Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

New high speed compression - lz4

94 views
Skip to first unread message

Dagen Brock

unread,
Mar 18, 2013, 12:08:48 PM3/18/13
to
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

Bill Buckels

unread,
Mar 18, 2013, 3:57:24 PM3/18/13
to

"Dagen Brock" <Dagen...@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



Egan Ford

unread,
Mar 18, 2013, 6:20:17 PM3/18/13
to
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).

Dagen Brock

unread,
Mar 28, 2013, 6:09:16 PM3/28/13
to
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! :)

Dagen Brock

unread,
Mar 28, 2013, 6:15:21 PM3/28/13
to
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.

Egan Ford

unread,
Mar 28, 2013, 6:37:08 PM3/28/13
to
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.
0 new messages