Ein Aprilscherz ?? APG ATI RADEON HD3650 für powermac g5 8-
  • Just looking around
    filos
    Posts: 13 from 2014/4/20
    From: darmstadt
    ich habe das ebay gefunden

    http://www.ebay.de/itm/142223434944?_trksid=p2060353.m1438.l2649&ssPageName=STRK%3AMEBIDX%3AIT

    den verkäufer habe ich angeschrieben
    und er sagt das es unter mac osx 100 % läuft
    aber nur mit einen dvi ausgang

    fake oder real 8-) ich denke der hat den ati kext geändern oder ??
    und im netzt finde ich nix darüber
  • »08.04.17 - 11:15
    Profile
  • Moderator
    Kronos
    Posts: 1639 from 2003/2/24
    Wahrscheinlich geflasht wär also problematisch unter MorphOS.

    Wobei ja angeblich RadeonHD Karten auch mit x86 BIOS gehen sollen.
    Die Frage ist halt was du damit willst was eine normale Apple-AGP 9600-9800 nicht leisten könnte.

    Edith:

    Wie z.B. die die du anscheinend grad selber bei iBäh anbietest .........

    [ Edited by Kronos 08.04.2017 - 12:43 ]
  • »08.04.17 - 11:40
    Profile
  • Jim
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Jim
    Posts: 3947 from 2009/1/28
    From: Delaware, USA
    If you are talking Apple compatible cards, then you are still talking AGP (since the PCIe Mac aren't supported).

    I don't know if Bigfoot intends to continue to support AGP variants of Radeon video cards moving forward.
    He was working with R600 video cards recently to add acceleration, but I'm not sure how far along that work is, and the HD3650 AGP card I've bought him has not been shipped yet (its still here on my desk).

    But if he does continue to support AGP cards, you would probably be better off with an unflashed PC HD3650.

    I don't know of any Apple roms that are compatible with PPC systems.


    in any case, Kronos is right, for now your best bet is a 9600 (its what I use).
  • »08.04.17 - 12:12
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 8977 from 2003/5/22
    From: Germany
    > Wobei ja angeblich RadeonHD Karten auch mit x86 BIOS gehen sollen.

    Nicht nur angeblich. Allerdings nicht unter Mac OS X (und OpenFirmware). Insofern scheint mir die Aussage des Verkäufers sehr zweifelhaft.
  • »08.04.17 - 12:30
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 8977 from 2003/5/22
    From: Germany
    > I don't know if Bigfoot intends to continue to support AGP variants of
    > Radeon video cards moving forward.

    I'd think if he supports PCIe-based Radeon HD 2000/3000/4000 cards he won't artificially restrict support for AGP variants of the same cards.
  • »08.04.17 - 13:09
    Profile
  • Order of the Butterfly
    Order of the Butterfly
    DanicaTalos
    Posts: 352 from 2010/10/15
    From: 01101110011100...
    Für Linux funktionieren diese Karten eher. Allerdings sehr buggy und ohne 3D-Acceleration.
  • »08.04.17 - 13:27
    Profile
  • Just looking around
    filos
    Posts: 13 from 2014/4/20
    From: darmstadt
    das einzige was ich gefunden habe ist das

    http://www.insanelymac.com/forum/topic/107526-ati-radeon-hd-pkg-installers-105x-106/

    unter ubuntu läuft glaube ich nur die hd 6570 aber pcie 8-/
  • »08.04.17 - 14:11
    Profile
  • Jim
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Jim
    Posts: 3947 from 2009/1/28
    From: Delaware, USA
    Quote:

    Andreas_Wolf wrote:
    > I don't know if Bigfoot intends to continue to support AGP variants of
    > Radeon video cards moving forward.

    I'd think if he supports PCIe-based Radeon HD 2000/3000/4000 cards he won't artificially restrict support for AGP variants of the same cards.


    AGP variants require a bridge chip, and that hasn't worked well with the R500 cards.
    So, it isn't an artificial restriction, supporting AGP variants requires extra work.

    And right now, I've shipped Mark an AGP 4650 (so he has an R700), but the R600 AGP cards are still on my desk (each package to South Africa costs me about $100) and I still have one more card (an GCN card) I want to include in this package.

    I'll try to get the remaining R600s out by the end of April, but Mark is already working on the R600 drivers, so he doesn't have cards to test an AGP driver with (yet).

    In my defense, he wasn't sure he was going to continue to support the AGP cards as there aren't any good Linux driver examples. But since there are only R600 and R700 cards remaining (that support the AGP interface), I thought I'd send them anyway.

    [ Edited by Jim 08.04.2017 - 10:15 ]
  • »08.04.17 - 14:54
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 8977 from 2003/5/22
    From: Germany
    > das einzige was ich gefunden habe ist das
    > http://www.insanelymac.com/forum/topic/107526-ati-radeon-hd-pkg-installers-105x-106/

    Da geht's aber nicht um PPC-Macs.
  • »08.04.17 - 15:01
    Profile
  • Jim
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Jim
    Posts: 3947 from 2009/1/28
    From: Delaware, USA
    Quote:

    DanicaTalos wrote:
    Für Linux funktionieren diese Karten eher. Allerdings sehr buggy und ohne 3D-Acceleration.


    From what Mark said the last time he mentioned it, I think his driver is actually likely to be pretty good.
    He was working on overlay functionality at the time, and the acceleration functions probably are tricky, as this is the first family with no 2D primitives.
    But in his usual positive perspective he said that creating the 2D functions from the 3D primitives will give him a better understanding of the 3D functions.

    The guy's attitude is kind of inspirational.
  • »08.04.17 - 15:23
    Profile
  • Just looking around
    filos
    Posts: 13 from 2014/4/20
    From: darmstadt
    Quote:

    Andreas_Wolf schrieb:
    > das einzige was ich gefunden habe ist das
    > http://www.insanelymac.com/forum/topic/107526-ati-radeon-hd-pkg-installers-105x-106/

    Da geht's aber nicht um PPC-Macs.


    stimmt ich habe es voll übersehen
  • »08.04.17 - 15:33
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 8977 from 2003/5/22
    From: Germany
    > AGP variants require a bridge chip, and that hasn't worked well with the R500 card.
    > So, it isn't an artificial restriction, supporting AGP variants requires extra work.

    I'm not sure how to conclude the requirement of extra work just from the fact that the faster AGP transfers don't work with the AGP variants. So yes, their bandwidth is lower than it could be, but where does the extra work come from for supporting the AGP variants?
  • »08.04.17 - 15:36
    Profile
  • Jim
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Jim
    Posts: 3947 from 2009/1/28
    From: Delaware, USA
    Quote:

    Andreas_Wolf wrote:
    > AGP variants require a bridge chip, and that hasn't worked well with the R500 card.
    > So, it isn't an artificial restriction, supporting AGP variants requires extra work.

    I'm not sure how to conclude the requirement of extra work just from the fact that the faster AGP transfers don't work with the AGP variants. So yes, their bandwidth is lower than it could be, but where does the extra work come from for supporting the AGP variants?


    The reason the transfer rate is basically 1X is that it has to go across the bridge chip, and Mark does not have documentation of how that chip works.
    But it isn't transparent, it needs to be programmed, so whatever work around he managed to implement with the R500 driver did require additional work.
  • »08.04.17 - 15:45
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 8977 from 2003/5/22
    From: Germany
    > The reason the transfer rate is basically 1X is that it has to go across the bridge chip,
    > and Mark does not have documentation of how that chip works. But it isn't transparent,
    > it needs to be programmed, so whatever work around he managed to implement with
    > the R500 driver did require additional work.

    So this existing "work around" for the bridge chip of the R500 cards doesn't work with the bridge chip of the R600 (or R700) cards? Pity.
  • »08.04.17 - 22:47
    Profile
  • Jim
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Jim
    Posts: 3947 from 2009/1/28
    From: Delaware, USA
    Quote:

    Andreas_Wolf wrote:
    > The reason the transfer rate is basically 1X is that it has to go across the bridge chip,
    > and Mark does not have documentation of how that chip works. But it isn't transparent,
    > it needs to be programmed, so whatever work around he managed to implement with
    > the R500 driver did require additional work.

    So this existing "work around" for the bridge chip of the R500 cards doesn't work with the bridge chip of the R600 (or R700) cards? Pity.


    I have no idea, Andreas.
    You'd have to ask Mark. I know far less about the entire matter than he does.
    It was his idea to use the AGP X1950Pro to develop the R500 driver (he asked for it).
    Frank already had a really nice PCIe variant.

    And I've never looked into bridge chip functionality, and I'm just going on what little he mentioned to me.
    I don't even know if they are the same components (on the R500, R600, and R700 cards).

    Mark just mentioned that it was not well documented.
    I'm hoping that he can get it to work better with these cards, because it limits the utility of the R500s in AGP systems.
    Otherwise I'd be using one right now.

    An 3650 or 3850 would make a nice upgrade.

    However, its worth mentioning (yet again), that buying hardware that support hasn't been announced for is a bad idea.

    [ Edited by Jim 08.04.2017 - 18:40 ]
  • »08.04.17 - 23:36
    Profile
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Andreas_Wolf
    Posts: 8977 from 2003/5/22
    From: Germany
    >> So this existing "work around" for the bridge chip of the R500 cards doesn't work with
    >> the bridge chip of the R600 (or R700) cards?

    > I have no idea, Andreas. [...] I've never looked into bridge chip functionality [...]. I don't
    > even know if they are the same components (on the R500, R600, and R700 cards).

    If they are (or if they aren't, but the existing "work around" for the bridge chip of the R500 cards also works with the bridge chip of the R600 (or R700) cards), future extra work to support AGP variants of R600 (or R700) cards shouldn't be required.
    Searching the WWW I've found that the bridge chip has kept the same name (RIALTO, as mentioned by me 5½ years ago) throughout all card generations up to and including R700 cards. Of course, it could still be incompatible revisions of the same chip or even completely different chips despite the same name being communicated externally.
  • »09.04.17 - 00:22
    Profile
  • Jim
  • Yokemate of Keyboards
    Yokemate of Keyboards
    Jim
    Posts: 3947 from 2009/1/28
    From: Delaware, USA
    Quote:

    Andreas_Wolf wrote:
    >> So this existing "work around" for the bridge chip of the R500 cards doesn't work with
    >> the bridge chip of the R600 (or R700) cards?

    > I have no idea, Andreas. [...] I've never looked into bridge chip functionality [...]. I don't
    > even know if they are the same components (on the R500, R600, and R700 cards).

    If they are (or if they aren't, but the existing "work around" for the bridge chip of the R500 cards also works with the bridge chip of the R600 (or R700) cards), future extra work to support AGP variants of R600 (or R700) cards shouldn't be required.


    That is a safe assumption.
    I am hoping Mark can get the transfer speeds up above what the R500 has managed.
    Still, its not a worse solution than OS4 users plugging PCIe video cards in PCI adapters.
  • »09.04.17 - 00:35
    Profile