Dagen, Thanks, I'll have to check it out.

BTW, I have been using this: 
http://atariarea.krap.pl/x-asm/inflate.html.  Supports higher 
compression, e.g. gzip.  Probably slower (and the code is larger). 
zopfli should be supported as well (on my list to check).

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


#691

FromDagen Brock <DagenBrock@gmail.com>
Date2013-03-28 15:09 -0700
Message-ID<165fabd1-1c59-4a27-92f7-d4e7d7d2c2c0@googlegroups.com>
In reply to#682
That's a really good point.  Indeed lz4 is faster, but compresses less, while gzip is slower with better compression ratios. 

It depends on your use case, and while the inflate lib is bigger, I think the compression should more than make up for that if you are compressing >~2k.  

The appeal of the lz4 decompressor was the tiny size and speed.  But if I hit size limits somewhere it's good to now I could switch to this inflate library. 

Isn't it great we have so many options!  :) 

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


#692

FromDagen Brock <DagenBrock@gmail.com>
Date2013-03-28 15:15 -0700
Message-ID<7d61cb23-a248-4b19-9120-6e4851dfc5c8@googlegroups.com>
In reply to#691
Woops.  I was making that statement about speed base on benchmarks from *modern platforms*.  I was only extrapolating that it would be the case for 6502 implementations too.  I could be wrong about that, so take my statements about speed with a grain of salt.  But I do believe it to be the case. 

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


#693

FromEgan Ford <datajerk@gmail.com>
Date2013-03-28 16:37 -0600
Message-ID<kj2gmk$j7p$1@news.xmission.com>
In reply to#691
On 3/28/13 4:09 PM, Dagen Brock wrote:
> Isn't it great we have so many options!  :)

It is.  And these are "new" options--makes the Apple II more exciting 
knowing that dedicated developers are not done with it.

[toc] | [prev] | [standalone]


Back to top | Article view | comp.sys.apple2.programmer


csiph-web