You can order the company’s “open letters”
Fragment 1/11.
You can order “open letters” from the (c)Nemo company by mail. (An IS-DOS-format disk with lots of letters in text format.)
Address of (c)Nemo:
198261, St. Petersburg, P.O. Box 213.
Don’t forget to include 5 rubles’ worth of postage stamps.
(c)Nemo company phone: (812)159-5569, from 22:00 to 24:00 MSK.
You can get the latest price list from (c)Nemo at the address above
and place an order.
╔══════════════════════════════════════════════════════════════╗
║ Contact with the (c)Nemo company in the Moscow region: ║
║ ordering KAY motherboards, HDDs and various peripherals, ║
║ receiving correspondence. Shipping by rail via Moscow is ║
║ possible. ║
║ (095)333-1471 Boris Sharapov aka OPER ║
║ from 19:30 to 23:30 MSK ║
║ ZXNet : Boris Sharapov (500:95/333) ║
║ INTERNET: zxnet@iname.com ║
║ ( Subj:500:95/333 ) ║
║ 117279, Moscow, General Antonov St., bldg. 2, apt. 106 ║
╚══════════════════════════════════════════════════════════════╝
∙ ∙ ∙
╔══════════════════════════════════════════════════════════════╗
║ 'ALL KINDS OF RAM DISKS ARE NEEDED, ALL KINDS MATTER' ║
╚══════════════════════════════════════════════════════════════╝
Area : .....(censored :-)
From : Alexey Voskresensky (Las)
To : All users of KAY-1024
Subj : NEW! — 'RAM DISK EMULATING THE VG93'!
Date : 05-06-00
Time : 14:00:00
Hello to all users of KAY-1024!
Hi, guys! It’s Las from FLASH GROUP again, and I’d like to tell you in more detail about my creation for KAY-1024.
This creation is my RAM disk driver, flashed into the KAY ROM. It emulates the operation of a real disk drive, a disk-drive controller and, naturally, the “brain” of the disk controller—the КР1818ВГ93 chip, with all its ports, commands and the consequences that follow from them. :) A bit of know-how, by the way. ;)
So long!
Lyokha, aka Las.
P.S. This article could be considered an advertisement. :) If anyone gets offended, it’s not my fault! :P
P.P.S. I hope the respected “coders” will find something of interest in this account.
P.P.P.S. Sorry for the complicated didactic constructions. :[] :P :) I wanted to make it simpler, but it turned out… Anyway, you be the judge, not me. I did what I could.
You Can Order “Open Letters” from the Company
Fragment 2/11.
LAS,Flash inc
<────────────
the country should
know its heroes
by sight
:-)))
%%%%%%%%%%%%%%
“ALL KINDS OF RAM DISKS ARE NEEDED, ALL KINDS OF RAM DISKS ARE IMPORTANT”
(an epoch-making work by Dr. Laspersky)
A PREFACE FOR NEWBIES. “WHAT IS THIS THING—A RAM DISK?”
Lately, users have started getting hard drives. I don’t need to explain what those are. :) A HDD controller was created for the KAY-1024—a small board that plugs into one of the KAY’s slots and serves to exchange data between the HDD and the computer. But this little board alone isn’t enough for productive work. It turns out :) that you also need software support for these “three pillars”—in other words, a driver. I’m going to try to tell you about one of the drivers: the RAM-disk driver (don’t confuse it with the hard-drive driver!).
Nemo acted sensibly: HDD_BOOT was burned into the TR-DOS ROM; it loads and starts iS-DOS from the hard drive. In iS-DOS, we choose the disk we need from the DISK IMAGES (something like .trd files for a Spectrum emulator on PC) stored on the hard drive. Using the hard-drive driver (which has absolutely nothing to do with the RAM-disk driver), iS-DOS reads the contents of the DISK IMAGE from the hard drive and (most importantly!) COPIES this DATA into the KAY-1024’s RAM. Then you can press Reset, keep working in iS-DOS, or go have a beer—but the main thing is, don’t unplug your KAY from the mains :), because the disk-image data has been copied and is stored in RAM. And as long as the computer is switched on, the contents of the disk image won’t be destroyed. In the KAY-1024, the disk-image data is stored in the upper 640 Kb of RAM.
But what good is it having some data somewhere in RAM? How can an ordinary (and not-so-ordinary :)) user make use of it? So that the user could “see” this virtual RAM disk and use it successfully, St Petersburg programmer Konstantin Verbov wrote a RAM-disk driver. It was burned into the TR-DOS ROM and, at certain moments, took control so that you could use this 640K of RAM like an ordinary
In fact, you absolutely do NOT need to have a hard drive and iS-DOS to use a RAM disk. A “bare” RAM disk can be created using TR-DOS itself—for example, by formatting it, since the “RAM-disk driver (c) Konstantin Verbov” lets you at least clear the system track, and then you’ve got “like, a clean disk for real!” You’ll understand my irony by reading just…
the first and second chapters of this tale.
In general, this driver was the “first swallow,” doing a pretty good job of what it was designed to do. But... Not everything is as smooth as it itches... Read on.
CHAPTER ONE. “A PEBBLE IN SOMEONE ELSE’S GARDEN.”
After using Nemo’s KAY megabyte RAM disk for a while, I started going nuts over the INABILITY to LOAD all sorts of programs from the RAM disk that DIRECTLY USE the VG chip in their operation. I really wanted to run “FUT_2.xx” or “RDS” and format, restore, and check the RAM disk... :):)
But that’s just a digression. Here’s the main problem: some programs with MUSIC-, TURBO-, and STEALTH-LOADers won’t load.
What we’ve got: 10% of programs have to be stored not on the hard drive, but on floppy disks. Or kept in a special archive on the hard drive, copied to a floppy when needed, and then run from there. But that’s not all!!! (Annoying, huh?!)
CHAPTER TWO. “THE PATHOLOGIST’S HOUR.”
Captain Nemo (owner of Nemo Ltd.) used to ship the KAY-1024 with ROM chips :)) containing the latest version of Kostya Verbov’s RAM-disk firmware, namely version 9901. Let’s dig deeper into this “wonder” to find out where the smell is coming from.
More or less significant changes (the rest are minor details) affected the following OS TR-DOS subroutines:
- Activating a logical device (selecting a drive, configuring it)
- Reading/writing a sector
-
Formatting a disk
Item 1. Activating a logical device (A, B, C, D)
(A reminder: the RAM disk is logical device.) How it works:
A call to the driver has been added to the subroutine to check which drive needs to be activated. If the drive being activated is <NOT “C”>, we exit the driver and configure the drive as usual. But if drive
is being activated, we ignore the access to the VG chip, set the appropriate DOS system variables, and exit. Significant drawback:
A program can select a drive by accessing the VG93 ports directly (say, system register #FF), bypassing this subroutine. Then we won’t activate or get any RAM disk at all.
Item 2. Reading/writing a sector
How it works:
Once again, the driver is called to check the current drive (which drive to read/write the sector from). If the current drive according to the DOS system variables is not C, we exit and read/write as usual. If the current drive is C, that means we have to work with the RAM disk. This is more complicated. In brief: we mark the current RAM page, use the DOS system variables to determine the current track/sector, transfer its contents from the required page of the RAM disk (i.e. “load” them) to the required location, find the active RAM bank using the marker, switch it in, change the TR-DOS system variables to the next track/sector, and exit. Writing to the RAM disk works the same way, only in reverse: we transfer the contents of the current sector to the RAM disk.
Major drawback:
Again, if the program reads/writes not through TR-DOS, but with its own disk loader/saver, then, naturally, we get nothing—even if the RAM disk was activated successfully at some point.
Item 3. Formatting a disk
How it works:
The disk-formatting subroutine once again calls the driver to check which logical drive is active. If the active drive is not C, it returns to standard formatting. If the RAM disk is to be “formatted,” the following happens: we zero the RAM disk’s system track, set the parameters of the service sector 9 according to TR-DOS standards (SS/DS, 40/80, free, etc., and so on), and exit, assuming that “formatting” was successful. Note that the entire RAM disk area (640 Kb) is not completely zeroed (“formatted”).
Major drawback:
If some program (for example, “FUT”, “RDS”, “DCU”, etc.) tries to format drive C using its own tools, naturally, it won’t work—in all likelihood, drive A will be formatted instead.
Minor drawback:
This is probably not even a drawback; I’m just nitpicking. :D When the RAM disk is “formatted,” only the system (0) track is “formatted” and initialized. All the others are left exactly as they were. In principle, most programs don’t give a damn about this, but still: if you’re going to “format” a disk, do it properly.
The biggest “ugh”
For this driver to work properly, the TR-DOS system variables must be present. That’s only natural: the driver was written on the assumption that it would work only through certain DOS subroutines, so:
FIRST CONCLUSION (which everyone has known for ages :)):
When calling the main TR-DOS system functions through #3d13, in most cases everything works fine.
CONCLUSION TWO (also obvious, and the most important :)):
Normal operation of the “RAM-disk driver (c)Konstantin Verbov” alongside programs that control the VG-93 controller (and its ilk) directly is IMPOSSIBLE for a number of reasons.
CONCLUSION THREE (probably too categorical):
We don’t need this driver!!! (Since we found a worthy replacement for it. What did you think I started this whole fuss for? :):):))
CHAPTER THREE. “EUREKA!”
Yes, guys, after a while I found myself wondering:
- Could the VG93 be emulated in real time, so that programs working with the VG93 directly wouldn’t notice they were being fooled?
Some people, for example, Андрей Ларченко (MOA), wrote that this was IMPOSSIBLE. I quote:
“Unfortunately, the Z80’s speed, even in Turbo mode, is not enough to fully emulate the 1818VG93 controller chip in real time.”
That claim made me furious: the Z80 can do anything—it’s only clumsy hands that can’t. I spent about a month thinking about this problem. The main difficulty wasn’t emulating the VG chip, but making the emulator “real-time” (read: accurate) and “invisible” to programs. I had to make the program work with the RAM disk exactly as it would with a real floppy drive, a real disk, and a real VG chip, so that it wouldn’t even occur to the programs to think: “Something’s not right here. Looks like someone’s slipped me some kind of emulator.” :):)
I wrote the program to emulate the VG chip and the floppy drive, with the RAM disk “inside” it, fairly quickly. The hardest part remained: feeding programs the VG chip’s port states and everything else. I also had to find some way to tell whether the “regular” floppy drive was working or the RAM disk was—this was no easy task either, since the driver mustn’t use any TR-DOS system variables to do its job. In fact, it mustn’t use even a single memory location, except in the upper memory of the KAY-1024 (the RAM disk’s 640K of memory).
In the end, after a month and a half of pondering and suffering, everything was finally written and debugged.
CHAPTER FOUR. “IF YOU DON’T PRAISE YOURSELF...”
You can order “open letters” from the company
Fragment 6/11.
All the bizarre loaders/savers work now! “BATTLE COMMAND 128K” by Stalker runs perfectly; “Format Utility” formats, checks and restores the RAM disk without even realizing it’s working with RAM rather than a disk. :) “Advanced Disk Service” works like a beast, and when you run its disk and drive tests, it even shows the disk’s rotation speed (the RAM disk!!!) as 300.0 rpm. All the test functions pass with flying colors, just as they do on a real floppy drive. In short, you now have a fully functional virtual floppy drive with a disk inside it :), and all operations (reading, writing, positioning the virtual heads, formatting, etc.) run faster than on a real floppy drive.
Everything works beautifully—I didn’t even expect it to work this well! :)
CHAPTER FIVE. “A LITTLE UNRESTRICTED TECHNICAL INFORMATION.”
ADVANTAGES:
- The driver doesn’t use, reserve, or require any DOS system variables.
- No program knows that when it works with drive
, it’s working with an emulator. - TR-DOS also “doesn’t know” that when it works with drive
, it’s working with an emulator. - All states of all VG93 ports are emulated.
- Absolutely all known :) VG93 commands are executed.
- The RAM disk “spins” at a specific speed.
- Index pulses are present.
- Physical track length is about 6300 bytes.
- When a track is read, everything is read correctly (gaps, marks, headers, etc.), but when the operation finishes, the “data lost” bit is set in the status register (#1F), just as it should be on a real controller.
- Number of “virtual” heads = 2.
- Number of physical tracks = 80.
- The “Maximum Tracks” check in “DCU”, “ADS”, “FUT”, “RDS”, etc. will return 80 tracks.
- Interrupt mode and status have no effect on the driver’s operation.
- The contents of no registers are corrupted.
- The emulator code is located in previously unused areas of the DOS ROM.
- The RAM disk data area occupies banks 24 through 63 of KAY-1024—that’s 640K (2560 TR-DOS sectors).
- The RAM disk operates silently: the heads don’t clatter, the disk doesn’t rustle, and the drive doesn’t spit oil :) :) :)
STRANGELY ENOUGH, ONE DRAWBACK:
- Sometimes (but not always!) when determining whether the RAM disk is active or inactive, the driver uses the stack to store AF once.
DRAWBACKS THAT OCCUR ONLY WHEN THE RAM DISK IS IN USE:
- When the RAM disk is in use, system variables are set up in bank 23 of KAY-1024 RAM (for reference: banks 0–15 = 256K, as in 'SCORPION-256'; the rest is KAY-1024 extended memory). The variables and its own stack are located from address #FE00 and occupy about 400 bytes (variables and its own stack);
- When the RAM disk is in use, 4 bytes of the stack are used (for saving AF,BC once);
- When the RAM disk is in use, SCREEN 0 (#4000) is enabled;
- Formatting the RAM disk as MS-DOS, IS-DOS, CP/M, or God knows what else is possible, but when reading a track, the sector size will be 256 bytes, with 16 sectors per track, etc. (the TR-DOS system's track format). This is due to the RAM disk's 640K memory limit, which does not allow non-standard tracks to be stored in their entirety, including gaps, labels, etc. (in short, there isn't enough room :));
Postscript: “WHERE TO CALL WITH ANY QUESTIONS AND WHERE TO BUY”
Phone number for people in St. Petersburg and other regions:
(812)587-94-09, from 20:00 to 22:00. Ask for the author, Alexey, or contact the (c)Nemo company.
╔══════════════════════════════════════════════════════════════╗
║ Latest IS-DOS Updates ║
╚══════════════════════════════════════════════════════════════╝
September 1999
14.9.99: ch+2.com — Fixed an indication error.
20.9.99: format, doctor, msd_form, trdoctor: Crashes when loading from a hard drive have been fixed (the floppy driver was not initialized).
21.9.99: tv — Fixed: return from <9>.
21.9.99: isdos.sys — All characters permitted in MSDOS are now allowed in filenames: !@#$%&'(){}^~-. Two bytes shorter.
22.9.99: sv_trd: Fixed an error in floppy-driver configuration on exit. The output filename is based on the disk name.
October 1999
22.9.99: svimage: Replaces spaces in the disk name with "" in the filename.
10.10.99: mkdir.com/res — Extensions for directories. (mkdir.res: +23 bytes.)
11.10.99: is_dos.sys — Extensions for directories are now allowed. +38 bytes. (Beta version)
11.10.99: shell.com — 80-character path. (+48 bytes in 2 channels)
11.10.99: rename.com — Extensions for directories.
11.10.99: arzt+.com — Compatibility with system changes from 21.9.99 and 11.10.99 / !@%&'(){}~ and directory extensions / (No indication yet, -21 bytes).
11.10.99: filecopy — 92-character path.
13.10.99: find — Extensions for directories.
13.10.99: is_dos.sys — A small indication bug was caught: an extra space was printed.
14.10.99: new_cats — Not only replaces directory levels from the old format with the new one, but also corrects those shifted after move.
14.10.99: is_dos.sys - Strengthened protection against crashes on
corrupted directories. In theory, it should display the
panel of an open directory at any level, rather than
crashing like iSDOS'96 or complaining about Error 223
like iSDOS'98.
14.10.99: arzt+ - Foolproofing: directory nesting depth up to 32.
Display of directory extensions.
15.10.99: ed1.ovr - A nasty bug was caught, showing up as a "!"
in an empty replacement line (Semyon Semyonych!...
jr z around a POP... What a stunt!)
15.10.99: from_msd - BUG - Block 0 of segmented files wasn't read.
15.10.99: unicolor - BUG - Error 170 when called with a parameter
17.10.99: ed+5.ovr, calculat.com - DEC/HEX/OCT/BIN (+640 bytes)
17.10.99: ch+2.com - Directory extensions
17.10.99: scan.res - Directory extensions (-4 bytes)
17.10.99: netto - Directory extensions
28.10.99: mkdir.res - BUG - The last character in the name wasn't cleared.
28.10.99: coca - Directory extensions
2.11.99: ide_tune - /A../H switches to set the device configuration
in command mode. (For hdd > 96 Mb).
13.11.99: find, ch+2 - Error in current path display.
14.11.99: from_trd - BUG - Drive A: wouldn't open after H:
14.11.99: contig.res - BUG - Extra POP.
14.11.99: ibm_is+, is_ibm+.com/.res - Progress window. Files
larger than 255 blocks are also made contiguous whenever
possible. (+195 bytes).
- Year 2000 -
15.1.00: ibm_is+, is_ibm+.com/.res - Fixed: hanging on
1-block files.
4.3.00: date.com - Y2K BUG: So it turns out the year 2000 is a leap year!
4.3.00: format, doctor - Fixed: bad blocks in file file0.bad
weren't accounted for on tracks 80 through 84.
5.3.00: bad.com - Fixed: got stuck in a loop if bad blocks
were at the end of the disk.
6.3.00: create.com - see bad.com
6.3.00: mkdir.com - Removed the limit on the 6th nesting
level. coca didn't work correctly with it.
7.3.00: trd_ram - Checks for the presence of the file boot on
the TRDOS quasi-disk.
7.3.00: nmi.com - Support for ZS-Chic and Chic for Profi.
╔══════════════════════════════════════════════════════════════╗
║ IDE HDD Statistics ║
╚══════════════════════════════════════════════════════════════╝
Here is a list of hard drives connected by A.G. Leontyev to a
KAY computer using the IDE interface from (c)Nemo
1. Conner CP 3000 40 MB (2 pcs.) OK
12 V * 250 mA + 5 V * 275 mA
2. WDAC 140-07M 42 MB 5/17/97 OK
3. WD 93044-A 4/27/782 "tin can" -
- hangs
4. ? 250 MB - ever busy -
5. IBM D30-A190 WDA-L42 5/17/977 OK
25 mm
- hung on create/bad/abc
- DI in the driver helped
You can order “open letters” from the company
6. Seagate ST351A/X 5/17/980 24.5mm OK?
DI helped
7. QUANTUM Plus Impulse P40A 9 OK
5/17/965 41.5mm bad cyl 0
(wire in the connector)
(we decided that controllers should
consist only of 555-series chips)
8. Seagate ST351A/X 00KC365686 OK
02.01.05
5/17/724 5/17/964 5/17/980
(the controllers are joking)
9. Seagate 00CS253503 03.02.03 OK
10. Caviar 280 10/17/980 85.3 MB OK
11. Seagate Technology ST157A -
"tin can" 2A*12V+1.5A*5V!
"lies" about 6/26/560 (5/17/977)
- hangs while writing
12. NEC GCA2142 D3735 7/17/740 -
1993.06 (with a little wire underneath)
0.8A*12V+0.7A*5V
- hangs while reading (demon), and
almost immediately (sometimes on block 20,
sometimes on block 30).
- On the 256 board, demon ran briskly,
but check hung on block 1280. When writing:
Error 7
13. IBM WDA-L42 D30_A190 5/17/977 -
Slow on RESET
160 kbps linear read ide+3.blk
(bad.com: 34000/67"=127 kbps)
- Hangs on KAY-1024
- On KAY-256 it hangs, but very
rarely.
14. Conner 81 MB (82992 kB) OK
"the hard drive with a nut"
(it was mine for a week)
15. Microsciense International Corp. -
Singapore 1990 5/17/97
8040-00 5V*0.26A+12V*0.35A
- Reads OK
- create and abc hang while writing,
although demon writes OK...
16. QUANTUM ELS42A made in Japan OK
5/17/968 0.23A*5V+0.185A*12V
1992
- Perfect
17. Conner Peripherals 120MB OK
- CP30104H NBQ3SE 2.77
Malaysia 1993 8/39/762
5V*250mA+12V*275mA
(actually 116 MB)
- works like clockwork. Mine
18. Samsung 125 MB 21A (APOLLO) -
SHD-31 16/25/615
- hangs while writing, though not
right away.
19. Seagate ST3144A 124.6 MB OK
00KK011606 REV184.0
15/17/1001
- slow on RESET and, accordingly,
won't boot via Ss/Reset
20. Seagate 120MB ST3144AT V190.00 OK
- slow on RESET
15/17/1001
21. Conner Peripherals 256 MB OK
- Why on earth did I give it away??...
22. QUANTUM LPS170A 1914049712 -
15/22/1011 (162.9 MB)
- reads perfectly, but hangs or
errors while writing, though not
right away. create worked OK, but
mkdir keeps crashing:
Error 86, Error 7.
23. MiniScribe Corp. 8225 XT DS 21919 -
Longmont, Colorado, USA 1988-1989
12V*1A+5V*.7A
- Ports cannot be read
24. Seagate ST351A/X Started smoking -
25. Seagate ST351A/X 00CS900772 OK
03.02.05 5/17/980
Turned out to be 5/17/979 (hit the limit)
26. Seagate st3096AT 00CU065677 -
V190.00 10/17/1024
Ports could be read on the second try
Only block 0 can be read.
Then it hangs
TOTAL:
4. ? 250 MB - always busy -
10. Caviar 280 10/17/980 85.3 MB OK
14. Conner 81 MB (82992 kB) OK
1. Conner CP 3000 40 MB (2 units) OK
17. Conner Peripherals 120MB OK
21. Conner Peripherals 256 MB OK
13. IBM WDA-L42 D30_A190 5/17/977 -
5. IBM WDA-L42 D30_A190 5/17/977 OK
15. Microsciense International Corp. -
23. MiniScribe Corp. 8225 XT DS 21919 -
12. NEC GCA2142 D3735 7/17/740 -
16. QUANTUM ELS42A made in Japan OK
22. QUANTUM LPS170A 1914049712 -
7. QUANTUM Plus Impulse P40A 9 OK
18. Samsung 125 MB 21A (APOLLO) -
9. Seagate 00CS253503 03.02.03 OK
26. Seagate ST3096AT 00CU065677 -
19. Seagate ST3144A 124.6 MB OK
20. Seagate ST3144AT 120MB V190.00 OK
25. Seagate ST351A/X 00CS900772 OK
8. Seagate ST351A/X 00KC365686 OK
6. Seagate ST351A/X 5/17/980 24.5mm OK?
24. Seagate ST351A/X Went up in smoke -
11. Seagate Technology ST157A -
3. WD 93044-A 4/27/782 "tin can" -
2. WDAC 140-07M 42 MB 5/17/97 OK
╔══════════════════════════════════════════════════════════════╗
║ KAY-1024 RAM Memory Structure ║
╚══════════════════════════════════════════════════════════════╝
1997 1992 1987 1982
KAY-1024 Scorpion-256 Amstrad-128 Spectrum-48
------------------------------------------------------
4 ▐----------┐ ▌ ▐ ▐
banks ▐ │ ▌ ▐
▐--------┐ │ ▌ ▐
▐ │ │ ▌
12 ▐ │ │ ▌
banks ▐ │ │ ▌ ┌─────────────────────────────────────┐
▐-------┐│ └───Є│ 64 KB (4 banks) │
8 ▐ ││ │ processor-supported physical memory │
banks ▐ ││ │ address space │
▐-----┐ │└─────Є┤ │
▐ │ │ └─────────────────────────────────────┘
▐ │ │ ┌─────────────────────────────────────┐
▐ │ │ │ 64 KB + 192 KB (16 banks) │
▐ │ │ │ used to store user programs and the │
40 ▐ │ │ │ IS-DOS OS body │
banks ▐ │ └──────Є└─────────────────────────────────────┘
▐ │ ┌─────────────────────────────────────┐
▐ │ │ 96 KB (8 banks) │
▐ │ │ area partially reserved for system │
▐ │ │ shadow software │
▐ └────────Є└─────────────────────────────────────┘
▐ ┌─────────────────────────────────────┐
▐ │ 672 KB │
▐ │ Electronic disk drive "C" area │
| | TR-DOS system, used in IS-DOS |
| | for storing “TR-DOS blanks” |
| | (disk images) |
KAY-1024 expanded memory addressing
+-----+---------+------------------+
| bit | PORT-bit| RAM size, kB |
+-----+---------+-----+-----+------+
| 0 | 7FFD-0 | | | |
| 1 | 7FFD-1 | 128 | | |
| 2 | 7FFD-2 | | 256 | |
| 3 | 1FFD-4 +-----+ | 1024 |
| 4 | 1FD-7 +-----------+ |
| 5 | 7FFD-7 | |
+-----+---------+------------------+
The TRDOS RAM disk starts at the 24th bank (#18 hex), i.e. 7FFD=#00, 1FFD=#90, and occupies the last 640 kB (40 banks) of the KAY-1024 computer’s memory. In iSDOS, the easiest way to access it is through the electronic disk driver, starting from block 512 (#200 hex). The electronic disk in iSDOS on KAY-1024 starts at the 16th bank; the 256K computer’s memory is used last, and it’s best not to use the 128K banks in it at all.
° ° °
Share your thoughts about the article