• Please review our updated Terms and Rules here

8bit Friendly ISA VGA cards

<I'm somewhat confused by your results: You have a VGA (PVGA1A chipset) card installed, connect to a VGA monitor, but you're only able to init and see CGA modes? What happens when you run a program that supports 256-color VGA graphics?>

First, thank you for responding; I have two programs, one is the MSD tool which is the Microsoft Diagnostic Version 3, program that checks everything and gives you a synopsis of what it found. For the video card, the MSD states the video card is a CGA card as seen by IMG_3084 (62872). In addition, I have a program that I believe was written in the early 1990s by WizardWare. The set of programs consists of 12-360 KB floppies to display VGA responses on the screen. For example, Mandelbrot & Julia Set Generator is the first diskette. The response is, when triggered, Out of Range and freezes the machine. These two tests are telling me the card is not VGA or configured as such. On the other hand, if I run Xtree Gold for more than 25 rows; the card will handle that successfully.

IMG_3059.jpgIMG_3084.jpg

A picture of the card would help identify it. Attached is image IMG_3059 ( 62871), which shows the 8-bit VGA card written in the inquiry.

The assembly code does give a response of the lower-res showing "Program terminated normally." No address as was shown in your example, however.

So your saying that neighbor the config or the executable will do anything for me. So what can be done to get VGA or do I already have VGA and don't realize it?
 
I think you fully do have VGA (I can see it on your monitor!), but the motherboard DIP switches and possibly the video card DIP switches are set to CGA and so the system is confused. AT&T ROM BIOS 1.43 is required to set the 6300 to "other video adapter". Is your 6300 showing ROM BIOS 1.43 on startup? If so, checking the AT&T 6300 reference, setting DIPSW-1 SW5 and SW6 to ON might solve your issue.

If the debug test worked, then your WizardWare programs are defective and/or buggy, or they are getting confused with the system reporting it has a CGA card installed. The fact that you can see anything at all on your monitor means you do have a VGA card and it's definitely working.
 
Again thank you for your feedback. Sorry, I have been busy responding to your suggestions. I had already upgraded the AT&T 6300 bios from 1.21 to 1.43 before the install of the video card. In addition, the dip switches were configured before the install of the Paradise PVGA1A card and execution of the MSD program. I have a 5.25-inch floppy that may house the programs: sys and executables for the PVGA!A card. However, I believe the floppy is a 1.2 MB HD type. I have an HD floppy drive, but my quad-core machine doesn't recognize it. The HP WX8600 is too new for that type of floppy drive. Right now my posture is to accept what I have because it works.
 
If you can transfer VGAFIX and run it, it will allow more software to run. Currently you have a functioning VGA card in the system, but the BIOS is setting up the BDA for CGA which is going to make some programs force CGA mode even though VGA exists.
 
I tried the suggested VGAFIX and didn't see any difference initially. I did rerun the VGA programs that I wrote about in an earlier post and while I didn't see a red error of being out of limits the font was smaller near something equal to VGA size; but still out of limits. I figure the 256 KB of RAM on the Paradise may be the root cause of the message. Thanks, for the help here.
 
Almost nothing of what you wrote makes sense. First of all, this is the first time you've said anything about being "out of limits". Your VGA monitor showed an image, so I don't understand what you're talking about. Secondly, 256K is normal for the PVGA1A and is identical to stock VGA so there's nothing wrong with that.
 
Almost nothing of what you wrote makes sense

... I am likewise confused, but... since we're talking about what VGAFIX can or can't do and since "font size" was mentioned, I'll toss out something I noticed that could possibly be a thing for some people that need the program to fix up a bad BDA?

I noticed the other day I was unable to use MODE to set the 43 and 50 line text modes on my Tandy 1000. Long story short, the problem is that MODE relies on ANSI.SYS, and since ANSI was being loaded before VGAFIX could run out of autoexec.bat ANSI.SYS was under the impression I had a CGA card and thought the hardware was incapable of extended text modes. I can't think of much that really needs ANSI.SYS, but if you do have something that falls into that category then you might have something of a pickle on your hands.

So far I've played a little with NNANSI, an ANSI.SYS replacement that can be loaded from the command line, and if loaded with the "d" option to embed itself in the device chain properly "mode 80,50" will indeed work. Kind of thinking of taking a poke at making a version of VGAFIX runnable from config.sys, though, because NNANSI seems like it might not be 100% compatible with IRCjr.
 
Ok, I agree the "out of limits" error message doesn't make any sense to me either. It must be indictive to the program being used. If the error message said out of memory, that would make sense, agreed. I am going to try a picture next as soon as I can find a DOS picture viewer. I do believe I am seeing 132 characters on screen and that is new. So the tool is working.
 
So you are running a mode switching program that came with the Paradise card, and we were talking about VGAFIX which is a small program that tries to force computers to recognize that a VGA card is installed. When troubleshooting computers remotely, it's important to be as specific as possible so that there are no misunderstandings.
 
(NOTE: All pictures referred to in the course of the text posting are seen at the end of the posting.)

Picture number 1 shows the current configuration of my AT&T 6300. The computer houses the basic 6300 architecture along with boards for a Perstor HD controller, WD Paradise VGA video card with 256 KB video RAM, Boca 2MB RAM Card EMS 4, MS Bus mouse card, and I/O card for COMM3 with LPT2.

The following shows results from Checkit's Memory Map as shown in picture 2 highlighting the video RAM, there is a problem with the Paradise VGA video RAM as seems to use only half of the 256 KB video RAM. Checkit's Testing of the Video Memory generates a parity error as shown in picture 3. In addition, Checkit's testing for color produces just B&W (i.e., color is lost) with different grayscale levels as shown in picture 4. In addition, picture 2 shows the addresses C000-C600, 24 kB, as the Video ROM address. Picture 5 shows the unknown ROM, which appears to be nothing identifiable except for the 8 kB ROM covering addresses C600-C800; the Prester HD Controller addresses are C800-CA00 as shown in picture 7. The E000 is the Boca RAM card as shown in picture 6, however, I have Quart Deck QRAM.sys set up for /R:1 /FL=0, which is applying 64 kB to memory from the RAM card, however, I am reading their manual to attain a better understanding of EMS application. The last R in the memory map of picture 2 for the machines 1.43 ROM as shown in picture 8.

Lastly, a summary of the InfoPlus is included below:
InfoPlus Results

Page 1
Infoplus 1.58 Page 1 - Machine & ROM Identification
Generated: Fri, Aug 28, 2020 at 2:59:45 pm
-------------------------------------------------------------------------------
Machine type: (unknown machine - model byte 00)
BIOS source: COPYRIGHT (C) OLIVETTI 1984
BIOS date: 04/03/86A
BIOS extensions:
Segment Size Copyright notice
C000 24K 003056-007COPYRIGHT PARADISE SYSTEMS INC. 1987,1988, ALL RIGHTS
RESERVEDVGA=U
C800 8K COPYRIGHT(c) PERSTOR SYSTEMS INCORPORATED, 1987-88
-------------------------------------------------------------------------------
Infoplus 1.58à Page 2 - CPU Identification
Generated: Fri, Aug 28, 2020 at 3:01:28 pm
-------------------------------------------------------------------------------
CPU: 8086
Interrupts acknowledged immediately after segment register change: no
Coprocessor: 8087
Infinity: projective
Rounding: to nearest or even
Precision: 64 bits
BIOS coprocessor flag set: yes
BIOS Weitek coprocessor flag set: Not Applicable

Infoplus 1.58à Page 3 - RAM Identification
Generated: Fri, Aug 28, 2020 at 3:01:56 pm
-------------------------------------------------------------------------------
Total conventional memory (bytes): 655360 (640K)
Free conventional memory (bytes): 414752 (405K)
Extended memory (from BIOS call): N/A
XMS driver present: no
DPMI driver present: no
Expanded memory:
Interrupt vector: 0B87:1893
Driver: EMMXXXX0 status: available version: 4.0
Page frame segment: E000
Total EMS memory: 2048K available: 2048K
VCPI capable: no
Handle Size Name
------------------------------------------------------------------------------
Infoplus 1.58à Page 4 - Memory Block Listing
Generated: Fri, Aug 28, 2020 at 3:02:23 pm
-------------------------------------------------------------------------------
MCB PSP Parent Size Env Owner Interrupts
0B85 0008 0014 21136 DOS
10AF 10B0 10B0 4704 COMMAND.COM 2E
11D6 10B0 10B0 64 COMMAND.COM
11DB 10B0 10B0 256 O COMMAND.COM
11EC 3A00 3A00 80 COMMAND.COM
11F2 11F3 10B0 16896 COMMAND.COM 08 0B 2F 33
1613 161B 10B0 96 O PARADISE.EXE
161A 161B 10B0 1056 PARADISE.EXE 10
165D 1665 10B0 96 O XTG.EXE
1664 1665 10B0 119136 XTG.EXE 30
337B 1665 10B0 7168 XTG.EXE
353C 1665 10B0 8176 XTG.EXE
373C 1665 10B0 8176 XTG.EXE
393C 1665 10B0 3104 XTG.EXE
39FF 3A00 3A00 2640 COMMAND.COM 22 23 24
3AA5 3A00 3A00 256 O COMMAND.COM
3AB6 3ABE 3A00 96 O INFOPLUS.EXE
3ABD 3ABE 3A00 414752 INFOPLUS.EXE

-------------------------------------------------------------------------------
Infoplus 1.58à Page 5 - Video Identification
Generated: Fri, Aug 28, 2020 at 3:02:57 pm
-------------------------------------------------------------------------------
Active video subsystem: VGA + analog color
Inactive video subsystem: (none)
Initial video mode: 80 x 25 color
Current video mode: 3 (80 x 25 color text)
Current display page: 0
Graphics modes: 3
Video buffer (offset): 0000
Video buffer size (bytes): 4096
Active display port: $3D4 (color)
CRT mode register: $09
Current palette: $30
Colors: o++++++++++++++++o
Current colors: white on blue
Text rows: 25
Text columns: 80
Scan lines/character: 16
Cursor scan lines: 6-7

-------------------------------------------------------------------------------
Infoplus 1.58à Page 6 - Video Information
Generated: Fri, Aug 28, 2020 at 3:03:50 pm
-------------------------------------------------------------------------------
Display adapter: VESA version 1.0
OEM ID: Western Digital Inc Manufacturer: Western Digital Inc
Video modes supported:
Number: 0100 Mode: Color graphics BIOS output support: yes
Screen size: 640x400 Character size: 8x8 Colors: 256
Memory model: Packed-pixel Memory planes: 1 Memory banks: 1
Number: 0102 Mode: Color graphics BIOS output support: yes
Screen size: 800x600 Character size: 8x8 Colors: 16
Memory model: 4-plane Memory planes: 4 Memory banks: 1
Display adapter: VGA
Chipset: Western Digital/Paradise PVGA1A
Video: 8-bit ROM: 8-bit Frequencies are: fixed-sync
Font Address
INT 1FH C000:0C00
INT 43H C000:0800
ROM 8x14 C000:1000
ROM 8x8 (lo) C000:0800
ROM 8x8 (hi) C000:0C00
ROM 9x14 C000:1E00
ROM 8x16 C000:1F2D
ROM 9x16 C000:2F2D
Display type: color
Memory: 256K
Palette registers: 00 01 02 03 04 05 14 07 38 39 3A 3B 3C 3D 3E 3F
Border color: 00 Color page: $00 Paging mode: 4 pages of 64 registers
-------------------------------------------------------------------------------
Infoplus 1.58à Page 7 - Keyboard & Mouse Information
Generated: Fri, Aug 28, 2020 at 3:05:00 pm
-------------------------------------------------------------------------------
Keyboard:
BIOS support for enhanced keyboard: no
Insert: off Caps Lock: off Num Lock: off Scroll Lock: off
Buffer start: 0040:001E end: 0040:003E size (keystrokes): 15
Internal modem [IBM Convertible]/serial printer [PCjr]: no
Game port: no, and NOT set in BIOS equipment word
Mouse: yes Buttons: 2
Microsoft version standard: 8.20
Type: InPort Interrupt: IRQ3
driver type: TSR cursor type: software no. of MDD's: 37
interrupt rate: 30Hz acceleration profile: 2 BallPoint: no
.INI file path: C:\DOS\MOUSE.INI
pointer - style: normal size: small size change: no
EGA register support: yes version: 1.52
Save state buffer size (bytes): 399
Mickeys/pixel (horizontal): 50 (vertical): 50
Double speed threshold: 50
Current display page: 0
Language: English

-------------------------------------------------------------------------------
Infoplus 1.58à Page 8 - Parallel/Serial/Sound Ports
Generated: Fri, Aug 28, 2020 at 3:05:22 pm (Note: Reordered before second i/o
card installed.)
-------------------------------------------------------------------------------
Printers: 1 Serial ports: 1
Device: LPT1 Device: COM1
Base Port: $0378 Base port: $03F8
Timeout: 20 UART: 8250
Busy: yes Timeout: 1
ACK: no Baud rate: 2400
Paper out: no Data bits: 8
Selected: no Parity: none
I/O error: no Stop bits: 1
Timed out: no Break: no
RLSD: no
Sound cards: RI: no
Ad Lib (or compatible): no DSR: no
Sound Blaster: no CTS: no
Roland MPU-401: no dRLSD: no
Gravis Ultrasound: no -dRI: no
dDSR: no
dCTS: no

-------------------------------------------------------------------------------
Infoplus 1.58à Page 9 - DOS Information
Generated: Fri, Aug 28, 2020 at 3:05:46 pm
-------------------------------------------------------------------------------
DOS version: 6.22 revision: A File Control Blocks:
in HMA: no in ROM: no amount: 4
OEM serial number: FF Stacks:
System date: Friday, 8-28-2020 amount: 9 size each (bytes): 128
System time: 3:05:30.15 pm
Command load paragraph: 3ABE ------ International Information -----
Ctrl-C check: off Global code page:
Disk verify: off Active: 437 Default: 437
Switch prefix character: / Country code: 1
\DEV\ prefix for devices: optional Country: United States
Reset boot: preserve memory Thousands separator character: ,
Boot disk was: C: Decimal separator character: .
DOS critical flag: 010F:0320 Data-list separator character: ,
DOS busy flag: 010F:0321 Date format: USA (mm-dd-yy)
Printer echo: off Separator character: -
PrtSc status: ready Time format: 12-hour
Memory allocation: first fit Separator character: :
DOS buffers: 10 Read-ahead: 0 Currency format: $x,xxx.yy
File handle table: 3ABE:0018 Case map call address: 010F:0CF5
length: 40 used: 5
Open file handles:
Table at: 010F:00CC table size (handles): 5
AUX open mode: read/write sharing mode: compatible inherit: no
attributes: (none)
remote: no date: 1-1-1980 time: 12:00:18 am
size (bytes): 0 (0.0K) DOS device
CON open mode: read/write sharing mode: compatible inherit: no
attributes: (none)
remote: no date: 1-1-1980 time: 12:00:18 am
size (bytes): 0 (0.0K) DOS device
PRN open mode: write sharing mode: compatible inherit: no
attributes: (none)
remote: no date: 1-1-1980 time: 12:00:18 am
size (bytes): 0 (0.0K) DOS device
Table at: 0E3C:0000 table size (handles): 35
(none used)

-------------------------------------------------------------------------------
Infoplus 1.58à Page 10 - Multiplex Programs
Generated: Fri, Aug 28, 2020 at 3:06:27 pm
-------------------------------------------------------------------------------
Multiplex interrupt ($2F)
DOS: yes
DRIVER.SYS: yes
DISPLAY.SYS: no; OK to load
ANSI.SYS: no; OK to load
EGA.SYS: no; OK to load
PRINT: no; OK to load
ASSIGN: no; OK to load
SHARE: no; OK to load
NLSFUNC: no; OK to load
GRAFTABL (4.0-): no; OK to load
GRAFTABL (5.0+): no; OK to load
NETBIOS append: no; OK to load
NETBIOS network: no; OK to load
SHELLB: no; OK to load
XMA2EMS: no; OK to load
APPEND: no; OK to load
GRAPHICS.COM: no; OK to load
Crit.err.handlr: no; OK to load
Multiplex interrupt ($2F)
CDROM: no; OK to load
Network: no; OK to load
DOSKEY: no; OK to load
DOS Extender: no; OK to load

-------------------------------------------------------------------------------
Infoplus 1.58à Page 11 - Environment Variables
Generated: Fri, Aug 28, 2020 at 3:07:51 pm
-------------------------------------------------------------------------------
Program name: C:\TESTING\INFOPLUS\INFOPLUS.EXE
Environment:
Segment: 3AA6 Size: 256 Used: 61 Free: 195
Variables:
COMSPEC=C:\COMMAND.COM
PROMPT=$p$g
PATH=C:\;C:\DOS;C:\WP51\

-------------------------------------------------------------------------------
Infoplus 1.58à Page 12 - Device Drivers
Generated: Fri, Aug 28, 2020 at 3:08:13 pm
-------------------------------------------------------------------------------
Device Units Header Attributes Strategy Interrupt
NUL 010F:0048 1000_0000_0000_0100 010F:0DC6 010F:0DCC
EMMXXXX0 0B87:0000 1000_0000_0000_0000 0B87:0012 0B87:001D
CON 0070:0023 1000_0000_0001_0011 0070:06F5 0070:0700
AUX 0070:0035 1000_0000_0000_0000 0070:06F5 0070:0721
PRN 0070:0047 1010_0000_1100_0000 0070:06F5 0070:0705
CLOCK$ 0070:0059 1000_0000_0000_1000 0070:06F5 0070:0739
4 0070:006B 0000_1000_1100_0010 0070:06F5 0070:073E
COM1 0070:007B 1000_0000_0000_0000 0070:06F5 0070:0721
LPT1 0070:008D 1010_0000_1100_0000 0070:06F5 0070:070C
LPT2 0070:009F 1010_0000_1100_0000 0070:06F5 0070:0713
LPT3 0070:00B8 1010_0000_1100_0000 0070:06F5 0070:071A
COM2 0070:00CA 1000_0000_0000_0000 0070:06F5 0070:0727
COM3 0070:00DC 1000_0000_0000_0000 0070:06F5 0070:072D
COM4 0070:00EE 1000_0000_0000_0000 0070:06F5 0070:0733

-------------------------------------------------------------------------------
Infoplus 1.58à Page 13 - DOS Drive Information
Generated: Fri, Aug 28, 2020 at 3:08:55 pm
-------------------------------------------------------------------------------
LASTDRIVE: E:
Logical drives: A: B: C: D:
Diskette drives: 2
Sectors/track: 9
Bytes/sector: 512
On time (ms): 500
Off time (s): 2.0
Head settle time (ms): 15
Current drive and path: C:\TESTING\INFOPLUS
Drive type is: physical
removable: no door lock: no
JOIN'd: no
SUBST'd: no
Volume label: PRIMARY
Media: fixed disk
Clusters: 37993
Sectors/cluster: 4
Bytes/sector: 512
Total space (bytes): 77809664 ( 75986K)
Free space (bytes): 60835840 ( 59410K)

-------------------------------------------------------------------------------
Infoplus 1.58à Page 14 - BIOS Drive Information
Generated: Fri, Aug 28, 2020 at 3:09:22 pm
-------------------------------------------------------------------------------
BIOS disk parameters:
Unit: 128 129 130 131
Type: fixed disk fixed disk fixed disk fixed disk
Drives: 2 2 2 2
Heads: 6 6 15 15
Cylinders: 819 819 917 917
Sectors/track: 31 31 31 31
Specify bytes: (y for more)
Off time (s):
Bytes/sector:
Sectors/track:
Gap length:
Data length:
Gap length for format:
Fill byte for format:
Head settle time (ms):
On time (ms):
Infoplus 1.58à Page 15 - Partition Table Listing
Generated: Fri, Aug 28, 2020 at 3:09:43 pm
-------------------------------------------------------------------------------
Partition table data:
Unit: 128
Partition: 1 2 3 4
Bootable: yes no no no
Starting head: 1 - - -
Starting sector: 1 - - -
Starting cylinder: 0 - - -
System ID: Big DOS-16 - - -
Ending head: 5 - - -
Ending sector: 31 - - -
Ending cylinder: 818 - - -
First partition sector: 31 - - -
Sectors in partition: 152303 - - -

The question here is what happened to the remaining 128 KB of Video RAM? Can any of the video ROM\RAM be moved to the EMS on the BOCA Card using the quarterdeck software? Can the Other items are shown in the Reserved memory be loaded high into the EMS?


picture 1.jpgpicture 2.jpgpicture 3.jpgpicture 4.jpgpicture 5.jpgpicture 6.jpgpicture 7.jpgpicture 8.jpg
 
The question here is what happened to the remaining 128 KB of Video RAM? Can any of the video ROM\RAM be moved to the EMS on the BOCA Card using the quarterdeck software? Can the Other items are shown in the Reserved memory be loaded high into the EMS?

Nothing happened to it. VGA's memory, however much is on the card, is remapped around the 128k space in the memory map between A0000-BFFFF. This is how Check-it displays the memory map for any VGA-equipped machine.

As for the option ROM identification, as long as it identifies "ROM" where ROM is supposed to be, which it looks like it does, don't worry about it. Check-it is terrible at identifying what ROMs are actually for. (For instance, so far as I can tell it *always* IDs a ROM starting at C000 as a "Video BIOS", even if it's, say, the BIOS on an XTIDE card in a CGA machine.) The "shades of gray instead of color" problem is probably monitor ID related. Long story short: Ancient VGA cards like yours used a certain pin on the connector to tell if they were connected to a color or monochrome monitor; grounding it (I believe) makes it think a mono monitor is attached. Later VGA cards/monitors reused this line to do I2C signaling for DDC monitor information. This reuse can cause problems for old video cards, causing them to sporadically misidentify what kind of monitor is attached. My guess is that's what's happening there. Only sure way to stop this is cut the offending pin.

So far as I'm concerned the only red flag, and it's a big one, in your information here is the memory on the VGA card is failing the memory test. That's a problem. If you have another machine to stuff the card into try running the memory test on it there and see if it follows it; if it does, well, you have a bum VGA card.
 
There's some odd stuff going on here, both with the VGA and with your system config:

Boca 2MB RAM Card EMS 4, MS Bus mouse card, and I/O card for COMM3 with LPT2.

It always helps with troubleshooting to remove things you won't ever use. The bus mouse card, COM3, and LPT2 all use different IRQs -- if you'll never have a need for COM3/LPT2 you might as well remove the card.

Total conventional memory (bytes): 655360 (640K)
Free conventional memory (bytes): 414752 (405K)

If that's what you see at DOS running "mem" or "chkdsk" then you've got too much loaded.

165D 1665 10B0 96 O XTG.EXE
1664 1665 10B0 119136 XTG.EXE 30
337B 1665 10B0 7168 XTG.EXE
353C 1665 10B0 8176 XTG.EXE
373C 1665 10B0 8176 XTG.EXE
393C 1665 10B0 3104 XTG.EXE

When running diagnostic programs, you have a clean-booted system. It looks like you ran infoplus from an Xtree Gold session.

I also saw the PARADISE.EXE loaded. That shouldn't be running for most normal VGA operation. Does CheckIt/Infoplus have the same results if you don't run PARADISE at all? Is there a reason you need PARADISE loaded?

It may help for you to post your CONFIG.SYS and AUTOEXEC.BAT just to see if there is something wonky in there.

Active video subsystem: VGA + analog color

This is good! Your system is recognized as running VGA.

The question here is what happened to the remaining 128 KB of Video RAM? Can any of the video ROM\RAM be moved to the EMS on the BOCA Card using the quarterdeck software? Can the Other items are shown in the Reserved memory be loaded high into the EMS?

As explained, this is just testing semantics. There is no issue with your VGA RAM size. Also: Video RAM can not be exchanged with EMS and vice versa. EMS provides a 64K "window" into the full EMS RAM space, VGA does the same thing at a000.

So far as I'm concerned the only red flag, and it's a big one, in your information here is the memory on the VGA card is failing the memory test. That's a problem. If you have another machine to stuff the card into try running the memory test on it there and see if it follows it; if it does, well, you have a bum VGA card.

So, I might have an explanation for the VGA grayscale issue and possibly the memory test issue: The M24/6300/6060 has bus issues where 16-bit port I/O is performed in the wrong order (hi/lo bytes swapped). If CheckIt was trying to set the palette by sending 16-bit words instead of one byte at a time to the palette registers, this might be the result. It's just a guess.

A second VGA card -- one that is known to work in 8-bit slots -- should be tested. Or, the Paradise card should be tested in the same way it generated errors on a second machine. Otherwise you can't be sure if the problem is M24/6300-related or VGA card related.
 
So, I might have an explanation for the VGA grayscale issue and possibly the memory test issue: The M24/6300/6060 has bus issues where 16-bit port I/O is performed in the wrong order (hi/lo bytes swapped). If CheckIt was trying to set the palette by sending 16-bit words instead of one byte at a time to the palette registers, this might be the result. It's just a guess.

It's a little weird/suspicious that the addresses it's saying are failing on the VGA memory test are every other byte; that *could* legit be some problem with the system that it's doing something wrong when it's writing 16 bit quantities to an 8-bit device. But if it's that it must be happening on the motherboard level, the card the OP is using only has an 8-bit bus connector. The only hypothesis I can summon for it, hypothetically, would be if the standard video in an M24 is set up to be fully 16 bit and for some reason the bus-sizing signal for it is still being generated for addresses in the associated range? ...

But if that were the case wouldn't video be trashed all the time, not just the Checkit memory test?
 
Without a 256-color MCGA test, we can't know if there is actually an issue with the memory or not. The errors are reported in graphics-mode range, not in the b800 textmode range.
 
Without a 256-color MCGA test, we can't know if there is actually an issue with the memory or not. The errors are reported in graphics-mode range, not in the b800 textmode range.

Squinting at the photo of the card it looks like the card's memory is four banks of 4464 DRAMs, so if it *is* actually bad memory it could simply be in a section of RAM that's not normally mapped to the text region. But I'd think corruption should show up in any EGA or better graphics mode, since those all use the A000 page. If each bank is a plane then in the 16 color modes it might only affect one color?
 
Re: the problem I mentioned about VGAFIX sometimes not doing the needful if you have a driver like ANSI.SYS that fails to detect you have a VGA card because you can't run it until autoexec.bat, I went ahead and wasted the time to hack together a fix. I took the guts of VGAFIXC and wrapped it inside some code that lets the resulting EXE run from either config.sys or the command line. It *should* just exit and not leave anything resident, but I've just started messing with it. Seems to work, it'll probably melt your computer into slag as I am as of yet pathetically unskilled in 8086 assembly.

(Source code compiles under MASM 6.)

View attachment vgafudge.zip
 
Trixter: I removed the second serial/parallel card as well as the bus mouse as you suggested and the full memory of the Paradise card returned according to check it as it passed the video memory test. Of course, the reversing of the color that AT&T/Olivetti implemented for msb & lsb cannot be corrected for the testing to allow the color to be displayed during the text and graphics tests. I am OK with that not being accomplished because of that reversal as a result of the design employed by Olivetti. Thank you for your guidance.
 
I am pretty skeptical that this bus reversal issue is responsible for the monochrome palette; I would think there would be all sorts of side effects if writes to the VGA registers were getting clobbered. It’s almost certainly the monitor ID thing.
 
Back
Top