Atari 8-bit Software Preservation Initiative

[walter_J64bit]

I'll just keep my PC running.

[Farb]

It seems I had a network misconfiguration that was not allowing peers to connect. I think it's fixed now so we'll see if things start working for folks.

[_The Doctor__]

okay... what's the IP I perma banned people who sat a zero for more that 48 hours

 

still got nothing coming

 

I will manually add the peer ip and port I leach at 330MiB down and feed at 25 MiB up

Edited May 19, 2017 by _The Doctor__

[AtariGeezer]

Looks okay to me, there are files that are bigger than 3 MB and conversational German at 271 MB which leaves 184 MB for all the rest which are less than 1 MB and most of those are less than 100k...

[_The Doctor__]

cool, the numbers are good but the torrent is still stalled correct?

[Farb]

Ok, I was trying to be clever and set up a torrent server so there was always a seed. That seems to have failed miserably, so here's a torrent file created the same way I've done the previous ones...

Atari 8bit Preserved Software 2017-04-30.torrent.zip

[AtariGeezer]

Yep, no peers are currently downloading ...

 

New torrent downloaded 100% and now seeding...

Edited May 19, 2017 by AtariGeezer

[Clay Cowgill]

I'm at 100% now, I cranked up my upstream limit and will leave it seeding for overnight and some time tomorrow. Thanks for all the work on thie Farb!

[_The Doctor__]

it is rolling along but rather slowly.... only 300 KiB/s I sure hope everyone delete all parts and files of the failed torrents as the name and folders may resend whatever was part of the old torrent

 

2 seeds found

 

I am currently uploading 3x more than I am getting...

Edited May 19, 2017 by _The Doctor__

[Clay Cowgill]

I turned my upstream throttling off entirely, but it's not going above ~500kB/s. Probably Comcast choking off torrent traffic or something...

[Kyle22]

480K here, also Comcast.

[Clay Cowgill]

480K here, also Comcast.

Very suspicious...

[_The Doctor__]

I was able to feed someone at 1MiB I could feed 25 more at that rate....

[xdslx]

two peers except me , both are stuck at %70.7 . one is from us other uk.

[xdslx]

its so small , can somebody upload it to mega instead of torrent ?

[Mclaneinc]

The new torrent was full speed for me and now seeding...

 

Thanks Farb etc...

[+remowilliams]

its so small , can somebody upload it to mega instead of torrent ?

 

EDIT: link updated

 

https://mega.nz/#F!U0llGDpQ!F3FIe1kW4_zbYBWyHElfvw

 

[sargie]

Not having much luck with the torrent - Stuck at 70 odd percent now for 12 hours

[+remowilliams]

Not having much luck with the torrent - Stuck at 70 odd percent now for 12 hours

 

Use the new torrent file from this post http://atariage.com/forums/topic/234684-atari-8-bit-software-preservation-initiative/page-26?do=findComment&comment=3764564

 

It is active and seeded.

[Triads]

Not having much luck with the torrent - Stuck at 70 odd percent now for 12 hours

Im seeding the previous torrent now

[sargie]

Thanks! Is working now - almost done!

 

Will leave it seeding until the next one!

[xdslx]

 

https://mega.nz/#F!dpUjQJYJ!dJsO7bbRK63dtNUII63zUg

thank you very much , theperhim gary kach

[ijor]

Here is a clean working dump of Zeppelin. As the name of the file implies, this title works only on an 810 drive. It will randomly fail on a 1050 and probably on everything else as well. So if using Altirra, make sure the drive is configured as 810. Will elaborate later ...

 

I wanted to make a detailed article about this because it is quite interesting. But never found the time. So I am posting here the main concept "as headlines" without much of a redaction. Technically oriented people will probably understand, I hope

 

- Zeppelin uses the so called skew align as one of its protections.

- Skew align refers to a specific angular alignment of the sectors across different tracks.

- Most titles with this protection were formatted aligned with the index hole.

- So that the same physical sector number is located at almost the same angular position in all the tracks.

- The typical protection check performs something like reading physical sector one of a track, then sector one of an adjacent track, and measures the total time taken. It should be one full revolution.

- Some protections measure the actual RPM before, others assume a nominal value.

- The specific drive and firmware timings, such as seek time and settle delay, are not relevant. It always takes one full revolution because the revolution time is much bigger.

- The total time since the drive starts reading the first sector until being ready to read the other one in the adjacent track is in the order of 150 ms; while a full revolution takes 208 ms.

- So the drive always needs to wait one full revolution disregarding if it was ready 20 ms earlier or later.

 

- Zeppelin disks are is not aligned to the index hole. The skew align was achieved likely using timing.

- That means that sectors with the same physical number in different tracks are not at the same angular position.

- It also makes the formatted alignment not that accurate because it is more influenced by the drive RPM.

- The angular distance between specific sectors on adjacent tracks is not constant. This is probably on purpose to make reproduction of the alignment more difficult.

- But note that yet the alignment across multiple copies is, of course, (more or less) constant for a given pair of tracks.

- Protection is checked similarly as the typical case, measuring the elapsed time since reading physical sector one of a random track, until reading also physical sector one on an adjacent track.

- But because of the not constant alignment, it doesn’t always take one revolution. The software has a table for the expected time for each track.

- And because now in some tracks it takes less time than one revolution, the internal drive timings become relevant and significant.

- The protection assumes the timing matches a standard 810 drive. Just about any other drive would randomly fail the protection.

[Farb]

This is fantastic information - thanks, Ijor. I've updated the information for the Zeppelin dump on the preservation website with this info.

[DjayBee]

 

- But because of the not constant alignment, it doesn’t always take one revolution. The software has a table for the expected time for each track.

 

Slight correction: It's actually two tables to diffenrentiate between PAL and NTSC.