P96 V3.5.0 released

Caution: Non registered users only see threads and messages in the currently selected language, which is determined by their browser. Please create an account and log in to see all content by default. This is a limitation of the forum software.


Please understand that you need to create an account to be able to post, guest posting was disabled as an anti spam measure.

  • The new version V3.5.0 contains updates for the Trio64 (Cybervision), IMSG300 (Visiona) and ET4000 (Merlin/oMNiBus) drivers, and it adds new drivers for Vooodoo Banshee, the Voodoo 3 and Voodoo 4/5 class of graphics chips. This requires a PCI bridge board that is supported by the OpenPCI.library, which is not part of the package, but can be freely downloaded.


    V3.5.0 also contains an early alpha version of the framebuffer driver for Indivision ECS V4. This driver needs a processor with MMU and an installed MMU library to work. Note that you may have to update the FPGA core of your Indivision ECS V4 card with the latest one from the Wiki. Supported colour modes are 8-bit (256 colours) and 15/16 bit (65k colours), dual screens with screen dragging and panning for oversized screens. Note that Indivisino ECS V4 does not yet have a blitter, so don't expect high speed!


    All changes are summarized in the Wiki article of P96.

    Unfortunately, this current shop system is still giving us a hard time, with lots of problems during the order process, and even more problems when we attempt to update the downloadable files, so we don't even try for this new version. We'll send the new version via eMail upon request, but will also pro-actively send out eMails over the course of the next few working days.


    We with everyone happy holidays!

  • Nice!


    How is working frame buffer for the Indivision ?


    Does it means we could enjoy faster 320x256 8 bit fast (without c2p) for games like Doom? Will it have bitter/faster operation later ?

    Could t also be done on the Indivision AGA mk3?


    Thanks for your work. It sounds great 👍


    Merry Christmas 🎁!

    Cheers

    Nicolas

  • Does it means we could enjoy faster 320x256 8 bit fast (without c2p) for games like Doom? Will it have bitter/faster operation later ?

    Please read this thread for answers about that.


    Could t also be done on the Indivision AGA mk3?

    No, the memory of that product is wired in a way that it doesn't allow random access. Although Indivisino ECS V4 looks very similar to the predecessors, the wiring between memory and FPGA is quite different.


    Jens

  • ...as mentioned in the product description, yes. Unfortunately, I now have trouble with my eMail server as well. I already contacted my admin.


    Jens

  • I'm prioritising current orders, but have been sending out the updates to customers "backwards" until late october as of now. If you want your update sooner, send me an eMail, don't forget to include the last paid order ID.


    Jens

  • Hi,

    I've a question for S3Virge diver.

    I've got a S3 Virge DX 4MB PCI on Mediator 1200 TX+, I'm using P96 3.5.0 (S3Virge.card 43.71 and S3Virge.chip 43.48), openpci.library v14.4, but the performance is quite low according to P96Speed, it seems like hardware/blitter acceleration is off.

    I've to set some particular tooltype on monitor icon?


    Thanks


    Here are tooltype on monitor icon and p96 test:


    (ABOVE=<name>)

    (BELOW=<name>)

    (BIGSPRITE=YES|NO)

    (BLACKLEVEL=Black|Pedestal)

    BoardType=S3Virge

    (BORDERBLANK=YES|NO)

    (BORDERKLUDGE=YES|NO)

    DISPLAYCHAIN=NO

    (FAKENATIVEMODES=YES|NO)

    FASTDRAWLINE=YES

    (FONTSFIX=YES|NO)

    GRANTDIRECTACCESS=YES

    IGNOREMASK=YES

    INTERRUPT=NO

    (INVMODE=YES|NO)

    LEFT_OF=Voodoo

    (MEMORYCLOCK=75)

    (MEMORYTYPE=New|Old)

    (MONITORHEIGHT=<value in mm (100..2000)>)

    (MONITORWIDTH=<value in mm (100..2000)>)

    (NOBLITTER=YES|NO)

    (NOMASKEDBLITS=YES|NO)

    (NOPLANAR2CHUNKYBLITS=YES|NO)

    (OVERCLOCK=YES)

    (QUIET=YES|NO)

    (RIGHT_OF=<name>)

    SettingsFile=SYS:Devs/Picasso96Settings

    (SOFTSPRITE=YES|NO)

    (SWITCHTYPE=NONE|JAVOSOFT)

    (SYNC=Tessellated|Composite|Separate|Green)

    (SYSTEM2SCREENBLITS=YES|NO)

    (TRUECOLORKLUDGE=YES|NO)

    (TRUECOLORPIXSHIFT=<value in pixel (0..256)>)

    (WAITSTATES=<value in cicli di clock (2..4)>)


    P96Speed 1.2 - ©`97-99 by Jens Langner

    .-----------------------------------------.

    | Computer......: Amiga 1200/AGA |

    | CPU...........: 68060/50 MHz |

    | OS / WB.......: V47.102/V47.4 |

    | SetPatch......: V47.8 |

    | Chip/Fast.....: 1.9MB/128.0MB |

    | Graphics card.: NO GRAPHICS CARD |

    | GFX system....: Picasso96 v |

    | Resolution....: 640 x 480 x 8 |

    | Depth/Colors..: 256 colors |

    | Testlength....: 13 |

    +-----------------------------------------+

    | Description...: all |

    | : |

    | : >ÚŽ |

    | Owner.........: Sante |

    | E-Mail........: ziosante@libero.it |

    `-----------------------------------------'

    .============= SPEEDRESULTS ==============.

    | RectFill()................ 959 op/s |

    | RectFill() Pattern........ 951 op/s |

    | WritePixel().............. 1221 op/s |

    | WriteChunkyPixels()....... 451 op/s |

    | WritePixelArray8()........ 440 op/s |

    | WritePixelLine8()......... 1110 op/s |

    | DrawEllipse()............. 1065 op/s |

    | DrawCircle().............. 1072 op/s |

    | Draw().................... 1077 op/s |

    | Draw() Hor/Ver............ 1109 op/s |

    | ScrollRaster() X.......... 211 op/s |

    | ScrollRaster() Y.......... 210 op/s |

    | PutText()................. 958 op/s |

    | BlitBitMap().............. 991 op/s |

    | BlitBitMapRastPort()...... 658 op/s |

    | BitMapScale()............. 94 op/s |

    |--------------- Intuition ---------------|

    | OpenWindow().............. 31 op/s |

    | MoveWindow().............. 860 op/s |

    | SizeWindow().............. 23 op/s |

    | CON-Output................ 118 op/s |

    | ScreenToFront()........... 46 op/s |

    `========================================='

  • I've just done that test and after I tested also changing other tooltype like:

    SYSTEM2SCREENBLITS=YES

    NOMASKEDBLITS=NO

    NOPLANAR2CHUNKYBLITS=NO


    but the peformance remains the same

  • Why do you avoid the NOBLITTER tooltype? It may be switched off all the time, so the tweaking you're doing may just do nothing if the blitter is switched off in the first place.


    Jens

  • In previous post I meant I made these tests:


    1) setting only NOBLITTER=NO


    2) setting

    NOBLITTER=NO

    SYSTEM2SCREENBLITS=YES

    NOMASKEDBLITS=NO

    NOPLANAR2CHUNKYBLITS=NO


    3) all tooltypes disabled, i.e. between (), so they take the default values


    in all cases the performance of S3Virge is the same and it is very low

    just an example with P96Speed:

    WritePixel() on native AGA screen PAL at 640x512x8 scores 37.072 op/s

    WritePixel() on S3Virge screen at 640x480x8 scores 1.221 op/s


    This looks very strange to me, it should be faster

  • Can you run MuScan on the machine and post the output here? Also, what kind of accelerator are you running? I see 68060/50, but not the type of accel. There are huge differences in access time to the A1200 bus.


    Jens

  • Looking at the speed test results again, I can see that WritePixel is indeed slow, but RectFill isn't that much slower, meaning, the blitter is active. Something else is causing the slowdown, and it may be cache (hence the question for MuScan output), the Mediator hardware (which needs to do banking, which is slooooow), a missing SetPatch (btw, what version do you run?) or the OpenPCI implementation.


    BTW, did you ever try the original Mediator PCI drivers? I believe that S3 is supported by their drivers.


    Jens

  • S3 Virge driver from Elbox works, it is faster on P96Speed but it has various bugs and it does not support all features from P96 v3+.

    I have a TF1260 accelerator. SetPatch is 47.8, I'm running OS3.2.2.1. Openpci.library actually is v14.4 from Thomas Ritcher


    Here Muscan output: muscan.txt

    Here lspci output: lspci.txt


    thanks for the help

  • What's the screenmode you're using? If it's the 16-colour "planar" mode, then it's expected to be slow, as that's emulated. Please try a 256-colour mode.


    Thor also analyzed your MuScan output. The verdict is: It's an architectural problem of Mediator. The large PCI address space must be somehow mapped into 4MByte of Z2 space, which also affects the IO registers. When doing "writepixel", a series of bank switch request will happen, which make things slower.


    Unfortunately, the blitter registers can't be reached through the legacy IO registers (which are faster to switch to), so all we could do is to use legacy IO registers where possible, but it will not solve the root cause of mapping the huge address space to the relatively small Zorro space.


    If Elbox would change the way they map memory and registers, i.e. keep the blitter registers available all the time while being able to move through actual memory space on a different banking register, that would help big time. The expected value for WritePixel() is around 14.000, so about an order of magnitude faster than your current measurement.


    If you have the MMU-based memory expansion feature from Elbox enabled, please switch that off. Your accelerator should have enough memory.


    Jens

  • thanks,

    the primary problem is the 4MB Window jumper I must set on my Mediator to make it works with TF1260.

    Maybe TF1260 needs a firmware upgrade to make it works with 8MB Window jumper setting....

  • The last reply was more than 365 days ago, this thread is most likely obsolete. It is recommended to create a new thread instead.