@8089.ADF - ATI GRAPHICS ULTRA MCA Video Acc., Ver 2.2, VGA ENABLED
mach32 Drivers and Utilities (ATI Graphics Ultra Pro)
M32UTL.ZIP
Installation and configuration utilities (INSTALL.EXE v2.3) for mach32
products - EISA and MCA config files, DPMS, VVESA, VDIF, VGRAF and DOS
mouse drivers included.
Identifying ATI Products by Direct (Visual) Inspection
Mach32
Setup Tips (ATI Site) Broke
TLC34076 Video Interface Palette Data Manual Datasheet Release Notes for XFree86[tm] 3.3.3.1
ATI GUP Board ATI Graphics Ultra Pro
RAMDAC Older boards use an ATI 68875 CFN, newer ones use a TI TLC 34076-135FN. The TI chip responds as an ATI 68875 under the install program and other diagnostic applications BIOS ST M27C256B-12 28 pin DIP
112-19201-106 Odd Jumpers JU5 ROM
Enable/Disable Jumper JU6 DAC Select Jumper
JU8
Compatibility/Performance Jumper JU9 IRQ
Jumper JU10 VGA
Disable Jumper ATI18811-1 Clock Chip >I'm having problems getting X running on my new machine running linux. I have an ATI Ultra XLR 2MB Video card, and a Gateway 1572FS monitor. Robert Seward wrote: ATI 18811-1: The monitor times given for the Gateway 1572FS
did not work quite right for me - the video was not
centered. I modified it slightly, but I don't have
the numbers with me. ATI GUP Alone in a 95 Take this with a grain of salt. The Farmer City Madman did it without the nutroll. 12 Nov 2005 I have replicated the GUPification process. FIRST and MOST IMPORTANT - get the @8089.ADF from M32UTL.ZIP
or from Peter's site (above) extract all files
to a floppy. You now should have a lone GUP. Single Card Settings
Single GUP Under W95 :*** The GUP uses IO of 2E0-2E7. Non-configurable. How can I set the IO ranges used by the GUP? Addresses used under W95
Trivia Hi Jim, ROM BIOS Address If you are
using extended memory managers such as QEMM, 386MAX or EMM386,
be sure to use an EXCLUDE statement to protect the mach32 ROM
Address space. (ex. X=C800-C9FF) COM4 Conflict The
GUP uses I/O port address from 2E0-2EF. This will conflict
with systems using COM port 4. You must disable COM port 4
in order to use the GUP. Network Cards Many
NICs have a default I/O port base address of 2E8. This
conflicts with the GUP. The NIC must be reconfigured, since
the address of the GUP is not configurable. Memory Aperture If your system is fully populated with system memory (My interpretation is the 640K conventional plus 384 upper), the memory aperture feature must be disabled. (So with the 90/95 machines, I think Memory Aperture disabled would be a good bet. Mine's disabled) What is this Terminator Plug?
Some personal experiences with
my GUP- INSTALL Menu You must run the ATI setup utility called INSTALL.EXE to initialize the GUP. This utility is DOS based, so you may have to boot with a DOS floppy with the utility on it. Set the card to VGA. Use the Windoze or OS/2 drivers to change the configuration later. The Main Selection Menu has six sub-menus. Set
Power-Up Configuration
Custom
Monitor Configuration
Power-Up Video Mode Memory Aperture VGA Memory Size
Mach32 Memory Aperture Configuration Issues ATI Tech Note #067 Introduction Description The main reason for a direct memory aperture feature is that it allows an application to write to the video display memory directly, bypassing the overhead of the standard "paging" system required by VGA and SVGA cards. The major benefits of a memory aperture are apparent when dealing with graphics which are not "vector-based", such as continuous tone bitmapped images. In an aperture-disabled configuration, these graphics, which cannot be easily translated into vector-based commands for the mach32 accelerator, must be passed through the standard "video memory window" in the A000-AFFF region, just above 640k. Since the standard VGA window is only 64k in size, moving a large image onto the screen can involve several operations. With a memory aperture available, a single operation can write the image from system RAM into video RAM. On EISA, 32-bit MCA and VLB architectures, there is an additional performance advantage as compared to ISA. These busses support 32-bit data transfers, and generally operate at higher speeds than the 16-bit ISA bus. Thus, while the main function of the mach32 chipset - offloading screen vector processing from the CPU - is unaffected by the hardware platform or the state of the memory aperture, the processing of bitmapped images will be more rapid on machines in which the memory aperture feature is available, and even faster on a 32-bit high speed bus. The "memory aperture" feature of our ISA mach32 products can only be enabled in systems having less than 12MB of RAM installed. This is due to the 16MB address limitation on the ISA bus. Our memory aperture can be enabled only within the addressable space on the bus. If this space is occupied by system RAM it is impossible to have the "aperture" share this area. There is a feature in the INSTALL program which checks to see how much memory is in the system. If 16MB are detected in an ISA system, INSTALL will display a warning. Unfortunately, some versions of INSTALL will display the warning message whether you've attempted to enable the "memory aperture" feature or not. If your "power up configuration" shows the memory aperture to be disabled, you can ignore any warning messages. If you have an ISA system with less than 12MB of RAM installed and the Memory Aperture feature will not enable, there is likely a problem either with the memory manager, or with the motherboard DIP switches or jumpers. Disable any memory management software (EMM386, QEMM, 386MAX, etc) and try again. If this doesn't help, check the CMOS setup to be sure that the correct amount of system RAM is being reported. Make sure that the system CMOS is not set to "cache" the memory area in which the mach32 memory aperture will be installed. Some CMOS setups allow you to define a "cachable memory size" option. Make sure that this is set to the actual amount of system RAM installed. Other setup routines will allow you to define a "non cachable block". If this is how your CMOS operates, be sure to define the memory aperture location as a "non cachable block". It may be necessary to adjust the motherboard DIP switches or jumpers to properly reflect the amount of system RAM installed. Check your motherboard manual for details. On an ISA system, the default aperture state is "disabled". The default positioning of the aperture, once enabled, is at the 10MB location. We have found that 12MB is more appropriate for systems in which a "non-cachable block" must be defined in the CMOS. The aperture size will be 4MB for a 2MB mach32, and 1MB for a 1MB mach32. INSTALL versions 2.1 and later will automatically place the memory aperture at an appropriate location if you choose to enable it. ATI mach32 products designed for other bus specifications (EISA, 32-bit MicroChannel, and VESA Local Bus) fall under the same sort of limitations as the ISA models, but the maximum memory address for the aperture is much higher (128MB), so the possibility of a conflict with installed system RAM is slight. The recommended memory aperture location for the EISA and VLB mach32 products is at the default 124MB setting. INSTALL version 2.1 or later will automatically place the aperture in this location if you elect to enable it. Very early versions of the mach32 Windows driver
would not operate if the memory aperture was
disabled. This problem has been resolved in more
recent versions of the driver. Ensure that you're
using a driver version of 2.0 or later to take
full advantage of the memory aperture feature. Mach32 Only Gets 640@256 or 800@16 in Win9x When a mach32 product is installed in Windows 95 or Windows 98, only the following modes are available:
This problem occurs because the mach32 board is configured for "factory defaults". The Microsoft-supplied driver for the mach32 reads information regarding mode support directly from the card. If it reads a "factory default" configuration, only the modes shown above are made available. The mach32 product must be configured to match the capabilities of the MONITOR attached, before WIN95 will allow full access to resolution and color depth combinations.. This is accomplished using the DOS-based INSTALL program which comes with the mach32 product (mach32 disk #1). The INSTALL program can also be downloaded from the ATI website or BBS. Look for M32UTL.ZIP. Instructions:
If you have any questions about the information above, contact ATI Technical Support and reference INFO-BASE file # 2981 GUP Under Linux Brad Parker says: I've never been able to get the ATi GUP to run using the Mach32 xserver, however here is the why and how of getting it to display using the generic SVGA xserver. Why: The ATi GUP is a Vesa compliant video card. It also has a firmware configuration utility available. I drive a fixed frequency IBM 6091-19 with mine. As the monitor only works at three specific frequencies, I used the DOS firmware configuration utility to program in my specific needs to the onboard eprom on the GUP. How: There are two ways to do this. And there is the Easy way, which is an
interactive adjustment program. supplied with the ATi
driver diskettes, available on the ATi web site. This
DOS program works a lot like the size and sync
adjustments on a multisync monitor. It reminds me of the
AppleVision software for the Mac. As I knew from the IBM
specs what the 6091-19 needed, I just picked the closest
preset, which happened to be the Vesa 1280 x 1024 @ 60
Hz mode. This is known as Vesa mode 107h (See Peter
Wendt's excellent page on the XGA-2 card for a list of
other Vesa modes). I then entered the custom mode. Next
I displayed the test pattern with the ATi setup program
and used the arrow keys and what not to get the size,
centering, sync polarity etc.. correct for my monitor.
Then I saved those settings, which get stored on the
GUP's eprom chip under, you guessed it, Vesa Mode
107h (I gave it a user friendly name, but 107h is where
it really is). When setting up your xserver for Linux,
choose the generic SVGA server. It's smart enough to
figure out that the ATi GUP is a Vesa compliant card.
You edit the XF86Config file leaving all the entries for
the display settings that correspond to the ones you
have programmed in the ATi card using the above
procedure. Don't worry about the dotclocks and such,
just use the basic resolutions, color depth and vertical
refresh rates that are closest to your monitors modes.
Delete all of the others that aren't supported by your
monitor or the GUP. Now when you start the xserver, it is
simply going to send a Vesa mode change to your Vesa
compliant ATi GUP card. The card will switch to your
custom mode that corresponds to the Vesa mode, like 107h
in my particular case. I'm fairly certain that it just
ignores the dot clock settings and all. Now when you toggle through the various modes while running X, the generic SVGA xserver just sends out various Vesa mode changes. See the SVGA howto for details on which modes correspond to which resolutions. I've also used the ATi GUP in this manner with a Compaq SVGA and IBM 9524 (14P) and 9525 (15P) monitors. It was significantly easier with the multisyncing 14P and 15P than with the fixed frequency 6091-19. Basically with the multisyncs I just used whatever was the default Vesa modes and adjusted the monitors to work with that. Hope this sheds some light on the subject. ASIC Revision > I was unable to install NT with linear video memory access enabled. From Brad Parker "there is a timing problem in The Revision 3 MACH 32 CHIP ITSELF!! Its a hardware problem. No amount of messing with the software will fix it. check the revision of the Mach32 chip. if the rev number ends in 3 ie: 6880003 replace it with a revision 6 chip and all should be ok." (Ed. uh, it's not socketed, so how to easily swap in a new ASIC is beyond me) (Ed. OR
Control Panel>Display>Display
Properties>Display Type, look in the lower left hand
at Adapter Information, Chip Type, and it will say
Mach32 rev. x). AdapterId 8089 ATI GRAPHICS ULTRA MCA Video Accelerator (VGA ENABLED) 10/02/94 v2.2 FixedResources (IO) Address of BIOS ROM A duplicate choice is required to
fix configuration problem on Model 76 |