nicholaA@moravian.EDU (03/04/89)
Here's some information I thought everyone might be interested in knowing,
but didn't feel it warranted being in the documentation to ShrinkIt -- some
timing information I did while I was rewriting the compress/uncompress
routines in ShrinkIt.
More than one person has accused ShrinkIt of being slower than the current
generation of "whole-disk" packers, even though it almost always compresses
things smaller... After reading this table, you'll be able to tell that
just isn't so.
-------------------------------------------------------------------------------
Shrink Bard's Tale II, 1567 Blocks used, Skewed 2:1
ShrinkIt 0.95 - 5:08, 1470 Blocks
ShrinkIt 1.0 - 4:55, 1470 Blocks
DigiPack GS - 4:57, 1520 Blocks
Unshrink Bard's Tale II, skewed 2:1
ShrinkIt 0.95 - 2:34
ShrinkIt 1.0 - 2:30
DigiPack GS - 3:38
Shrink AppleLink PE System Disk, 1574 Blocks used, Skewed 4:1
ShrinkIt 0.95 - 4:35, 916 Blocks
ShrinkIt 1.0 - 4:28, 916 Blocks
DigiPack GS - 4:26, 1022 Blocks
Unshrink AppleLink PE System Disk, Skewed 2:1
ShrinkIt 0.95 - 3:14
ShrinkIt 1.0 - 3:00
DigiPack GS - 3:10
Shrink IIGS Source Code Sampler Disk #1, 1537 Blocks used, Skewed 4:1
ShrinkIt 0.95 - 3:48, 612 Blocks
ShrinkIt 1.0 - 3:48, 612 Blocks
DigiPack GS - 3:51, 802 Blocks
Unshrink IIGS Source Code Sampler Disk #1, Skewed 2:1
ShrinkIt 0.95 - 2:51
ShrinkIt 1.0 - 2:37
DigiPack GS - 2:56
All of these figures were timed taken from my Apple 3.5" drive, written
to my CMS SD60 hard drive. Each time the readings were taken, my IIgs had
at least 800k free for DigiPack to read the entire 3.5" disk into memory,
process it, then spit it out. (I'm not sure if the code is actual 16-bit code)
The disks processes represent a very good compressable disk (the source
code sampler), a very bad compressable disk (bard's tale II, because most
of the data on the disk is already compressed via Huffman compression), and
a mediocre compressable disk, the ALink system disk. At the time, I wasn't
trying to do this with scientific precision, I just needed some benchmarks
to shoot for in improving the algorithms.
andy
-------------
Andy Nicholas CsNET: shrinkit@moravian.edu
Box 435, Moravian College InterNET: shrinkit%moravian.edu@relay.cs.net
Bethlehem, PA 18018 liberty!batman!shrinkit@sun.com
uucp: rutgers!lafcol!lehi3b15!mc70!shrinkit
AppleLink PE: ShrinkIt rutgers!liberty!batman!shrinkit