Ability to run more than one extraction at once.

[ci-macpaw]

Originally reported on Google Code with ID 176

joe.roback:

Sometimes I am extracting multiple archives from either multiple sources,
Gigabit Network Storage devices or RAID devices which can easily sustain
the IO throughput necessary to have more than one extraction. It would be
nice to have 2 things.

1) Ability to specify a default number of "active" extractions.
2) Ability to click on a queued extraction to "force" start it. Kind of
like how there is a `X` icon on the right to cancel an extraction, if there
also was some kind of play button to start an extraction.

[ci-macpaw]

joe.roback:

you know, if implemented, user can receive a warning if they enable more than 1 or 2
extractions at once, stating they should know what they are doing and that it may
hurt extraction performance.

This kind of power-option you will never see in Mac's built-in extractor. I am
probably only one of very few extracting more than one archive at a time from
separate sources, so I understand the low/maybe priority. I'll have to stick to
command extraction for now.

[ci-macpaw]

paracelsus:

Generally if you run extractions directly on local disks, the problem with running multiple
ones is that seeking 
overhead will kill your throughput. Of course, if the extraction process is particularly
slow, you can still afford 
that, but in the end it all turns into a big mess trying to figure out when an archive
can safely be started in 
parallel.

Like said, It's something I've kept in mind but never figured out how to do right.
I think I'll leave the issue open 
for future versions, though, in case a good behaviour can be worked out at some point.

[ci-macpaw]

joe.roback:

I guess. Its certainly your decision as the developer. And I do understand it can be
an easily abused option. But almost no extraction is multithreaded, leaving only
multiple extractions as a way to maximize hardware / extraction times.

For instance, the unrar code (including RARLabs own C++ code even compiled with Intel
Compiler) at maximum extracts at 15-20 MB/s on an 8-core Xeon with 8GB of memory.
Even laptop disks can sustain 80MB/s on a sequential write. I can unrar 4 archives
at
once before I start seeing a degradation in performance. My most common setup looks
like this

Linux Box (8-core xeon with single 7k SATA disk) over a gigabit network to a Macbook
Pro 17" with 7.2k harddisk. I can sustain 50MB/s easy with 3 archives using the
command unrar tool.

Again, I not trying to force or heavily suggest you to do anything you don't want to
do. I've always liked the idea of a replacement for the built-in Mac OS X extractor
tool with more power-user-like options, and I think theunarchiver is really it, so
I
just want to provide helpful, useful feedback. :-)

[ci-macpaw]

paracelsus:

The code has always been designed with this in mind, but in the end I decided that this
feature would add 
unnecessary complexity for very little actual gain. The cases where this is useful
are very few, and only give a 
very small increase in speed. Mis-using the feature loses a whole lot of speed, and
just results in confusion for 
users.