💾 Archived View for spam.works › mirrors › textfiles › computers › ultra.txt captured on 2023-06-14 at 16:06:38.
-=-=-=-=-=-=-
? Gravis Ultrasound Tech Specs ?
????????????????????????????????
The Unofficial Dox
Written December, 1992
By Tran of Renaissance and Joshua Jensen
OK, folks, here it is. This is the programming information that Gravis
didn't and won't give you any time soon. These dox have been obtained
through long hours of debug sessions and experimentation.
Let's clear up some things that have been debated in comp.sys.ibm.pc.soundcard
for quit sometime now:
? The GUS can play 32 mono samples at once, fully pannable through 16
different positions.
? This is still a bit unclear. The SDK says that samples have to be
32-byte aligned in memory. I've been following this. Maybe that is
not the case, though, and I haven't tested it.
Credits: Tran came up with the initial information about programming the
card. Josh documented it and included actual assembly language information
in this archive on how it works. It's relatively simple, but some things are
unclear and are explained here in the dox and in the code.
??????????????????????????????????????????????????????????????????????????????
Methods ?
?????????
@DELAY = (in byte 300h)*7
Description: All this function does is give a short idle wait for the card
to do whatever processing it needs to.
??????????????????????????????????????????????????????????????????????????????
Probe Ultra Sound ?
???????????????????
out BASE+103h, 4Ch
out BASE+105h, 0
@DELAY
@DELAY
out BASE+103h, 4Ch
out BASE+105h, 1
Poke Data Byte Loc: 0h, Byte: AA ; Don't think this HAS to be an AA.
Poke Data Byte Loc: 100h, Byte: 55 ; Don't think this is needed.
Peek Data Byte Loc: 0h
Store Byte
out BASE+103h, 4Ch
out BASE+105h, 0
Restore Byte
Is it AA? If so, then we have found a GUS.
??????????????????????????????????????????????????????????????????????????????
Test for amount of memory installed on UltraSound ?
???????????????????????????????????????????????????
Poke Data Byte Loc: 40000h, Byte: AA
Peek Data Byte Loc: 40000h
Is it an AA? If not, then there is 256k of DRAM. Exit.
Poke Data Byte Loc: 80000h, Byte: AA
Peek Data Byte Loc: 80000h
Is it an AA? If not, then there is 512k of DRAM. Exit.
Poke Data Byte Loc: C0000h, Byte: AA
Peek Data Byte Loc: C0000h
Is it an AA? If not, then there is 768k of DRAM. Exit.
Poke Data Byte Loc: 0FFFFFh, Byte: AA
Peek Data Byte Loc: 0FFFFFh
Is it an AA? If not, then there is 1024k of DRAM. Exit.
Description: Poke these bytes on the boundaries of 256k, 512k, 768k, and
1024k. If the memory isn't there, then it won't return what
is poked.
??????????????????????????????????????????????????????????????????????????????
Peek Data Byte ?
????????????????
out BASE+103h, 43h
out BASE+104h, low word address
out BASE+103h, 44h
out BASE+105h, high byte address
in BASE+107h, INBYTE
Description: The card's memory can be from 00000h to fffffh. INBYTE is the
byte value returned from the card that was at that memory
location.
??????????????????????????????????????????????????????????????????????????????
Poke Data Byte ?
????????????????
out BASE+103h, 43h
out BASE+104h, low word address
out BASE+103h, 44h
out BASE+105h, high byte address
out BASE+107h, OUTBYTE
??????????????????????????????????????????????????????????????????????????????
Set Voice Frequency ?
?????????????????????
out BASE+102h, voice number
out BASE+103h, 1
out BASE+104h, word frequency number (actual frequency / 19.0579083837)
Description: Tran thought he'd give you the exact value :) Dividing by 19
will give enough accuracy for the task, though.
??????????????????????????????????????????????????????????????????????????????
Set Voice Balance ?
???????????????????
out BASE+102h, voice number
out BASE+103h, 0ch
out BASE+105h, byte balance (0-0fh)
Description: The Balance byte can range from 0 to 15. 0 is far left,
15 is far right, and 7 is right in the middle.
??????????????????????????????????????????????????????????????????????????????
Set Voice Volume ?
??????????????????
out BASE+102h, voice number
out BASE+103h, 9
out BASE+104h, word volume value (0-0ffffh, log ... not linear)
??????????????????????????????????????????????????????????????????????????????
Set Loop Mode ?
???????????????
out BASE+102h, voice number
out BASE+103h, 80h
in BASE+105h, TEMP
out BASE+103h, 0
out BASE+105h, (TEMP & 0e7h) or MODE
Description: This is not really necessary. Actually, the Set Loop Mode and
Stop Voice Immediately could be one function.
??????????????????????????????????????????????????????????????????????????????
Stop Voice Immediately ?
????????????????????????
out BASE+102h, voice number
out BASE+103h, 80h
in BASE+105h, TEMP
out BASE+103h, 0
out BASE+105h, (TEMP & 0dfh) | 3
@DELAY
out BASE+103h, 0
out BASE+105h, (TEMP & 0dfh) | 3
Description: Dunno why the extra delay and duplicated code is there.
??????????????????????????????????????????????????????????????????????????????
Start Voice Playing ?
?????????????????????
out BASE+102h, voice number
out BASE+103h, 0ah
out BASE+104h, word BEGIN >> 7
out BASE+103h, 0bh
out BASE+104h, word BEGIN << 9
out BASE+103h, 2
out BASE+104h, word START >> 7
out BASE+103h, 3
out BASE+104h, word START << 9
out BASE+103h, 4
out BASE+104h, word END >> 7
out BASE+103h, 5
out BASE+104h, word END << 9
out BASE+103h, 0
out BASE+105h, MODE & 0fch
??????????????????????????????????????????????????????????????????????????????
Read Voice Location ?
?????????????????????
out BASE+102h, voice number
out BASE+103h, 8ah
; in BASE+104h, low byte TEMP0
; in BASE+105h, high byte TEMP0
in BASE+104h, word TEMP0
out BASE+103h, 8bh
; in BASE+104h, low byte TEMP1
; in BASE+105h, high byte TEMP1
in BASE+104h, word TEMP1
LOC = ((TEMP0 << 7) | (TEMP1 >> 9)) & 0xfffff
Note: Word read at BASE+104h seems to work same as sequential byte reads
from BASE+104h then BASE+105h.
??????????????????????????????????????????????????????????????????????????????
??????????????????????????????????????????????????????????????????????????????
Ports ? As you can see, there are a lot of blanks to be filled. Hope Gravis
??????? will take a hint soon.
BASE - write only, mix control
?????????????????????????????????
? 7 ? 6 ? 5 ? 4 ? 3 ? 2 ? 1 ? 0 ?
?????????????????????????????????
? ? ? ? ? ? ? ?
? ? ? 0 = Line in on. 1 = Line in off.
? ????? 0 = Output on. 1 = Output off.
????????? 0 = Mic in off. 1 = Mic in on. ???
BASE+102h - Active voice select (byte port)
BASE+103h - command port (byte port)
Bitmap so far:
?????????????????????????????????
? 7 ? 6 ? 5 ? 4 ? 3 ? 2 ? 1 ? 0 ?
?????????????????????????????????
? ? ? ? ? ? ? ?
? ? ? ? ? ?
? ? ? ? ?????
? ? ? ????????? Command for voice
? ? ?????????????
? ????????????????????????? Command for card.
????????????????????????????? Read from voice.
0h - write voice mode (loop and data type) - byte on 105h
?????????????????????????????????
? 7 ? 6 ? 5 ? 4 ? 3 ? 2 ? 1 ? 0 ? Designated as MODE through dox
?????????????????????????????????
? ? ? ? ? ? ? ?
? ? ? ? ? Voice On/Off (If both set to 0, voice on.)
? ? ? ????? Voice On/Off (If both set to 1, voice off.)
? ? ????????? 0 = 8 bit data. 1 = 16 bit data.
? ????????????? 0 = No loop. 1 = Loop Sample.
????????????????? 0 = Go forward. 1 = Go backward.
1h - voice frequency - word on 104h
2h - loop start location / 80h - word on 104h
3h - loop start location << 9 - word on 104h
4h - loop end location / 80h - word on 104h
5h - loop end location << 9 - word on 104h
9h - voice volume - word on 104h
0ah - voice location / 80h - word on 104h
0bh - voice location << 9 - word on 104h
0ch - voice balance - byte on 105h
0eh - Unsure. Has something to do with turning - byte on 105h
stereo on and off. See U_Reset for an
example of usage.
43h - low word of DRAM address - word on 104h
44h - high byte of DRAM address - byte on 105h
4Ch - Unsure exactly of what this does - byte on 105h
However, a 1 must be output to BASE+105h for
the card to do memory writes. At least, that's
the only way I could get it to work.
80h - read voice mode - byte on 105h
8ah - read voice location - bytes on 104h, 105h
8bh - read voice location - bytes on 104h, 105h
Notes: Any >= 80h plus a number might be the read mode of the value - 80h.
In other words, to read the voice volume (this is untested), you'd
output an 89h through BASE+103h and then read BASE+105h.
BASE+104h - word writes/reads
BASE+105h - byte writes/reads
??????????????????????????????????????????????????????????????????????????????
Notes ?
???????
To stop an instrument when the sample ends (not immediately), change its
loop type to 'no loop'.
If you are making a lot of changes to just ONE voice, the
OUT BASE+102h, Channel needs only to be done once.