From one different dump it's hard to make any solid conclusion.
If some other dump shows same picture - it's artifact of some ASP testing.
If found other dump with few bits changed - it's sign of faulting ROM.
Currently, I assume this is faulting ROM case, as ALL other dumps is absolutely the same in low 6k of ROM.

mattw wrote on 2023-10-02, 21:24:

maybe do not analyze them just upload them here - one with v0.02 and one with v0.03. thanks!

here you go.
i just zipped both complete folders to not forget anything.

In both versions its 0x80 ... but the checksums seem not to be identical... maybe its really a failing chip...

Next version of sbdump will be with integrated CRC32 calculation and comparing.
So it will be easy to do fast repeated dumps and check results.

DerBaum wrote on 2023-10-02, 21:48:

but the checksums seem not to be identical... maybe its really a failing chip...

it's actually very odd looking, could be something else fails in your system - RAM, HDD and they introduce that random byte(s) corruption.

I wonder if it's possible that newer integrated chips like "ViBRA 16XV" upgrade-able in software and silently the drivers over the years updated them from DSP V4.13 to V4.16 - I am starting to think that, because even very reputable sites like retronn.de:

https://retronn.de/imports/hwgal/hw_ct4170.html

list CT4170 as "V4.13", but I have 4 x CT4170 cards here and they are all DSP V4.16 - that really starts to look very suspicious to me. I found already several websites list CT4170 as V4.13, that's really strange, because Retronn.DE card is made 1997 as mine four. I don't know...

mattw wrote on 2023-10-03, 00:43:

I wonder if it's possible that newer integrated chips like "ViBRA 16XV" upgrade-able in software and silently the drivers over the years updated them from DSP V4.13 to V4.16 - I am starting to think that, because even very reputable sites like retronn.de:

https://retronn.de/imports/hwgal/hw_ct4170.html

list CT4170 as "V4.13", but I have 4 x CT4170 cards here and they are all DSP V4.16 - that really starts to look very suspicious to me. I found already several websites list CT4170 as V4.13, that's really strange, because Retronn.DE card is made 1997 as mine four. I don't know...

Im willing to bet its something like this;
019720 is probably the 4.13 version (Black audio sockets)
049730 the 4.16 one (Colored audio sockets)

You can also check the back of the card and in the top right it should have the manufacturing date (should say something like 94v-a 4398 which is the 43th week of '98)
Creative done the same on the ct1740, the early versions 039237 have 4.04 or 4.05 while the last version 069346 has 4.11

mattw wrote on 2023-10-03, 00:43:

list CT4170 as "V4.13", but I have 4 x CT4170 cards here and they are all DSP V4.16 - that really starts to look very suspicious to me. I found already several websites list CT4170 as V4.13, that's really strange, because Retronn.DE card is made 1997 as mine four. I don't know...

May be early 97 cards was 04.13, but late 97 was 4.16 ))
Integrated flash memory, complicated update interface, hidden software update... For only one update done... Not trying to sell marketing tales based on this... Dangers of firmware exposure during update... Too much trouble for so little gain))

I have two vibra 16Cs, one CT4180 from 9th week 1998 but it has 4.13 and a CT2960 from 1996. I also have a vibra 16S that is older, which I believe is also 4.13. Yet to see a vibra with 4.16

DerBaum wrote on 2023-10-02, 20:00:

ct2940 has a real opl and on ct3600 and ct 2950 it is not present.
I dont know how they decide to use the opl or the creative emulation. maybe in software by a setting a byte. But it also can be done on the hardware side to disable CQM when the OPL is present.

I have a CT2950 with YMF289B instead of CQM. As for how they choose to use OPL or CQM, perhaps one needs to actually examine the PCB layout to find out.

It could be as simple as hard-wiring a specific I/O pin high or low and the DSP only needs to check the state of that pin to know what's connected there.

As for the later, integrated versions, depending on the design, it may work in a way that whether the DSP is integrated or discrete does not matter (so the same FW code can be used as-is).

I'm not sure about the difference between CQM and real OPL. I have AWE64 Gold which supposedly uses CQM and I could not hear any difference in the games I usually play.

I have about 30 SB 16/AWE cards I can dump, IIRC. I just need to go back and re-read what kind of contraption I need to build.. Do you still want me to do that? I can make a list of the models I have.

appiah4 wrote on 2023-10-03, 07:17:

I have about 30 SB 16/AWE cards I can dump, IIRC... I can make a list of the models I have.

please, do make at least a rough list.

appiah4 wrote on 2023-10-03, 07:17:

I just need to go back and re-read what kind of contraption I need to build.. Do you still want me to do that?

In the mean time I will write a small guide - not to waste time read - it's very simple - I needed to do "building", because I did not have running PC system with ISA slots - if you have one is more or less, swapping SB cards and running a tool.

Maelgrum wrote on 2023-10-03, 01:34:

May be early 97 cards was 04.13, but late 97 was 4.16 ))
Integrated flash memory, complicated update interface, hidden software update... For only one update done... Not trying to sell marketing tales based on this... Dangers of firmware exposure during update... Too much trouble for so little gain))

yes, I understand it sounds really far fetched, but neither I can proof it's that simple:

S95Sedan wrote on 2023-10-03, 01:19:

Im willing to bet its something like this;
019720 is probably the 4.13 version (Black audio sockets)
049730 the 4.16 one (Colored audio sockets)

because looking at the PCB dates and also the date code on "ViBRA 16XV" chip itself, it varies from 1996 to 1997 in my case, but all are V4.16: I will make you exact list of those 4 boards with dates. So, I am real interested someone with "ViBRA 16XV" and DSP 4.13 to report here.

DSP fw has absolutely no code related to FM (opl).
It's designed for SB command processing/midi handling/ASP stuff.
Access to FM chip done via Bus control/interface chip, and in case of external FM, is simple port decoding.

OK, below is the list of my CT4170 (ViBRA 16XV) cards - we need more people to give feedback on this, because IMHO date-code doesn't explain the DSP V4.13 vs DSP V 4.16, especially when I have card with RevA "ViBRA 16XV" chip made in 1997 that is still DSP V4.16, while all other 3 cards have RevB "ViBRA 16XV" chip made in 1998 - I mean at least I expect to change the Chip Rev when they changed the DSP version:

1* my CT4170 card 1 (DSP V4.16):
2- on PCB front side: "049730" - it doesn't look it's 1997-date-code though, because all cards I have, have that same number even those manufactured end of 1998 has that number
3
4- on PCB back side: "3797", i.e week 37 of 1997
5
6- ViBRA 16XV: "3597", i.e week 35 of 1997 (consistent with date-code on "PCB back side"), 
7
8- the chip is "CT2511-SAT, i.e. that has to be Rev. A, as others are "-SBT", i.e. rev. B
9
10===
11* my CT4170 card 2 and card 3 (they have the same markings and are DSP V4.16):
12
13- on PCB front side: same as card1
14
15- on PCB back side: "3898", i.e week 38 of 1998, over 1 year newer than card1
16
17- ViBRA 16XV: "9810", i.e week 10 of 1998 (kind of consistent with date-code on "PCB back side"), at least as far it's the same year of 1998
18
19- the chip is "CT2511-SBT, i.e. what has to be RevB chip
20
21===
22* my CT4170 card 4 (DSP V4.16):
23
24- on PCB front side: same as all previous cards, i.e. not a manufacturing date-code, maybe it's PCB revision number
25
26- on PCB back side: "4997", i.e week 49 of 1997, almost end of 1997
27
28- ViBRA 16XV: "9748", i.e week 48 of 1997 (consistent with date-code on "PCB back side")
29
30- the chip is "CT2511-SBT, i.e. what has to be RevB chip

mattw wrote on 2023-10-03, 00:37:

DerBaum wrote on 2023-10-02, 21:48:

but the checksums seem not to be identical... maybe its really a failing chip...

it's actually very odd looking, could be something else fails in your system - RAM, HDD and they introduce that random byte(s) corruption.

I have the feeling one of my ISA slots is dodgy on my test system. I have to investigate this. i will do some dumps of the ct3600 in another system and see if they change.

Please wait until next version of sbdump will be done.
With crc checking, and some fw classification mechanism integrated, it will be much easier to do massive dumps and checks.

Maybe this is of interest to you :
I found another number to better identify creative cards in my collection.

There is a black number printed on the back of the card on the top right corner of all creative cards wich exactly describes the type of the card, even if the stickers dont tell you.

In this case it is a CT4171

The number on the front, the number on the sticker and also the number in the serialnumber all say 4170 ... BUT the small black number on the top right says 71 . So it is a CT4171.

BTW. This is a 4.16 version.

I found this while cataloging soundblaster live cards.

mattw wrote on 2023-10-03, 10:32:

appiah4 wrote on 2023-10-03, 07:17:

I have about 30 SB 16/AWE cards I can dump, IIRC... I can make a list of the models I have.

please, do make at least a rough list.

I have these:

Creative Sound Blaster 16 CT1740
Creative Sound Blaster 16 CT2290
Creative Sound Blaster 16 Value CT2770
Creative Sound Blaster 16 CT2920
Creative Sound Blaster 16 VIBRA CT2800
Creative Sound Blaster 16 VIBRA CT2890
Creative Sound Blaster 16 VIBRA CT2940
Creative Sound Blaster 16 VIBRA Vaue CT2960
Creative Sound Blaster 16 VIBRA CT2980
Creative Sound Blaster 16 WavEffects CT4170
Creative Sound Blaster AWE32 CT3980
Creative Sound Blaster AWE64 Value CT4500
Creative Sound Blaster AWE64 Value CT4520

I have multiples of some, if that makes a difference. That's why I said about 30 😓 I think I have 7 AWE64s 🤣

Now, about that guide?

thank you @DerBaum - so based on the picture you posted above, now @S95Sedan theory about :

S95Sedan wrote on 2023-10-03, 01:19:

019720 is probably the 4.13 version (Black audio sockets)

is disproven. I really don't know - I doubt so much reputable website made a typo or a mistake CT4170 is DSP V4.13 and now all such cards we reported here are DSP V4.16 - still my hypothesis for a silent software/driver update from DSP V4.13 to V4.16 done on those cards makes the most sense, unless someone doesn't report here CT4170 with V4.13 DSP.

Idea is:
1. Calculate 6k CRC and full 8k CRC.
2. If 8k CRC is equal to known, write a dump 'ct1741_v413.bin' and note in log that is known version
3. If 6k CRC is equal to known, write a dump 'ct1741_v413i.bin' with 6k length and note in log that is known version (i - means integrated)
4. In all other cases, write a dump 'XXXXXXXX.413', where XXXXXXXX is 6k CRC

This filename classification helps for easy understanding what is happened and what to do.