Strangely high serial number for Fidelity SC9

This forum is for general discussions and questions, including Collectors Corner and anything to do with Computer chess.

Moderators: Harvey Williamson, Steve B, Watchman

Forum rules
This textbox is used to restore diagrams posted with the fen tag before the upgrade.
Larry
Senior Member
Posts: 2269
Joined: Wed Aug 01, 2007 8:42 am
Location: Gosford, NSW Australia

Post by Larry »

Steve B wrote: the replies to 1.h4 are NOT book moves so i recalled that incorrectly


SC9B=1.h4 b6 Regards
Steve
As a piece of trivia, the CB16 module does have a book reply
for 1.h4. It is 1...e5. This might be what you were thinking of.
There is another test for SC9B. Set up any position with KBN/k,
and the SC9A claims a drawn game by insufficient material after
each move. The SC9B does not claim a draw, but of course cannot
demonstrate the win.
L
Mike Watters
Member
Posts: 429
Joined: Fri Sep 26, 2008 12:31 pm
Location: Milton Keynes
Contact:

Post by Mike Watters »

spacious_mind wrote:Since you bought yours from a retailer as an upgraded version B. The only missing test is for other 4xxx serial number owners to do the same test to see if their computer is the same as yours.

It seems that you have found a simple test that anyone can do with their SC9's to confirm version.

Best regards
Hi Nick

See top of Page 3 of this thread.
Also there are a few words about the 9B in Computer Chess Digest 1984 Pages 114/115 -
http://www.chesscomputeruk.com/Computer ... t_1984.pdf

All the best
Mike
User avatar
spacious_mind
Senior Member
Posts: 4000
Joined: Wed Aug 01, 2007 10:20 pm
Location: Alabama
Contact:

Post by spacious_mind »

Mike Watters wrote:
spacious_mind wrote:Since you bought yours from a retailer as an upgraded version B. The only missing test is for other 4xxx serial number owners to do the same test to see if their computer is the same as yours.

It seems that you have found a simple test that anyone can do with their SC9's to confirm version.

Best regards
Hi Nick

See top of Page 3 of this thread.
Also there are a few words about the 9B in Computer Chess Digest 1984 Pages 114/115 -
http://www.chesscomputeruk.com/Computer ... t_1984.pdf

All the best
Mike
Thanks Mike, that's it, you can't beat a simple test to identify a version. Sorry I didn't see it earlier.

Regards
Nick
Nick
User avatar
Steve B
Site Admin
Posts: 10140
Joined: Sun Jul 29, 2007 10:02 am
Location: New York City USofA
Contact:

Post by Steve B »

Steve B wrote: as has been mentioned before here in different threads ....

in my A.. at the 5 second level... it will answer 1.h4 with either 1..e5 or 1..d5
only those two moves
with my "B" it replies to 1.h4 with 1..b6 only
as I said above ..the b6 test has been discussed before several times in the forum
this thread is certainly not the first time

The earliest thread I could find on it dates back to 2009
Larry first mentions the test 1..b6 for the SC9B
all credit goes to Larry

http://hiarcs.net/forums/viewtopic.php?p=40632#40632

8 Years Ago Regards
Steve
User avatar
spacious_mind
Senior Member
Posts: 4000
Joined: Wed Aug 01, 2007 10:20 pm
Location: Alabama
Contact:

Post by spacious_mind »

Steve B wrote:
Steve B wrote: as has been mentioned before here in different threads ....

in my A.. at the 5 second level... it will answer 1.h4 with either 1..e5 or 1..d5
only those two moves
with my "B" it replies to 1.h4 with 1..b6 only
as I said above ..the b6 test has been discussed before several times in the forum
this thread is certainly not the first time

The earliest thread I could find on it dates back to 2009
Larry first mentions the test 1..b6 for the SC9B
all credit goes to Larry

http://hiarcs.net/forums/viewtopic.php?p=40632#40632

8 Years Ago Regards
Steve
OK so it is a well done Larry! :)

Best regards
Nick
User avatar
Tibono2
Full Member
Posts: 696
Joined: Mon Jan 16, 2017 7:55 pm
Location: France
Contact:

Post by Tibono2 »

Hi,

As far as the b6 reply to 1.h4 is reliable, my SC9 is a 'b' one
Tibono2 wrote:Hi,

just received a new toy - nice Sensory 9 which reveals to be a "B" one (serial 32xxxx, reply to 1.h4 is b6) and 2Mhz clock.
...
Sensory Chess Challenger 9 SC9B 32381304 Friday, August 26, 1983

Maybe the production started a few months before the 9B officially reached the shops... or maybe my board label is one not be considered (model number SU9 being wrong, the whole data printed on this label is doubtful).

This does not help regards :?
Tibono
Larry
Senior Member
Posts: 2269
Joined: Wed Aug 01, 2007 8:42 am
Location: Gosford, NSW Australia

Post by Larry »

spacious_mind wrote:
OK so it is a well done Larry! :)

Best regards
Awww....shucks
Regarding the Fidelity endgame module never being released
for sale. Yes that's true, but the source code inside the module
did not go to waste. It was probably copied and pasted into the
ROM for the 9B, along with the slight speed increase.
Just my guess regards..
Larry
yoyo_chessboard
Member
Posts: 126
Joined: Mon Oct 17, 2016 5:44 pm
Location: South of France - cote d'azur

Post by yoyo_chessboard »

i tested my SC9A(s).

for serial: 32171952 clock: 1.88Mhz (verified with logic analyser)
and
ROM_START( fscc9 )
ROM_LOAD("101-1034a01", 0xc000, 0x2000, CRC(b845c458) SHA1(d3fda65dbd9fae44fa4b93f8207839d8fa0c367a) ) // HN48364P
ROM_LOAD("101-1034b02", 0xe000, 0x2000, CRC(cbaf97d7) SHA1(7ed8e68bb74713d9e2ff1d9c037012320b7bfcbf) ) // "
and PCB 5

larry's mate in 3: 75 seconds
colditz7: 3min40 -> 220 seconds (runned twice)
colditz11 = 5min 8s -> 308 seconds (runned twice)

colditz11: for a long time, the SC9 considered A8-D8 as the best move before proposing the expected move.

just a trick for those (as me) that don't know how to play black first
as i didn't manage to make the black play first using the D1 key (prior of after C5), i set up a position 'almost colditz 11' and played manually white queen to arrive in the colditz11 position.
immediatly the SC9A started its computations. i think that doesn't alter the results

second
serial: 23305465 clock : 1.60 Mhz (verified with logic analyser)
same ROMset
larry's: 75 seconds. very difficult to identify a real difference with 1.88 Mhz board
but for colditz7: 4min20 -> 260 seconds
40 seconds more than with 1.88 Mhz.

i haven't yet tested the colditz11 with that board.
i will do it this week-end.
User avatar
spacious_mind
Senior Member
Posts: 4000
Joined: Wed Aug 01, 2007 10:20 pm
Location: Alabama
Contact:

Post by spacious_mind »

Larry wrote:
spacious_mind wrote:
OK so it is a well done Larry! :)

Best regards
Awww....shucks
Regarding the Fidelity endgame module never being released
for sale. Yes that's true, but the source code inside the module
did not go to waste. It was probably copied and pasted into the
ROM for the 9B, along with the slight speed increase.
Just my guess regards..
Larry
If your guess is correct then that would make the 9B quite unique.
Nick
yoyo_chessboard
Member
Posts: 126
Joined: Mon Oct 17, 2016 5:44 pm
Location: South of France - cote d'azur

Post by yoyo_chessboard »

i asked 'blitzchess', a french collector, some help to have more information about SC9 family.

i was very kind and sent me 4 and half SC9.
why 4 and half : because there are 4 full boards and a mother board without sensory surface, because of an upgrade.
with my 3 three SC9A that makes a good starting point to reorganize that mess in SC9 family.

he is a very lucky man :
all 4 four of his SC9 have different PCB.
and he got a ROM revision nobody else spoke about prior, to what i know:

in fact there are now (at least)3 revisions of the PCB of the SC9 and there are also 3 revisions of the software running in them.

there is a pcb : 510-1046 revision B01 which is running software :101-1034A01 and 101-1034A02
found in SC9 : 21482215
never heard before

there is a pcb : 510-1046 revision C01 which is running software :101-1034A01 and 101-1034B02
found in SC9 : 32170966
-> this one is what we commonly call SC9A
this board is running at various speed , at least at 1.5, 1.6 , 1.9 Mhz
date code we have (at least) for that revision : 8242/8243 and 8325/8326


there is a pcb : 510-1046 revision D01 which is running software :101-1034B01 and 101-1034C02
found in SC9 : 33296781

-> this one is what we commonly call SC9B

i haven't measured the real clock at the moment.
the hardware revision consist of suppressing a chip or replacing one by another one.

so we have now a small problem of labelling :
how to call each of the three revisions:

how, for example, name the revision B board that is prior to what we call SC9A. (revision C board).

and do we agree to call SC9B only those boards that have pcb : 510-1046 revision D01 which is running software :101-1034B01 and 101-1034C02
we know that there are at least two variants of that board : with ou without the daughter board.

Blitzchess is also a happy owner of the SC9 running at 8 mhz.
i have made some photos of that board. I know that some members are against upgraded board, even if myself i consider that rather positive if well realised.


on my flickr albums you will find all photos i took about all that mess of boards and ROMs, making albums for each revision of the boards.
https://www.flickr.com/photos/149310091@N02/albums
as it is for further investigations the photos are heavy (6 Mbytes each) in order to zoom without loss of quality.
i haven't tried to compress them a little.

i have still many things to do :
measure each clockrate
dump the ROM for the very first board ( i didn't manage to dump them for the moment)

some questions for larry and mike :

on the lower side of the revision D board there is a date code on the board i have: it is said 83 34 as week 34 of 1983.
so it suggest that that revision B was "technically introduced" at end of august 1983, even if commercially it is rather in october 1983.
that matches with the datecode on the ROMs : week 43/1983 at least for what we know for the moment.
may be they were just waiting the ROM to be produced.
do you have a datecode on your boards ?

as you see bliztchess board hasn't the daughter board, just the sandwich with two TTL chips one on the other.

we have opened the pandora box :
i think there are many other things that we can deduce from those photos and those boards, and further investigations.

we have now more data to rewrite the roadmap of SC9 model.

may be we could open a new topic like' SC9 different revisions'
yoyo_chessboard
Member
Posts: 126
Joined: Mon Oct 17, 2016 5:44 pm
Location: South of France - cote d'azur

Post by yoyo_chessboard »

Hi all,
i have continued the measures around the SC9 family.
i have , at the moment, the following results :

http://www40.zippyshare.com/v/9rcdy5G8/file.html

i have used the larry's mate in three and colditz 7 and 11 positions.
i have also used a position that is in a fidelity add from the french europe echecs magazine n°322 from october 1985.
it is an add for the 'new excelllence' that is stronger than the SC9B.
it is the position 1 at the left bottom of the page.
it is simple to set up as there are only 4 figures and computtion time is mid-range : not too short (and you can't see any difference between two boards), not to too long to wait.

http://www40.zippyshare.com/v/2fIib56o/file.html

the boards measures :
all the SC9A that i have seen (some mines, some from Blitzchess collection)
the SC9B from Blitzchess

and also some special guest :
the SC9 that is running at 8 Mhz.
i dumped its ROM and it is a SC9B ROM.

my SC9A running at 4 Mhz (modified by myself).

i also added for reference a designer 2100 display just to see the progress in computation between 1982-1984 (SC9 range) and 1989-1990 (designer 2100 display).
i have modified my designer 2100 display and it is working at 10 Mhz.
i don't have results for a 'factory' designer 2100 display with is running at 6 Mhz.

i think there are many ways to read that spread sheet.

each owner can make the tests and compare with the results. The speed clock was measured with a logic analyser .
so the measure is precise and can be taken as a reference.
if you have the same results you have the same clockrate.

there are measures from the SC9B from Trilobyte (the Master Modder as called here) at 8 Mhz and we can see the interest to have a higher clockrate for a given sofware engine.

larry , you could verify if your SC9B at 8 Mhz gives the same results.

there is also now a clear way to classify SC9A and SC9B , according to the reply to 1h2-H4 but also by time taken to solve a given position, whatever the clockrate.

i also introduced the notion of distance :
an example of the real life. when you cruise from city A to city B, there is X kilometers.
if you cruise at 60 km/h you will take more twice time than if you cruise at 120 km/h.
but if you multiply time* speed you have a 'constant'.

for each position i have tried to compute that constant :
it could be translated by (take for example larry' mate in three in SC9B):
it takes 61 seconds at 1.9 Mhz to be solved.
so it takes about 115.9 Millions of CPU cycles to be solved.
it is a way to verify if you measure is 'logic', in the right range of values.


that ratio is also interesting to compare two software engines :
for larry's mate in three:
the SC9B has a ratio of 112/120 depending of clockrates and measures
the designer 2100 display has a ratio of 80, so we could say that it sees faster the right move.
we could say that the designer is in progress compared to SC9B for that field.

one surprise :
the designer 2100 display at 10 Mhz has better result on larry's mate in three.
but it takes more time to solve colditz 11 problem.
probably that the evaluation function is more complex and it takes more time to evaluate each node.
Larry
Senior Member
Posts: 2269
Joined: Wed Aug 01, 2007 8:42 am
Location: Gosford, NSW Australia

Post by Larry »

yoyo_chessboard wrote: that ratio is also interesting to compare two software engines :
for larry's mate in three:
the SC9B has a ratio of 112/120 depending of clockrates and measures
the designer 2100 display has a ratio of 80, so we could say that it sees faster the right move.
You are referring to this position, or a variant of it:
Fidelity SC9B 8mhz
[fen]8/8/4R3/8/8/4K3/8/4k3[/fen]
Mate in three: 13.8 seconds, give or take a quarter of a second. I used
the timer on my iPad.
Please excuse the reversed files. I still can't seem to figure out
how to do it right.
Larry
yoyo_chessboard
Member
Posts: 126
Joined: Mon Oct 17, 2016 5:44 pm
Location: South of France - cote d'azur

Post by yoyo_chessboard »

Larry wrote:I just occurred to me that a good position to test which speed SC9 you
have is this mate in 3 with white to move:
[fen]5k2/8/5K2/8/8/5R2/8/8[/fen]
The original SC9 will take around 70 - 75 secs depending on whether
it is the 1.5mhz or the slightly later 1.6mhz. The SC9B will mate in around
50 secs. White starts with *any* rook move.
If yours is a 3mhz or 3.2mhz it can be expected to take around 35 seconds.
You will have the level set at level 9 (infinite).
Hoping you get a nice surprise...
Larry

Hi larry
i referred to that position from your message of 27 march 2016
top of page 2 of that topic.

regards.
BillT
Member
Posts: 132
Joined: Sat Jun 09, 2012 9:31 pm
Location: Norfolk UK

Post by BillT »

Hi
This topic is still very interesting and I would like to help,so I have opened up my SC9B in order for you guys to see the details and dates of chips and crystal etc
Hopefully you will be able to view the photos using the link below.(not done this before --hope it works !)

https://drive.google.com/folderview?id= ... XNSSDJ5ak0

It'll be interested to hear your thoughts...if you need any more images etc let me know

I also did the same with my SC9A

https://drive.google.com/folderview?id= ... DNPcmozSDg

Fascinating regards
Bill
yoyo_chessboard
Member
Posts: 126
Joined: Mon Oct 17, 2016 5:44 pm
Location: South of France - cote d'azur

Post by yoyo_chessboard »

Hi Bill_T

your SC9A s very interesting (whatever real speed it is)

for the moment i have seen three couples of PCB/ROMs

PCB 510-1046B-01 with ROMS 101-1034A01 and 101-1034A02
PCB and ROMs identified for the first time in that topic and coming from Blitzchess's SC9.

PCB 510-1046C-01 with ROMS 101-1034A01 and 101-1034B02
what we commonly call SC9A and that is in MAME emulator

PCB 510-1046D-01 with ROMS 101-1034B01 and 101-1034C02
what we commonly call SC9B and that is in MAME emulator

but you have a mix of the two first :

PCB 510-1046C-01 with ROMS 101-1034A01 and 101-1034A02

that confirms that the first software revision is totally compatible with the 1046C board revision

i have tested that on my side but i didn't had a real proof, just trials that worked.
your board is an 'official' proof that it works.

when disassembling the code i saw very few differences with the binary we commonly call SC9A.
so the first and second binaries are swappable.
that also means that those who have the very first release with ROMS 101-1034A01 and 101-1034A02 can put ROMS 101-1034A01 and 101-1034B02

but on the contrary SC9A and SC9B binaries are not swappable as it.
there is one component that differs on the board , the LS244 is replaced by a LS251 and so the IRQ software routines that handles the sensory surface are different from SC9A and SC9B.

so a SC9A is not directly upgradable in SC9B from the sofware point of vue.
and i have not tried at the moment to make the sofware modifications to run a SC9B software on a SC9A PCB.

regards.
Post Reply