20.Jun.2026
|
Graphics card support: P96 3.6.3
The Amiga RTG system P96 has been updated to version 3.6.3. Customers who have purchased a license within the last 12 months can download the update for free from icomp.de. Here are the details of the changes:
- ENV:Picasso96/AmigaVideo can also be set to "off" to disable the sync impulses of the Amiga video output. This will request DPMS supporting monitors connected directly to this output to either power down, or switch to an alternative input, such as the graphics card output.
- The BltMaskRastPortBitMap() function mis-computed the offset into the mask if the source was interleaved and the source Y position was different from 0. This caused bad rendering when blitting interleaved bitmaps with non-zero offsets.
- P96Prefs warned on untested video modes, even if these modes were disabled. It now only warns you on untested enabled modes.
- If a dragged screen on a video card without DAC switching or mode mixing defines less colors in its color map than the background screen, then some colors of the background screen were not initialized, leaving false colors there, even though they are not used by the dragged foreground screen.
- Blitting overlapping bitmaps with the CPU with different bitmap base addresses might have selected the copy direction only on the coordinates, but not on the addresses, thus causing incorrect blit results. This must have been a very old defect that has been finally fixed.
- Added support for the Matrox G100 with SDRAM. Thank you, Tim!
- Fixed potential erraneous behaivor of Matrox cards when changing the resolution, in particular on MMS types.
- Fixed evaluating the tool types of the Matrox drivers.
- Matrox drivers did not work properly with interrupts disabled, i.e. INTERRUPT=NO as tool type.
- The VESA Bios of Matrox cards was only evaluated for the very first out of multiple chips on the MMU cards
- The image quality of the overlay on Matrox cards was worse than necessary.
- For some arcane reason, access to the Radeon VGA ROM does not work reliable on the 68030 if data-cache bursts are enabled. The card init code now temporarily disables the data cache and data cache bursting to make the driver also work on the 68030. Thanks Tomasz for all your patience for hunting this issue down!
- Switching between native and RTG screen with the VA2000 did not always set the clock of the native screen correctly. This has been fixed.
- A new tool type CAPTUREPREFETCH was added to the VA2000 driver by which the memory prefetch value for the flicker fixer screen can be set. This should rarely be necessary as the default value should be working fine and fits to the size of the screen on which the flicker fixer outputs the captured video.
- The VA2000 monitor switch function did not return the previous state of the switch as it should have.
- The fallback functions for YUV422 rectangle fill were not correct and failed to deliver the correct colors.
- The VUY422 modes where not correctly mapped for voodoo4 and voodoo5. All lower releases are handled correctly.
- For some reason, the documentation on the 422 modes was incorrect and U and V were interchanged. Software used the modes correctly, but documentation and reality were not aligned.
- Color conversion from RGB to YUV had the same defect and interchanged U and V, but the actual fill implementation on the voodoo, radeon and ragexl swapped colors back in proper order.
- The RTG library fallback function for filling rectangles in YUV were incorrect and failing.
- The Voodoo, Radeon and RageXL drivers no longer swap U with V to fix the core library defect.
- The AmigaVideo environment variable can now also set to "Off" which will disable the sync signals on the ECS or better chipset. This may help to signal an external monitor to switch away from the input connected to the Amiga video. The other recognized value is "31kHz" which will enforce an 31kHz refresh but on an active display to allow higher performance of audio playback.
- Fixed an off-by-one error in the bounds checking in the function computing the ideal filter coefficients for the Radeon overlay in YUV mode, causing distorted display in case the YUV window got scaled. Thanks Tim for reporting.
- The BlitTemplate function of the Radeon and RageXL drivers did not work correctly if the width of the bitplane did not correspond to its modulo value.
- The RageXL line drawer ignored the pattern offset, causing line patterns to be aligned incorrectly.
- The 3DLabsPermedia2 driver is completely new. This new driver adds the following features:
- 3DLabsPermedia2 supports now doublescan modes with the workbench-friendly 2:1 aspect ratio.
- The memory test of the 3DLabsPermedia2 is now more careful, and as such supports also PCI cards based on the same chip.
- The 3DLabsPermedia2 driver now also supports bitmaps wider than 2048 pixels and allows blitting within them. The hardware blitter is then operated line-by-line.
- The 3DLabsPermedia2 driver now also supports line drawing with patterned lines and all possible line draw modes.
- Prepared the driver to also support PCI cards based on the 3DLabsPermedia2 using SDRAM instead of SGRAM. Currently, only SGRAM is supported, though switches for SDRAM are already present, yet untested.
- Recomputed the clock rates for Permedia2. Instead of >600 possible clocks with barely different settings, clock rates are now at least 5%% away and thus distinguishable.
- The 3DLabsPermedia2 will now support screen dragging! For that, ensure that INTERRUPT=Yes is enabled.
- Added support for Oliver Achten's PCMCIA-based pVision card. This card has a full blitter, a sprite and supports screen dragging with full mode mixing.
- The maximal memory clock of the S3Virge is now 85Mhz.
- P96ScreenCx is now a bit more careful fitting boards to icons and is aware of multiple boards of the same board type.
(cg)
[News message: 20. Jun. 2026, 22:57] [Comments: 0]
[Send via e-mail] [Print version] [ASCII version]
|
