SABnzbd Forums

Hi, I just wanted to start off by saying sabnzbd is a great program and I love how frequently it's updated. I have come to appreciate how versatile and useful this program is.

I would like to suggest ICE ECC support. Can it be added to Sabnzbd?

What is ICE ECC? ICE ECC is an alternative to PAR2 files. It's faster creating AND recovering and supports multiple cores. I am unsure whether or not it supports 64bit.

Indirect reason: ICE ECC is able to create recovery files much faster than PAR2 programs can. I realize that has absolutely no benefit for sabnzbd because sabnzbd simply utilizes recovery files, it does not create them; however, it's worth mentioning.

Direct reason: ICE ECC is able to recover files much faster than PAR2 files. This would benefit everyone because it takes less time to recover files if the files include ICE ECC recovery files.

Bonus reason: Whenever Sabnzbd recovers files with Par2 files, it lags my computer senseless. I have tried limiting the recovery to a single core, I've tried reducing the priority, I've tried increasing the priority of *other* programs to compensate. Nothing seems to work. Whenever sabnzbd begins recovering files from a large multi gig download, my computer becomes unusable for about 2 minutes. I'm barely able to open the task manager, I'm barely able to move the mouse.

My computer is pretty good. I have 8 gigs of DDR3 1600 ram, a Core i7 870 quad core, a Radeon 5850 graphics card, a 650w PSU, a pretty clean system... my specs fit the bill nicely. This problem is not related to Sabnzbd, this is directly related to par2 recovery files. This problem occurs in other programs such as ParBuddy. This problem occured on my last computer as well. In short, this problem is directly linked to Par2 files and can be solved by switching to ICE ECC. Apparently, ICE ECC is a lot friendlier and allows you to continue using your computer during a recovery.

I know what you're going to say: "Nobody uses ICE ECC, so why should we add it to Sabnzbd? We would be wasting our time."

My response: Not a lot of people know about ICE ECC, so adding it to sabnzbd would cause people to take notice. Adding support would cause other program devs to add support and in a year or 2, everybody might be using ICE ECC. Not to mention, PAR2 files can also be included as a fallback in case someone doesn't have ICE ECC support in their usenet downloader.

All in all, I think ICE ECC should be given a fair consideration. It's a pretty good program.

Super bonus reason: Par2 successor may never come to fruition! It has been years since an update has been released or even talked about. Perhaps Par2 is dead? Food for thought! Maybe it's time to jump on the ICE ECC bandwagon?

Thanks for reading,
-whatevermynameis

NZBsHereGrabbinNZBs wrote: I know what you're going to say: "Nobody uses ICE ECC, so why should we add it to Sabnzbd? We would be wasting our time."

That's exactly my answer.
More importantly: the tool is Windows-only and closed-source.
Two very good reasons for not even wanting to support it.

Is there an alternative to ICE ECC that is superior to Par2 files and is also open source and cross platform?

None that I know of.
There are a few improved-performance variants of it. We use one for Windows and for OSX.
There's also an superior Japanese implementation, but its sources haven't been published and it's Windows-only.
Because SABnzbd is multi-platform and open source.
Supporting ice-ecc would go against our own goals, so we won't do anything to promote it.
If one day it would become popular with posters, we might reconsider our position.
We definitely will not start the ball rolling.

Yeah, I actually just contacted the guy and asked him if he ever planned on making ICE ECC open source / cross platform.

His answer was directly "no" and then he told me that he has plans of turning it into a pay-for program.

Maybe he'll make some money off of it, but my guess is... not very many people need a standalone repair/recovery program for personal use; therefore, I foresee extremely bleak financial outlook for the guy.

I don't claim to be an expert in any of this, but I would say it's better to integrate something like that into other applications--like Sabnzbd, or commercial applications... you know, where it's more important.

Just my 2cents.

I'm gonna make a new thread.

NZBsHereGrabbinNZBs wrote:Yeah, I actually just contacted the guy and asked him if he ever planned on making ICE ECC open source / cross platform.

His answer was directly "no" and then he told me that he has plans of turning it into a pay-for program.

Then I hope he looks forward to no one using his application.

If it isn't freely and widely available, no one will use ICE ECC. If no one is out in the wild using ICE ECC on usenet, there is no reason for us to implement it.

I didn't test ICE ECC, but just looking at the Test A from its ohw homepage (http://www.ice-graphics.com/ICEECC/IndexE.html) I don't see why it should be much better.
So QuickPar takes 03:14 and ICE ECC 02:37. 37sec less. But QuickPAr is far from the fastest par2 tool.
For example, phpar2 beats QuickPar. Example from the phpar2 homepage is 2:10 with QuickPar but only 44sec! for phpar2.

I wouldn't be suprised to see phpar2 or another optimized par2 (TBB?) beat ICE ECC.

edit:
OK; out of curiosity I made a small test. 413MB in 11 rar files on my notebook (Core 2 Duo 2GHz)

.\ICEECC.exe c /r10 /b487 .\test\*.rar
vs
.\phpar2.exe c -r10 -b487 test.par2 .\test\*.rar

ICE ECC took 24.43 sec, phpar2 only 18.69sec.

ICE ECC file size is 41,0MB, par2 file size is 42,3MB

Verify of complete files took 5.53sec with ICE ECC, 5.77sec with phpar2

recovery after I deleted a 14MB file took 12.51sec with ICE ECC, 10.92sec with phpar2

So in my opinion, ICE ECC sucks and there is no need to use it for anything!

Interesting. Thanks for taking the time to perform that test.

It says there is a 32bit version of phpar2, but is there a 64bit version? If not, would that make it difficult to implement into Sabnzbd?

If phpar accepts the same parameters and produces the same console output
it would be a drop-in replacement.
BTW: don't mistake being 64bit for being faster by definition.

It doesn't really matter to me whether 64bit is faster or better than 32bit.

There is no 64bit version because it contains hand written 32bit assembler code that would have to be rewritten for 64bit.

shypike wrote:If phpar accepts the same parameters and produces the same console output
it would be a drop-in replacement.

Well, in that case could we look into this? Because like the OP said, when processing a large download my quad core i7 also throttles to completel unusable for several minutes while par2 runs.