CaptainStarMac
Neuling
Thread Starter
- Mitglied seit
- 19.06.2001
- Beiträge
- 59
Es gibt ein neues Bios für Hercules 3D Prophet (Ur)MX- Karten !
Habs mir gleich mal gezogen aber noch nicht geflasht...
Hat das schon wer getestet ??
Hier die offiziellen Änderungen laut Nvidias BIOS Release Notes seit der Vorgänger Version:
03.11.00.12.00
Allows an analog monitor connected to an external DAC to operate
through the DFP/DVI interface. (This modification runs the BIOS in
1024x768 fixed timing mode and scales all displays to that resolution.)
Fix an issue where an adjustment to the horizontal timing to correct a
hardware issue caused the maximum resolution available on a fixed-
resolution display to be one resolution smaller than the display could
actually drive.
Fix bug where default TV format is not set up on fast-boot BIOS
execution, unless booting to a TV device.
Correct problem in EDID read routines for monitors that respond to
EDID2 and data reads, but return only 0FFh (256 0FFh - 00h checksum).
Add validation of header of read EDID before returning successful buffer
read in DDC routines.
Correct error in handling of Compaq FP500 flat panels, which caused
invalid data to be passed to panel parsing routines.
Add ability to set PBUS_DEBUG_CTRIM_7 on a per-memcfg strapping
basis (required for P44 DDR).
Fix jittery text when switching output devices from flat panel to TV.
Update save/restore function to save additional NV11 registers.
Remove check for digital monitor in ParseEDID1 and ParseEDID2.
Add code to save SI in Program_DTD.
Fix blank CRT when switching from flat panel to CRT when panel
resolution > 1024x768.
Add init of CR4A-4C.
Fix Chrontel selecting wrong TV standard.
Fix Chrontel TV detect broken.
Fix Philips selecting wrong TV standard.
03.11.00.13.00
Correct error in EDID2 processing that caused an analog monitor on the
external DAC to be detected as a digital FP, or a digital FP to be detected
as an analog monitor.
03.11.00.14.00
Fix bad reference in Geforce2 MX BIOS script that corrupted the memory
init tables.
Remove G2MX-specific register values from generic NV11.SCP file.
Create G2MX-specific script for build process.
Update build to use correct scripts.
03.11.00.15.00
Fix an EDID 1.0 parsing problem.
3.11.00.16.00 Modify ProgramI2C to return the ACK/NACK in the carry flag.
Modify OEMInitTV to exit if carry flag set during I2C transfer.
Modify Philips encoder_init routine to exit if carry flag is set during
I2C transfer.
Add changes to allow compatibility with the Conexant CX871 encoder.
Update fixed-flat-panel structure to include additional data:
Update fixed-flat-panel tables from 4 to 16.
Update BMP to include pointer to new panel structures.
Update BMP to version 5.11 to allow Resource Manager to identify
BIOS with fixed tables.
Update scripts to find BMP versions of 5.11.
Fix data definition for Panel Table 0 (SXGA+).
Make structure of BMP 0x05.0x11 consistent.
Update the BMP scripts for all the chips.
Modify NVResizeMemory to correctly determine the RAM width and
size. (Required for DDR memories because the memory controller
requires a signal back from the memory module to complete the memory
transfer. If no memory is present then the memory controller returns the
data in the FIFO.)
Add code to verify the last MB of memory is correctly sized.
3.11.00.17.00
Beep and abort on an invalid devinit opcode of 0x00.
Save/restore fixes.
Force 24-bit interface if 12-bit isn't strapped. (Corrects issues caused by
assuming 24-bit as the default power-up state.)
Clear TEST_CONTROL bit for PWRDWN_DAC to allow CRT detection to
work correctly. (The BIOS does not reset this bit if the detection code is
called from within an SMI and the driver is running.)
3.11.00.18.00
Clear TEST_CONTROL bit for PWRDWN_DAC to allow monitor detection to
work correctly.
Fix AND/OR.
DWORD align PCI expansion ROM data structure per PCI 2.2 spec, section 6.3.1.2.
Update HandlerNoSTI to return via a FAR RET.
Disable interrupts in FP_DDC_Init and change call to HandlerNear()
to HandlerNearEntry() because the former executes a STI. This fixes a
problem with J-DOS.
Change FPEndSetMode() to use the value 32h for the shadow lock when
an FP is installed.
[ 16. Juli 2001: Beitrag editiert von: CaptainStarMac ]
Habs mir gleich mal gezogen aber noch nicht geflasht...
Hat das schon wer getestet ??
Hier die offiziellen Änderungen laut Nvidias BIOS Release Notes seit der Vorgänger Version:
03.11.00.12.00
Allows an analog monitor connected to an external DAC to operate
through the DFP/DVI interface. (This modification runs the BIOS in
1024x768 fixed timing mode and scales all displays to that resolution.)
Fix an issue where an adjustment to the horizontal timing to correct a
hardware issue caused the maximum resolution available on a fixed-
resolution display to be one resolution smaller than the display could
actually drive.
Fix bug where default TV format is not set up on fast-boot BIOS
execution, unless booting to a TV device.
Correct problem in EDID read routines for monitors that respond to
EDID2 and data reads, but return only 0FFh (256 0FFh - 00h checksum).
Add validation of header of read EDID before returning successful buffer
read in DDC routines.
Correct error in handling of Compaq FP500 flat panels, which caused
invalid data to be passed to panel parsing routines.
Add ability to set PBUS_DEBUG_CTRIM_7 on a per-memcfg strapping
basis (required for P44 DDR).
Fix jittery text when switching output devices from flat panel to TV.
Update save/restore function to save additional NV11 registers.
Remove check for digital monitor in ParseEDID1 and ParseEDID2.
Add code to save SI in Program_DTD.
Fix blank CRT when switching from flat panel to CRT when panel
resolution > 1024x768.
Add init of CR4A-4C.
Fix Chrontel selecting wrong TV standard.
Fix Chrontel TV detect broken.
Fix Philips selecting wrong TV standard.
03.11.00.13.00
Correct error in EDID2 processing that caused an analog monitor on the
external DAC to be detected as a digital FP, or a digital FP to be detected
as an analog monitor.
03.11.00.14.00
Fix bad reference in Geforce2 MX BIOS script that corrupted the memory
init tables.
Remove G2MX-specific register values from generic NV11.SCP file.
Create G2MX-specific script for build process.
Update build to use correct scripts.
03.11.00.15.00
Fix an EDID 1.0 parsing problem.
3.11.00.16.00 Modify ProgramI2C to return the ACK/NACK in the carry flag.
Modify OEMInitTV to exit if carry flag set during I2C transfer.
Modify Philips encoder_init routine to exit if carry flag is set during
I2C transfer.
Add changes to allow compatibility with the Conexant CX871 encoder.
Update fixed-flat-panel structure to include additional data:
Update fixed-flat-panel tables from 4 to 16.
Update BMP to include pointer to new panel structures.
Update BMP to version 5.11 to allow Resource Manager to identify
BIOS with fixed tables.
Update scripts to find BMP versions of 5.11.
Fix data definition for Panel Table 0 (SXGA+).
Make structure of BMP 0x05.0x11 consistent.
Update the BMP scripts for all the chips.
Modify NVResizeMemory to correctly determine the RAM width and
size. (Required for DDR memories because the memory controller
requires a signal back from the memory module to complete the memory
transfer. If no memory is present then the memory controller returns the
data in the FIFO.)
Add code to verify the last MB of memory is correctly sized.
3.11.00.17.00
Beep and abort on an invalid devinit opcode of 0x00.
Save/restore fixes.
Force 24-bit interface if 12-bit isn't strapped. (Corrects issues caused by
assuming 24-bit as the default power-up state.)
Clear TEST_CONTROL bit for PWRDWN_DAC to allow CRT detection to
work correctly. (The BIOS does not reset this bit if the detection code is
called from within an SMI and the driver is running.)
3.11.00.18.00
Clear TEST_CONTROL bit for PWRDWN_DAC to allow monitor detection to
work correctly.
Fix AND/OR.
DWORD align PCI expansion ROM data structure per PCI 2.2 spec, section 6.3.1.2.
Update HandlerNoSTI to return via a FAR RET.
Disable interrupts in FP_DDC_Init and change call to HandlerNear()
to HandlerNearEntry() because the former executes a STI. This fixes a
problem with J-DOS.
Change FPEndSetMode() to use the value 32h for the shadow lock when
an FP is installed.
[ 16. Juli 2001: Beitrag editiert von: CaptainStarMac ]
