Biblioteka/Arhitektura-i-Kompajleri
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
master
Arhitektura-i-Kompajleri/Computer hardware/VGA documents/WHATVGA.DOC
T
Dasko 4ce48dfb1a Dodajem knjige
2026-05-29 00:39:46 +02:00

1015 lines
44 KiB
Plaintext
Raw Permalink Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
WHATVGA version 2.00 29/Sep/95
Copyright 1991-95 Finn Th›gersen
All Rights Reserved
Command line switches:
whatvga [+acumos] [-ncr] [=1024] [/a]
where +chipset forces WHATVGA to proceed as if this chipset was found.
The test for this chip set will be executed in order
to check for memory and revision.
No other chipsets will be tested.
-chipset ensures that the tests for this chipset will NOT be
executed. Useful if a chipset is misidentified
or some earlier test gives problems.
=kbytes Forces the number of Kbytes of video RAM on the
adapter to the specified value.
/a Run automatic test and write results to a file.
The user responds with Y or N to each test and can
enter comments to the individual tests.
On startup you may be asked to select the video system
to test (say if you have a VESA driver loaded).
If the program seems to halt for more than a few seconds
try to press the enter key.
The recommanded way to use the program is to start it
with no command line parameters and play with it a bit.
See that the chipset, memory and DAC are ID'd properly
and that you can display at least some super VGA modes.
Then run the program with /a to do the full test.
For the /a test, the system should be as "clean" as
possible, as TSRs (including mouse and keyboard drivers)
and memory managers can sometimes affect the results.
Try booting from a floppy.
Comments should only be entered when something is wrong
or needs clarification. Ie. there is no need to enter
"Fine" or "All ok" thats what the yes/no queries are for.
When something is wrong try to be as descriptive as
possible (remember, I can't see the screen).
/c Disabled clock detection. Clock detection can lock up
some systems. For instance S3 systems will lock up if
a 0Hz clock is selected.
/d Debug option. display the name of each chipset before it
is tested.
/i Ignore the configuration file (WHATVGA.CFG)
/p PCI test. Disables the test for PCI devices. Note that
WHATVGA has not been tested with with PCI type 2 busses.
Note that if you use the + - or = commandline options the program should
pause, displaying:
Forced to .....
Press a key to continue
If this does NOT happen the option(s) probably were not recognised.
The options can also be set from a configuration file. On startup WHATVGA
checks for the file WHATVGA.CFG in the current directory. The file consists of
the general options at the top and then the chipset specific options in one
section for each chipset started by a [chipset] header.
Most options are of the keyword=value type. The keyword is not case-sensitive.
For Yes/No options a value of YES, Y, ON or 1 (all not case-sensitive) will
be takes as YES, all other values as NO. Leading and trailing spaces are
ignored. The command line switches will win out if there is a conflict.
General Options above the first chipset header:
AutoTest =Yes/No ;Enables the Auto Test feature if set to YES
ClockTest=Yes/No ;Disables Clock testing if set to NO
Debug =Yes/No ;Enables debug output if set to YES
PCItest =Yes/No ;Disables the PCI test if set to NO (same as /p).
Memory =number ;Forces the amount of video memory (in Kbytes)
Ignore =chipset ;Causes WHATVGA to skip the tests for this chipset.
Any number of IGNORE= lines can be added.
Chipset =chipset ;Forces WHATVGA to assume the chipset was found
Sample file:
Debug=Off ;No DEBUG option
AutoTest=0 ;No AutoTest
ClockTest=N ;Disable clock test
Memory=1024 ;Force video Memory to 1Mb
[S3] ;S3 chipset section
-103 ;Disable mode 103h
+4322 732 498 P8 1000 ;Adds mode 4322h, a 732x498 P8 mode with 1000
; bytes per scanline (use 0 bytes/scanline for
; default width)
If you have problems with the identification/startup:
- The screen goes blank or all white...
Try to run: WHATVGA /d
This will display the name of each chipset and waits for
you to press the Enter key BEFORE the chipset is tested.
When you have found the offending chipsets, try to run:
WHATVGA -Chips&Tech (assuming Chips&Tech is the culprit)
Report the problem.
- The chipset is misidentified.
Try to run: WHATVGA -chipset
where chipset is one of the following:
Chips&Tech Chips & Technologies
Paradise Paradise/Western Digital
Video7 Video7
Tseng Tseng Et3000, ET4000, ET4000/W32
Trident Trident 8800/8900/9000...
Everex Everex
ATI ATI VGA Wonder/Graphics Ultra +/Pro
Genoa Genoa 6x00 (not 5x00, 7x00 or 8x00)
Oak Oak 37/67/77/87
Cirrus Cirrus 5x0/6x0
Ahead Ahead A&B
NCR NCR 77c2x
Yamaha Yamaha 6388
S3 S3 86c series
ALG Avance Logic ALG series
MXIC MXIC 860x0
Realtek Realtek
PRIMUS Primus P2000
Cirrus54 Cirrus CL-GD54xx
Cirrus64 Cirrus CL-GD64xx
Weitek Weitek W5x86
COMPAQ Compaq IVGA/AVGA/QVision
UMC UMC 85c408, 85c418
HMC HMC HM86304
WeitekP9 Weitek P9x00 series
XGA XGA compatibles
AGX IIT AGX series
Mach32 ATI Mach32
Mach64 ATI Mach64
Matrox Matrox VGA
ARK ARK Logic
Acer Acer M31x5
Sierra Sierra SC15064 (Falcon/64)
MediaVision Media Vision Pro Graphics
MGA Matrox MGA Accelerator
Alliance Alliance ProMotion
- Memory Size is identified wrong.
Run WHATVGA =2048 (or how many Kbytes you have)
Report the problem.
Whatvga has several options in the main menu:
1 Test standard VGA
Allows you to test the standard VGA video modes.
Draws a test pattern, including a color palette
and writes basic information as text on the display.
This option is not shown for non-VGA devices.
See 2) for the details.
2 Test extended modes
Allows you to test the extended video modes of the adapter.
Draws a test pattern, including a color palette
and writes basic information as text on the display.
On the upper and left edge every hundred pixels is marked.
when you press a key the testpattern is drawn again, this time
in the largest size the available video memory will support
("Virtual Screen"), allowing you to scroll around in the display
using the arrow keys. You should be able to move in the entire
pattern, but not to move so far right or down that the display wraps.
If you can move past the right edge (or can't reach it) then the
scroll function probably needs a factor 2 (divide or multiply)
in the SetVStart function.
When you press the Enter key a textscreen is shown with the
statistics for the mode, this screen also shown if the rutine
to analyse the mode from the register values (AnalyseMode) reaches
the same result. If not the values are shown as "Calculated".
By pressing "d" or "D" you can see a dump of all the VGA (and extended)
registers. The registers can be saved (appended) to the file REGISTER.VGA
by pressing "f" or "F".
3 Search for modes
Sequentially attempts to set modes 14h to 7Fh, each time testing
if the mode was succesfully set. If so the testpattern is drawn
and the mode is analysed.
5 Hardware Cursor test
This item attempts to place a 32x32 hardware Cursor in the
selected mode. The cursor can be moved around on the screen
with the arrow keys.
6 BitBLT test
This item first draws the testpattern, then draws the color palette with
BitBLT operations at the centre of the display, then makes four blockcopies
via BitBLT of the central palette, 90 degrees apart, each slightly
overlapping the original. As the edges of the source is modified during
the copying these changes will (in part) be copied to the new images.
7 Line Draw test
This item attempts to draw lines from the center of the screen
in a "wheel" using different colors.
8 Read/Write bank test
This item copies the left half of the screen to the right half, reversing
the line order, thus copying from "low" to "high" banks using seperate
read/write banks.
0 Exit
Items 5-8 are only present if WHATVGA believes the adapter supports the
function, though you can always try the test by pressing the key.
Special functions for test purposes (may hang your PC or worse):
B Register Bit test.
You are asked for the base register (such as 3C0h, 3C4h, 3CEh or
3d4h), starting index and starting bit number (0-7).
Then you select the mode for the test.
Now the test pattern is drawn in the desired mode, but using the
entire video memory (like the "virtual screen").
The selected bit will be changed for ~0.5 sec, then restored for
another 0.5 sec. This repeats until another bit is selected by
-, + or * or the test is stopped by the Esc key.
- steps one bit back, + one forward - both changing the index if
necessary. * advances to bit 0 of the next index.
This can be used to figure out the function of individual bits, the
Display Start extension bits being the obvious example (and the
intended target), but also clock bits etc.
!!NOTE!! this may hang your PC or expose your video card or monitor
to conditions beoynd their intended operational specifications.
While damage to your monitor or video card is extremely unlikely
(I haven't heard of a confirmed case since the old IBM MDA monitor),
it is theoretically possible and my insurence company refuses to
recognize any responsibility what so ever :-).
Hanging your PC is however a very real possibility.
Note the bit you were testing, reboot (power off may be required)
and restart from the NEXT bit.
D DAC test submenu
Tests various special DAC functions:
2 - Test 24bit (16Mcolor) mode
3 - Test 32bit (16Mcolor RGBa) mode
5 - Test 15bit (32Kcolor) mode
6 - Test 16bit (64Kcolor) mode
Tests the switching between 256color mode and 15/16/24/32bit mode.
Besides the normal 256color test pattern a RGB color scale in the
selected mode is also shown (this will look weird in 256color mode)
Pressing + toggles between the two modes.
Esc or Enter terminates the test.
8 - Test 6/8bit mode
A test pattern is drawn in the selected 256color mode. The palette
is set up to use the full 256 of 16m palette (256 shades of each
basic color). When the DAC is in 6bit mode there are only 64 shades,
thus the color bars will "break". The + key switches between 6bit
and 8bit DAC mode. Esc or Enter terminates the test.
A - DAC Cmd register Analysis
This test examines the DAC command register which is accessed by
reading 3C6h several times. Note that only some DACs have this
register! The PEL register (3C6h) is set to 4 different patterns
(55h, AAh, 5Ah and A5h) and for each pattern 3C6h is read 8 times.
If the DAC has a command register it will show after 4 or 5 reads.
Some DACs have more than one "hidden" register at 3C6h and some DACs
revert to the PEL register after the first read of the command
register, while others allow multiple reads of the command register.
Then the command register is tested for read/writability, first bit
by bit (0 to 7) this is reported as: "Dac Single Bit Mask: EF" where
EF indicates that all but bit 4 is Read/Writable, then all values (0
to 255) are tested and reported as: "DAC mask: AA R/W: BB Change: CC
Set: DD Clear: EE" where AA shows the bits actually tested - FFh =
all (Some DACs have problems with some bits combinations), BB are the
Read/Writable bits, CC the ones that changed but aren't R/W, DD the
bits that are always set and EE the ones that are always clear.
If any patterns causes unexpected results they are shown as: " AA =
BB" where AA is the value written and BB the one read back.
All fields above are in hex.
C - Test Command register
Allows you to test the behaviour of the DAC command register. Apart
from the usual 256color test pattern a number of color scales are
drawn in (from top to bottom) 15, 16, 24 and 32 bit color format.
These will of course look hopeless in 256color mode, but allows you
to detect if you switch to one of these modes. The F1 - F8 keys will
toggle bits 0 - 7 of the command register. The value you are
currently trying to program, is shown at the top.
G - Gamma Test
Allows you to test if your DAC can handle gamma correction in Hi-
and True color modes. You are asked to select a Hi- or Truecolor
mode, the test pattern is drawn and now one of the basic colors
(Red, Green and Blue) will be inverted every time + is pressed.
M - Select base mode
Selects the mode to use for the DAC tests from the available 256
color modes. By default the 640x480 mode is used.
A test pattern is drawn in the selected 256color mode.
The palette is set up to use the full 256 of 16m palette (256
shades of each basic color). When the DAC is in 6bit mode there
are only 64 shades, thus the color bars will "break".
The keys 6 and 8 switches between 6bit and 8bit DAC mode
(if the 8bit DAC mode works :-) ).
Esc or Enter terminater the test.
Currently this only works for the AT&T (20c490/491), Sierra 24bit
(SC15021/25), Trident TDK8001 and Brooktree (Bt477,Bt484/5) DACs.
R Register Read/Write test.
You are asked for the base register (such as 3C0h, 3C4h, 3CEh or 3d4h),
the mask used for the index register. The mask is entered as an 8 character
string where each character controls one bit of the index (bit 0 first) and
can be 'x' (don't care), '0' (always 0), '1' (always 1) or ' ' (don't care
if the bit is Read/Writable in the index register, 0 or 1 if it is always 0
or 1). Any other characters are interpreted as ' '.
Finally you are asked for the mode to use for the test. After the test a
table showing the status of each bit tested is output starting from index 0
bit 0. Each bit is shown as: '.' (Masked), '0' (Always 0), '1' (Always 1),
'A' (Automatically changing), 'C' (Read/Writable, but changed by some other
bit) or 'W' (Read/Writable).
The tests:
ID: Is the Chip Set, DAC and video memory identified correctly?
Xres: Are the modes (standard and extended) displayed correctly?
Display type, Switch settings and BIOS versions can cause
some modes to be unavailable or out of sync.
Scrl: Can you display and scroll in the Virtual Screen ?
You should be able to move in the entire virtual screen,
but not to move so far right or down that the display wraps.
If you can move past the right edge (or can't reach it)
we probably miss a factor of two in the SetVStart function.
Also can you scroll horizontally in 1 pixel steps ?
If the display wraps when the start crosses the middle (or 1/4)
of memory we have a problem with the high bits (the Y variable)
in the SetVStart function or with an enable bit.
Also part of video memory can be inaccessible in modes that
normally only use <half the memory causing the testpattern to
be overwritten.
Anal: Are the modes (particularly the extended ones) analysed
correctly (Ie do the "Calc:" fields match the "List:" ones) ?
Some modes, such as 2 and 4 color and mode 13h are hard to ID
on many adapters.
Cur: Is the HardWare cursor shown correctly and can it be moved
around the display with the arrow keys ?
The cursor should look like a snipers sights with a dot in the
upper left corner.
Blit: Is the central palette drawn correctly ? Are the secondary palettes
copied correctly ?
Line: Are the lines drawn from the middle to fill a rectangle ?
RWbank: Do the two halfs of the screen match ?
Adapters tested:
Mem: ID: Xres: Scrl: Anal: Cur: Blit: Line: RWbank:
Acer:
M3135
M3145
Acumos: See Cirrus 5401 and 5402
AVGA2 (5402) 512 Ok Ok Ok Ok Ok* NA NA NA
Ahead: . . . . . . . . .
V5000 A
V5000 B 1M Ok Ok Ok Ok NA NA NA Ok
Alliance (ProMotion):
3210 1M Ok Ok Ok Ok
Appian: . . . . . . . . .
AGC-98032 2M No
ARK Logic: . . . . . . . . .
ARK1000VL 2M Ok Ok Ok* Ok Ok No No Ok
ARK1000PV
ARK2000PV 2M Ok Ok Ok* Ok Ok No No Ok
ATI: . . . . . . . . .
18800 NA NA NA
18800-1 512 Ok Ok Ok Ok NA NA NA Ok
28800-2 NA NA NA
28800-4 NA NA NA
28800-5 NA NA NA
28800-6 1M Ok Ok Ok Ok NA NA NA Ok
68800-3 (VGA):2M Ok Ok* Ok Ok NA NA NA Ok
(ACC):2M Ok Ok Ok Ok* Ok Ok* Ok* Ok
68800-6
68800-LX
68800-AX
88000GX (VGA):1M* Ok Ok Ok* Ok NA NA NA Ok
(ACC):4M Ok Ok* Ok* Ok Ok Ok* Ok* Ok
Avance Logic: . . . . . . . . .
ALG2101 1M Ok Ok Ok Ok Ok* Ok* Ok* Ok
ALG2201 1M Ok Ok Ok Ok No No No Ok
ALG2228 Ok Ok Ok Ok No No No Ok
ALG2301 2M Ok Ok Ok Ok No No No Ok
Chips&Tech: . . . . . . . . .
82c450 512 Ok Ok Ok Ok NA NA NA NA
82c451 NA
82c452 512 Ok Ok Ok OK Ok NA NA NA
82c453
82c455
82c456
82c457
F65510
F65520
F65530
F65535
F65540
F65545
F64200
F64300 2M Ok Ok Ok Ok Ok NA
Cirrus: . . . . . . . . .
510/520 NA NA NA
610/620 NA NA NA
5401(AVGA1) 256 No NA NA NA
5402(AVGA2) 512 Ok* Ok Ok Ok Ok* NA NA NA
5410 NA NA NA
5420 512 Ok Ok Ok Ok Ok* NA NA NA
5422 NA NA NA
5424 NA NA NA
5426 2M Ok Ok Ok Ok Ok* Ok* NA NA
5428 1M Ok Ok Ok Ok Ok* Ok* NA NA
5429 NA NA
5430 2M Ok Ok* Ok Ok Ok* Ok* NA NA
5434 1M Ok Ok Ok Ok Ok* Ok* NA NA
5452
6205 NA NA NA NA
6215
6225
6235
6410 512 Ok Ok Ok Ok* NA NA NA NA
6412
6420
6440
Compaq: . . . . . . . . .
IVGS
AVGA 512 Ok Ok Ok Ok NA Ok* NA NA
QVGA
QVGA1024
QVGA1280 2M Ok Ok* Ok* Ok* Ok* Ok* NA
Everex:
Genoa: . . . . . . . . .
51/5200
53/5400
6100
62/6300
64/6600
7900
HMC: . . . . . . . . .
HM86304 512 Ok Ok Ok* Ok NA NA NA NA
HM86314
IIT: . . . . . . . . .
AGX-xx (VGA): 1M* Ok Ok Ok Ok NA NA NA NA
AGX10/11
AGX-14 1M Ok Ok Ok* Ok Ok Ok Ok NA
AGX-15 2M Ok Ok Ok* Ok Ok Ok Ok NA
AGX-16 2M Ok Ok Ok* Ok Ok Ok Ok NA
Matrox: . . . . . . . . .
Impres (VGA):
(ACC):
Ultima (VGA): Ok Ok* Ok Ok
(ACC): 2M Ok*
Imp Pl (VGA): Ok Ok* Ok Ok
(ACC): 2M
Media Vision: . . . . . . . . .
PG 1024 2¬M Ok*
PG 1280 4M No
MXIC: . . . . . . . . .
86000 512 Ok Ok Ok Ok NA NA NA NA
86010
86100
NCR: . . . . . . . . .
77C20
77C21
77C22E 4M Ok Ok Ok OK Ok NA NA Ok
77C22E+
77C32BLT 2M Ok Ok Ok Ok Ok Ok NA Ok
Oak: . . . . . . . . .
OTI-37C 256 Ok Ok! Ok OK NA NA NA
OTI-067 256 Ok Ok! Ok OK NA NA NA
OTI-077 1M Ok Ok! Ok Ok NA NA NA Ok
OTI-087 1M Ok Ok! Ok Ok Ok* Ok
Paradise/WD: . . . . . . . . .
PVGA1A 512 Ok Ok Ok Ok NA NA NA NA
WD90C00 NA NA NA
WD90C10 NA NA NA
WD90C11 512 Ok Ok Ok Ok NA NA NA Ok
WD90c2x
WD90C24 1M Ok Ok Ok Ok Ok* Ok Ok
WD90c26
WD90C30 1M Ok Ok Ok Ok NA NA NA Ok
WD90C31 1M Ok Ok OK Ok Ok* Ok* NA Ok
WD90c33 2M Ok Ok Ok Ok Ok Ok* Ok* Ok
Primus: . . . . . . . . .
P2000 2M Ok Ok* Ok Ok NA Ok* NA Ok
Realtek: . . . . . . . . .
RT3105 512 Ok Ok Ok Ok NA NA NA Ok
RT3106 NA NA NA
S3: . . . . . . . . .
86c911 NA
86c911A/24 1M Ok Ok Ok Ok Ok* Ok* Ok* NA
86c801 2M Ok Ok Ok Ok Ok* Ok* Ok* NA
86c805 NA
86c805p 1M Ok Ok Ok Ok* Ok* Ok* Ok* NA
86c928 4M Ok Ok Ok Ok Ok* Ok* Ok* NA
86c864 NA
86c864P 2M Ok Ok Ok Ok Ok* Ok* Ok* NA
86c964 NA
86c964P 2M Ok Ok Ok Ok Ok* Ok* Ok* NA
86c732
86c764
86c764P 1M Ok Ok Ok Ok Ok* Ok* Ok* NA
86c866
86c868
86c968
Sierra:
SC15064
SiS:
SG86c201 1M Ok Ok Ok Ok Ok Ok
Trident: . . . . . . . . .
8800BR NA NA NA NA
8800CS 512 Ok Ok Ok Ok NA NA NA NA
8900B 1M Ok Ok Ok Ok NA NA NA NA
8900C 1M Ok Ok Ok Ok NA NA NA NA
9000 512 Ok Ok Ok Ok NA NA NA NA
9000B NA NA NA NA
9000i 512 Ok Ok* Ok Ok NA NA NA NA
9000C 512 Ok Ok Ok Ok* NA NA NA Ok
8900CL NA NA NA
8900D 1M Ok Ok Ok Ok* NA NA NA Ok
9200CXr 1M Ok Ok Ok Ok* NA NA NA Ok
9400CXi 2M Ok Ok Ok Ok* NA NA NA Ok
9420DGi 1M Ok Ok Ok Ok* NA Ok
9440AGi 1M Ok Ok Ok Ok Ok Ok
9660XGi
Tseng: . . . . . . . . .
ET3000 512 Ok Ok* Ok Ok NA NA NA Ok
ET4000 1M Ok Ok Ok Ok NA NA NA Ok
ET4000W32 1M Ok Ok Ok OK Ok* Ok* NA Ok
ET4000W32i A NA
ET4000W32i B 2M Ok Ok Ok Ok Ok* No NA Ok
ET4000W32p A 2M Ok Ok Ok Ok Ok* Ok* Ok* Ok
ET4000W32p B 2M Ok Ok Ok Ok* Ok* Ok* Ok* Ok
ET4000W32p C
ET4000W32p D 1M Ok Ok Ok Ok Ok* Ok* Ok* Ok
UMC: . . . . . . . . .
UMC85c408 512 Ok Ok Ok Ok NA NA NA Ok
UMC85c418 1M Ok Ok Ok Ok NA NA NA Ok
Video7/Headland: . . . . . . . .
G2-GC205 256 Ok NA
HT-208 512 Ok Ok Ok Ok Ok* NA NA NA
HT-209
HT-216 1M Ok Ok Ok Ok Ok! NA NA NA
Weitek: . . . . . . . .
W5086
W5186 256 Ok Ok* No Ok
W5286 1M Ok Ok* No Ok Ok
P9000 2M No
P9100 (VGA) 2M Ok Ok* No Ok Ok
(ACC) 2M No
Yamaha: . . . . . . . .
6388
NA Not supported
Ok Works
No Does not work
! Doesn't work in mode 13h
* See notes below
Notes:
ARK Logic 1000VL/2000PV. Scroll in 256/High/Truecolor modes is in units of
4bytes. So far I haven't figured out how to use the BitBLT engine.
ATI 18800-1 & 28800-6. Horizontal scroll in PK4 mode (and PL4 for 18800-1)
is in units of 2 pixels. The 28800-6 requires extensive reprogramming of
mode 0Dh (timing problems when full memory is enabled).
I suspect that all combinations of the 28800-4/-5/-6 and BIOS id 4 or 5 can
be found. Examples ?
ATI Mach64 VGA. Scroll in PL4 modes is in units of 2 pixels. There are some
HiColor modes (72h,73h,75h), but they are partialy controlled by the Mach64
engine, so have been disabled for now
ATI Mach32 (68800-3). Blit & Line fails 24bit, but works in 4/8/15/16bits.
Mach32 can not tell bytes/scanline or interlace (write only regs)
The VGA mode 13h wraps at 64K.
ATI Mach64 (88800GX). Scroll is in units of 8 bytes. The clock probing works,
but where is the baseline? BitBLT and Line draw does not work in 24bit modes,
works with 32bpp modes. The 16color (packed 4bit) modes shows a blank screen
Avance Logic ALG2101. Blit (Fill) & Line fails in 64Kcolor modes. The cursor
is double height in interlaced modes
Avance Logic ALG2201/2228/2301. Blit, Line & Cursor fails because the graphics
engine registers are memory mapped (but to where ?).
The test for the ALG2228/2301 is suspect. It works for the two cards I have..
Acumos AVGA2 and Cirrus Logic 5402/5420 (rev 1). The cursor does not work in
the 320x200 modes due to the clock doubling
Cirrus Logic CL-GD5426. The cursor does not work in the 320x200 modes due to
the clock doubling, and also not in the 16Mcolor modes.
BitBlt does not work in the 16Mcolor modes (copy ok, fill fails).
BitBLT in 16c modes ?
Cirrus Logic CL-GD5430,CL-GD5434. The cursor does not work in the 320x200
modes due to the clock doubling, and also not in the 16Mcolor modes.
BitBlt does not work in the 16Mcolor modes (copy ok, fill fails), also not in
16color modes and mode 13h.
Note: Not tested in 32bit (RGBa) mode (too little RAM)/missing modes.
Note: On the 5430 the 16color modes blanked (all white)) the 2nd MB.
Cirrus Logic CL-GD6410. Can't id the clocks.
Compaq AVGA:
BitBLT only works in extended 256color modes.
Compaq QVision 1280. The AVGA modes (0-13h,29h,2Eh) wraps at 1Mb. Some modes
(800x600, 640x480 16Mcolors) are broken - BIOS problem. AVGA Blit only works
in mode 2Eh, fails in all other AVGA modes.
Scroll in packed modes (except mode 13h) is in units of 4 bytes
HMC 86c304. Horizontal scroll in 256c modes is in units of 4 bytes.
It is still uncertain if we can distinguish between the '304 and the '314
IIT (VGA). Memory limited to 512K due to scroll & lack of modes. Banking can
access 1Mb, but scroll wraps at 512K.
IIT-14/15/16. Scroll is in units of 4 bytes.
Only tested w/1024x768 8bit modes due to lack of BIOS modes.
Matrox VGA. Mode 13h is redefined to allow full memory access, but the new
mode has a 62 kHz line sync (I.e. we need a /2 on the dot-clock).
Matrix Ultima. The card is ID'd, but memory, DAC and clock is not checked.
No functionality expect for the ID (I.e. no modes...)
MediaVision PG1024. The ID rutine could generate "false positives" and will
not work if the adapter has not been setup for a base I/O address (538h/...)
Also does not check for 768K/2.25M (256color/16Mcolor versions).
All Paradise: Scroll in extended 256color modes is in units of 4 bytes
WD90c31: Cursor is horizontally compressed in 32K/64Kcolor modes. Blit fails
(copy ok, fill fails) in 32K/64K color modes.
WD90c33: Blit & Line fails in 16Mcolor modes
Paradise/WD90c24: Scroll in extended 256color modes is in units of 4 bytes
The cursor moves too slowly vertically in some Flat Panel modes (the 200/350
/400 line ones), probably due to vertical stretching.
Primus P2000. Interlaced modes have a problem scrolling into the 2nd Mb where
the image from the 1st Mb is superimposed on the "real" image from the 2nd
Mb. Blit fails in 32k/64Kcolor modes (copy is ok, fill fails).
Oak-37C,-67. Extended modes and R/W Banks could not be tested due to the low-
spec cards used. There are several different OAK-37 BIOSes with diff. modes!
Oak-087. The hardware cursor only works in 256color modes.
S3 86c924. Cursor has problems in clock doubled modes (mode $D and $13)
BitBLT & Line does not work in 16color (planar) and standard VGA modes
Scroll in 256&Highcolor modes is in units of 4 bytes
S3 86c801. Cursor has problems in clock doubled modes (mode $D and $13)
BitBLT & Line does not work in 16color (planar), standard VGA and 24bit modes
Scroll in 256&Highcolor modes is in units of 4 bytes
S3 86c805p. Cursor has problems in clock doubled modes (mode $D and $13)
and locks up in the packed 16color mode. BitBLT & Line does not work in
16color (planar), standard VGA and 24bit modes
Scroll in 256&Highcolor modes is in units of 4 bytes
S3 86c928. Cursor has problems in clock doubled modes (mode $D and $13)
Scroll in 256/High/Truecolor modes is in units of 4 or 8 bytes
BitBLT & Line does not work in 16color (planar) and standard VGA modes
S3 Vision864. The cursor is broken in all VGA and 16color (planar) modes
Scroll in 256/High/Truecolor modes is in units of 4 or 8 bytes
BitBLT & Line does not work in 16color (planar) and standard VGA modes
S3 Vision964. The cursor is broken in all modes using the S3 cursor (16color
(planar) and some 256color modes), as if the cursor map is stored wrong.
Scroll in 256/High/Truecolor modes is in units of 4 or 8 bytes
BitBLT & Line does not work in 16color (planar) and standard VGA modes
S3 Trio64 (86c764). Cursor has problems in clock doubled modes (mode $D and
$13). Scroll in 256/High/Truecolor modes is in units of 4 or 8 bytes
BitBLT & Line does not work in 16color (planar) and standard VGA modes
Trident. The newer Tridents must have some extra bits for the horizontal
timing registers. This can be seen in the 320x200 24bit mode where the
Horizontal Total rolls over and thus becomes very small.
The 9000i has an onchip 15/16bit DAC, but apparently no one can get it to work
- not even the BIOS/VESA modes. I expect it requires 16bits per pixel clock.
The 9000C, 8900CL and 8900D can not reliably be distinguished from each other.
The 8900B/8900C has a problem with images crossing 512K in 16color modes.
It is ok if the entire image is either below or above 512K
Trident 9440. The clocks are detected from the internal clock chip, but are
off by as much as 10%.
Tseng ET3000AX. Mode 0Dh wraps at 256K. This mode does not appear to be
reprogrammable (unlike mode 13h and the other 16color modes).
Tseng ET4000/W32.
Cursor is too short horizontally in 24bit modes and 15/16bit modes.
Blit: Fill doesn't work in 24bit modes. Copy only works down left!!
Blit doesn't work in 16c modes
Tseng ET4000/W32i revB. Cursor has a coloumn of junk below the actual cursor
in interlaced modes (Happens with the W32i rev B & W32p rev A).
Also cursor is too short horizontally in 15/16/24bit modes and too large
in mode 13h (pixel doubling).
Blit is broken in all modes (Fill appears to work in 8/15/16bit modes, copy
fails in all modes - destination block is often "broken up" or "dirty").
Tseng ET4000/W32p revA. Cursor has a coloumn of junk below the actual cursor
in interlaced modes (Happens with the W32i rev B & W32p rev A).
The BitBlt does not copy "backwards". Lines appears "dotted".
Cursor is too short horizontally in 24bit modes
Tseng ET4000/W32p revB. Blit: Fill doesn't work in 24bit. Blit & lines fails
totally in 16color modes. Lines appears dotted.
Cursor hor. short in 15/16/24 bit modes, too wide in mode 13h.
Some problems w/Analysis due to the Chrontel DAC
Horizontal Scroll in "16bit DAC video path" mode is in units of 2 (or 4?) bytes
Tseng ET4000/W32p revD. Blit: Fill doesn't work in 24bit. Blit & lines fails
totally in 16color modes. Lines appears dotted.
Cursor hor. short in 24 bit modes, too wide in mode 13h.
Horizontal Scroll in "16bit DAC video path" mode is in units of 2 bytes
UMC UM85c408,UM85c418. Some of the modes are weird (BIOS problem).
Video7 HT-208 (actually G2 208). Hardware cursor is horizontally compressed in
256color modes and blank in mode 13h (the last is also true for the HT-216).
Also, I have my doubts as to the clock selection on the '208.
Weitek W5x86,P9100(VGA): Scroll only works in the first 256K (as std VGA).
Horizontal Scroll in ext 256color modes is in units of 4 bytes
DACs tested:
ID 8bit 15b 16b 24b 32b Cursor MPlex Clock
Acumos:
ADAC1 NA NA NA NA
AT&T: . . . . . . . . .
20c490 Ok Ok Ok Ok Ok1 NA NA NA NA
20c491 Ok Ok Ok Ok Ok1 NA NA NA NA
20c492 Ok NA Ok Ok OK1 NA NA NA NA
20c493 NA NA NA NA NA
20c498 NA
21c498 NA
22c498 NA
20c504
20c505
ATI: . . . . . . . . .
ATI68830 NA
ATI68860 Ok Ok Ok Ok Ok NA Ok
ATI68875 pin NA
ATI68880
Avance Logic: . . . . . . . . .
ALG1101 Ok* NA NA pin NA NA NA NA NA
ALG1201 Ok Ok Ok Ok1 NA NA NA
ALG1301
Brooktree: . . . . . . . . .
Bt477 Ok Ok NA NA NA NA NA NA NA
Bt481 Ok Ok Ok Ok Ok1 NA NA NA NA
Bt482 NA
Bt484
Bt485 Ok Ok Ok* Ok* NA Ok* Ok No NA
Chrontel: . . . . . . . . .
CH8391 Ok Ok Ok Ok Ok1 NA NA NA Ok
CH8398 Ok NA Ok Ok Ok1 NA NA No Ok
Chips&Tech:
64300 Internal OK Ok Ok Ok1 NA NA Ok
Cirrus Logic: . . . . . . . . .
CL-GD5200
542x internal Ok NA Ok Ok Ok NA NA NA Ok
543x internal
Diamond: . . . . . . . . .
SS24 Ok NA Ok Ok Ok1 NA NA NA NA
IBM:
RGB514
RGB524 Ok
RGB525
RGB526
RGB528
RGB530
RGB561
RGB624
ICS: . . . . . . . . .
5300
5301 Ok NA Ok Ok OK1 NA NA NA NA
5342
IC Works:
w30c498
w30c516 Ok Ok Ok Ok Ok NA Ok NA
Paradise/WD:
WD90c24 Intern Ok Ok* Ok* NA NA NA
Oak: . . . . . . . . .
OTI-66
OTI-66HC Ok NA Ok Ok NA NA NA
MUSIC: . . . . . . . . .
MU9c1710
MU9c1880
MU9c4870
MU9c4910 Ok NA Ok Ok Ok1 NA NA NA NA
MU9c9910 Ok NA Ok Ok Ok1 NA NA NA Ok
S3: . . . . . . . . .
86c708 (GenDAC)
86c716 (SDAC) Ok NA Ok Ok Ok* NA Ok
SGS-Thompson: . . . . . . . . .
STG1700 NA
STG1702 Ok Ok Ok Ok Ok1 NA NA NA
STG1703 Ok Ok NA
Sierra: . . . . . . . . .
SC11481 pin NA NA NA NA NA
SC11486 NA NA NA NA NA NA
SC11488 pin NA NA NA NA NA
SC11482 pin NA NA NA NA NA
SC11483 NA NA NA NA NA NA
SC11484 pin NA NA NA NA NA
SC11485 pin NA NA NA NA
SC11487 Ok NA Ok Ok NA NA NA NA NA
SC11489 pin NA NA NA NA
SC15021 Ok Ok Ok Ok Ok1 NA NA
SC15025 Ok Ok Ok Ok Ok1 NA NA NA
TI: . . . . . . . . .
TLC34058
TLC34075 Ok pin NA NA No NA NA No NA
TVP3010
TVP3020
TVP3025 Ok Ok Ok Ok No No Ok No Ok
TVP3026
Trident: . . . . . . . . .
TKD8001 Ok Ok Ok Ok Ok1 NA NA NA NA
9000i Internal Ok NA No No NA NA NA NA NA
9200CXr Intern Ok Ok Ok Ok Ok1 NA NA NA NA
9400CXi Intern Ok NA Ok Ok Ok1 NA NA NA NA
9420DGi Intern Ok NA Ok Ok Ok1 NA NA NA
9440AGi Intern Ok NA NA
UMC: . . . . . . . . .
UM70c178 Ok Ok Ok NA NA NA NA NA
UM70c188 Ok Ok Ok Ok1 NA NA NA NA
Winbond: . . . . . . . . .
W82c490 Ok* NA Ok Ok Ok1 NA NA NA NA
NA Not supported
pin Supported by a pin
Comments:
Ok1 - The DAC is switched into a 24bit mode, but the colors are swapped
around. This may be a problem with the dac mode (BGR rather than RGB)
or with the DAC starting from the wrong byte because the left overscan
area is also seen by the DAC
The Trident TKD8001 can be sent into a "no return" state, which I (and the
BIOS) can't return it from....
The Avance Logic ALG1101 id "ID'd" by assuming that any ALG chips that reports
a standard DAC must have an ALG1101... It is unclear how 16bit mode is set.
The Brooktree Bt484/485 can be switched between 8/15/16/32 bit modes, however
due to the different pixel multiplexing used for these modes the display
timings are often seriously disturbed. Also the DAC must be in "VRAM" mode.
The S3 SDAC can be switched between 8/15/16/32 bit modes, however due to the
different pixel multiplexing used for these modes the display timings are
often seriously disturbed.
Paradise/WD90c24 internal DAC can be switched to 15/16bit mode, but double
clocking mode (16bits/VCLK) must be set for it to work.
The Winbond W82C490 is ID'd as "AT&T 20c492", which is close enough
The ATI 68860 is ID'd from the Mach64 status regs. Switch to 15/16/24/32
bit mode works, but changes the horizontal resolution.
Clock chips:
Id: Rd:
DCS2824 Ok NA
ICD20c61 Ok NA
ICS2595
SC11412
The DCS2824 and ICD20c61 are both identified as "ICD20c61" which is ok as the
DCS2824 is an OEM'd ICS chip. The ICD20c61 and ICD20c61A arte however id'd.
Generel Problems:
Search for modes doesn't understand 2 and 4 color modes.
In general 2 and 4 color modes are only supported partially, and
fixing it is NOT a top priority :-).
Mode 0Fh and 11h (and other 2 color modes) has problems in some cases.
Mode 13h has a pixel double function. Most chips use Master Clock/2
(3C4h index 1 bit 3) for this, but some chips has their own system.
This causes problems with mode analysis.
Some of the extended modes may only be implemented on
some otherwise similar boards, depending on make and BIOS rev.
Utilities:
DMPMEM:
DMPMEM dumps the ROMs to files named DMPxxxx.ROM (where xxxx is
the hex address of the ROM Ie. DMPC000.ROM for VGA).
Interrupt and Font addresses are saved to the file VECTORS.ROM
As Mouse drivers, national keyboard drivers, memory managers
etc can obscure the results the program should be run on a clean
system (empty CONFIG.SYS and AUTOEXEC.BAT say from a floppy).
TABLE:
Takes the register sets dumped to the file REGISTER.VGA and writes
them to REGISTER.TBL. If a register has the same value in all register
sets, the value is only shown the first time.
As the same file names are used every time you will have to delete
and rename the files as needed.
VESALIE:
Can change the version and amount of memory reported by your VESA driver.
Can also hide the VESA driver, so other applications don't see it.
Parameters: /0 /1 /2 Makes the VESA driver report a specific VESA
VBE version (1.00, 1.01 or 1.02).
/on /off Turns the VESA VBE on or off. By turning the
VBE off you can prevent applications from using
the VESA interface. Also if you try to load
another VESA driver, it may refuse to install
while the original driver is visible.
/Mxx Forces the VESA driver to report the specified
amount (in Kbytes) of video memory.
/U Unload the VESALIE program.
Note that this NOT a VESA driver, it just changes the behavior of
an already installed driver.
SHOWTEST:
Analyses the results in the .tst files generated when WHATVGA is
run with the /a option (WHVGAxxx.TST).
Parameters: SHOWTEST [tstfil] [outfil] [/r]
When run without parameters it lists the test files in the current
directory, showing the filename, chipset, memory, DAC and chipname.
When run with the name of a test file, the test results in this file
are written to the outfile. If no outfile is specified, the name of
the test file is used, with the extension changed to .txt
If the /r switch is used the register dumps for each mode is written
to the outfile (or to .reg).
To do:
- S3 968/IBM RGB524 combo needs a lot of work
- Damn Weitek (W5x86) scroll function
- ET4000/W32 BitBLT/Linedraw
- Attemp W5x86/Trident/ARK BitBLT/Line draw
- Further work on CT64300, AGX-016, P9000, Matrox, PG1024/1280, Appian,
Alliance, SuperMAC and others
- Figure out how to deal with the AGX-16 which has TWO DACs!!
Reference in New Issue View Git Blame Copy Permalink