RobboBases

[Prima]

Addendum:

By the way Kingliveson, I think there's a mix-up with the new and old RobboTripleBases. In your directory, there are
1. RobboTripleBase/
2. RobboTripleBase -old/

I downloaded RobboTripleBase/ [the new ones] and setup everything. However, IvanHoe 9.46A kept hanging upon start-up and just would not play. It turns out, according to my log file, that the supposedly new RobboTripleBases were actually the old ones.

Which brings me to yet another concern:

The 1st RobboTripleBases = 625MB uncompressed of 1-5 & Z pieces;
2 = 2 Bytes
3 = 25.9 KB
4 = 3.46 MB
5 = 441 MB
Z = 180 MB

The 2nd [or Newer] RobboTripleBases = 800MB uncompressed of 1-5 & Z pieces;
2 = 2 Bytes
3 = 26.3 KB
4 = 3.39 MB
5 = 568 MB
Z = 228 MB

With your recent announcement of new RobboTripleBase [also 800 MB uncompressed of 1-5 & Z pieces], which turns out to be an old RobboTripleBases [according to the registration in my log files];
2 = 2 Bytes
3 = 26.3 KB
4 = 3.39 MB
5 = 568 MB
Z = 228 MB

I knew there was a third, different type of RobboTripleBases, although I wasn't sure "which" was "which". From what I've noticed, there appears to be a 2-different types of the OLD RobboTripleBases (one of them is 625 MB and the other is 800 MB uncompressed) and 1 New RobboTripleBase (also 800 MB).

So, which of the OLD RobboTripleBases is to be used or better? Was there a fault with one of the OLDER RobboTripleBases?

[kingliveson]

There is the old and new; the old RobboTripleBases use TRIPLE prefix, while the new ones use TRIPLE0, TRIPLE2, and TRIPLE3.

If the engine is hanging, it has nothing to do with passing it older TripleBases path. A little secret between you and I -- the older TripleBase works with the latest IvanHoe 999946a. Although there could be unforeseen consequences when mixing the old with new.

Be sure to check the md5sum with the archive provided.

[Prima]

There is the old and new; the old RobboTripleBases use TRIPLE prefix, while the new ones use TRIPLE0, TRIPLE2, and TRIPLE3.

If the engine is hanging, it has nothing to do with passing it older TripleBases path. A little secret between you and I -- the older TripleBase works with the latest IvanHoe 999946a. Although there could be unforeseen consequences when mixing the old with new.

Be sure to check the md5sum with the archive provided.

The TripleBases md5 sum are fine and I keep each TripleBases in separate folders:

1. RobboTripleBase-OLD-625MB => For the Ist publicly released TripleBases, totaling 625 MB in size when uncompressed.
Therefore, the path for this would be: C:\RobboTripleBase-OLD-625MB

2. RobboTripleBase-NEW-800MB => For the 2nd publicly released TripleBase [either in 2010 or early 2011], totaling 800 MB in size when uncompressed.
Therefore, the path for this would be: C:\RobboTripleBase-NEW-800MB

3. RobboTripleBase-OLD-800MB => Your recent link to "new TripleBase" in the 1st page of this thread, which turns out to be old TripleBases. Also 800MB. Hence why I named it "OLD"
Therefore, the path for this would be: C:\RobboTripleBase-OLD-800MB

I was surprised to learn IvanHoe 9.46a also uses old TripleBases. So I tested each of the three TripleBases with IH9.46a. Only #2 [the New TripleBases] turns out to be used by IH9.46a. Attached are my PolyGlot log-files, announcing which bases are actually used or loaded and which are not. See for your self...

[kingliveson]

You didnt check the md5sum, and by the way, you already have the newer RobboTripleBase.

I downloaded the files (newer TripleBases) that had been uploaded, and come to find out, the md5sum after being extracted do not match, so they probably got corrupted during compression.

Code: Select all

< 1852a9b6bbab847de939f727322c14b5  TRIPLE3.Kbbwrbp
---
> 29c8f91612265fc62ca2e1d972ae6c02  TRIPLE3.Kbbwrbp

Since you already have the newer TripleBases, the way you can help at this point is once they are re-uploaded, you could test them. Of course, I'll do a check here too.

Another thing, there should be no bulk-loading when using the older TripleBases -- it loads everything into RAM which is what bulk-loading does with the newer ones.

Edit:
When I said "be sure to check the md5sum with the archive provided," I wasn't quite clear. The md5sum in directory as the 7z files is for archive integrity check . At times if an archive wont decompress, it probably got corrupted during download. Other reasons could be using older version or other software that dont fully support the compression format. Now once the archive is extracted, there is another md5sum file for all content -- this is what was meant.

[kingliveson]

The newer RobboTipleBases have now been successfully re-uploaded. Apparently, they were corrupted when compressed. So, to be safe, integrity check for RobboTotalBase and older RobboTripleBase (RobboTripleBase-old) on the server was done -- they passed!

Note:
RobboTripleBase directory contains the newer format. RobboTripleBase previously had old format, and has been renamed RobboTripleBase-old. Sorry for any inconvenience.

md5sum integrity check is pretty straight forward on non-Windows OS.

Code: Select all

md5sum -c <md5sum_file>

The above is available in Windows with Cygwin environment installed. An alternative approach would be to get md5sum.exe or md5summer.

[Prima]

You didnt check the md5sum, and by the way, you already have the newer RobboTripleBase.

I downloaded the files (newer TripleBases) that had been uploaded, and come to find out, the md5sum after being extracted do not match, so they probably got corrupted during compression.

Code: Select all

< 1852a9b6bbab847de939f727322c14b5  TRIPLE3.Kbbwrbp
---
> 29c8f91612265fc62ca2e1d972ae6c02  TRIPLE3.Kbbwrbp

Since you already have the newer TripleBases, the way you can help at this point is once they are re-uploaded, you could test them. Of course, I'll do a check here too.

Another thing, there should be no bulk-loading when using the older TripleBases -- it loads everything into RAM which is what bulk-loading does with the newer ones.

Edit:
When I said "be sure to check the md5sum with the archive provided," I wasn't quite clear. The md5sum in directory as the 7z files is for archive integrity check . At times if an archive wont decompress, it probably got corrupted during download. Other reasons could be using older version or other software that dont fully support the compression format. Now once the archive is extracted, there is another md5sum file for all content -- this is what was meant.

You're right, I didn't check THIS particular "new RobboTripleBases", announced in page 1 of this thread. Somehow that slipped my mind, although my previous RobboTripleBases (the 625MB & 800MB) were checked and turned out ok. Thanks anyway.

[Prima]

The newer RobboTipleBases have now been successfully re-uploaded. Apparently, they were corrupted when compressed. So, to be safe, integrity check for RobboTotalBase and older RobboTripleBase (RobboTripleBase-old) on the server was done -- they passed!

Note:
RobboTripleBase directory contains the newer format. RobboTripleBase previously had old format, and has been renamed RobboTripleBase-old. Sorry for any inconvenience.

md5sum integrity check is pretty straight forward on non-Windows OS.

Code: Select all

md5sum -c <md5sum_file>

The above is available in Windows with Cygwin environment installed. An alternative approach would be to get md5sum.exe or md5summer.

In the http://chess.cygnitec.com/ RobboTripleBase download directory, the directory with RobboTripleBase/ actually contains the new TripleBases AND the RobboTripleBase-old/ contains old TripleBases, respectively? Thanks for the correction.

So, is this this new TripleBase (1-5 & Z pieces) from the http://chess.cygnitec.com/RobboBases/ directory the same "new TripleBase" (1-5 & Z) released early 2011 (or late 2010) from the site of IvanHoe's developers?

[Prima]

By the way, the NEW RobboTripleBase for 5.7z.001 - 5.7z.004 gives archive error when extraction is attempted. However, the Z.7z.001- .002 extracts fine. And thanks for all the corrections.

[kingliveson]

By the way, the NEW RobboTripleBase for 5.7z.001 - 5.7z.004 gives archive error when extraction is attempted. However, the Z.7z.001- .002 extracts fine. And thanks for all the corrections.

You already had the newest and lastest RobboTripleBase. Can you please test the md5sum of those archive? thanks. Also, which version of 7-zip are you using?

[Prima]

By the way, the NEW RobboTripleBase for 5.7z.001 - 5.7z.004 gives archive error when extraction is attempted. However, the Z.7z.001- .002 extracts fine. And thanks for all the corrections.

You already had the newest and lastest RobboTripleBase. Can you please test the md5sum of those archive? thanks. Also, which version of 7-zip are you using?

Using md5summer.exe, all TripleBases (excluding 5) checked-out OK. I could not extract the 5-TripleBases (5.7z.001 - 5.7z.004) from its archive. However, the new TripleBases for 5 pieces downloaded earlier this year from the IvanHoe main site, checked-out OK. It's attached.

The version of 7-zip used is 9.20. It had no problem extracting TripleBase Z (.001 & .002).