Announcement

Collapse
No announcement yet.

New Vulkan Extension Proposed To Help In Emulating AMD's Old Mantle API

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • microcode
    replied
    Originally posted by Quackdoc View Post

    "Preformed correctly" implies that they preform incorrectly on dx11. which would be things like rendering issues. "Preformed better" would be the wording you are looking for. If performance is your only goal, then sure, grvk has higher potential, but the games that support mantle generally don't need that much horse power to them, By the time GRVK is ready for usage, cards will likely be powerful enough to bulldoze through regardless. but the only game i've tried recently is Inquisition, so maybe that's not a great benchmark. in this case DXVK would certainly be "enough" on any modern gpu of decent power.
    This time I'll just leave it at saying that you really, really need to learn more before commenting on these topics again.

    Leave a comment:


  • Quackdoc
    replied
    Originally posted by microcode View Post

    That is 100% unrelated to the topic. We were talking about performance. The D3D11 renderers of these games will never perform better than the Mantle renderers of these games, given drivers of equal quality; and even with a relatively immature Mantle driver, the Mantle renderer will usually win on performance. That's the whole point, nobody here mentioned that the D3D11 renderer in any of these games "flat out fails".

    The rest of your response further confirms your ignorance. Just stop.
    "Preformed correctly" implies that they preform incorrectly on dx11. which would be things like rendering issues. "Preformed better" would be the wording you are looking for. If performance is your only goal, then sure, grvk has higher potential, but the games that support mantle generally don't need that much horse power to them, By the time GRVK is ready for usage, cards will likely be powerful enough to bulldoze through regardless. but the only game i've tried recently is Inquisition, so maybe that's not a great benchmark. in this case DXVK would certainly be "enough" on any modern gpu of decent power.

    Leave a comment:


  • microcode
    replied
    Originally posted by Quackdoc View Post
    I don't remember one that flat out fails on directx
    That is 100% unrelated to the topic. We were talking about performance. The D3D11 renderers of these games will never perform better than the Mantle renderers of these games, given drivers of equal quality; and even with a relatively immature Mantle driver, the Mantle renderer will usually win on performance. That's the whole point, nobody here mentioned that the D3D11 renderer in any of these games "flat out fails".

    The rest of your response further confirms your ignorance. Just stop.
    Last edited by microcode; 19 November 2021, 04:13 PM.

    Leave a comment:


  • Quackdoc
    replied
    Originally posted by microcode View Post

    The whole point of Mantle was to enable architectural changes in the renderers of these games, this concept was emerging at that time, and Mantle lit a fire under Apple and Microsoft's asses to release their versions of this; these renderers were only released because they performed better in practice, despite Mantle drivers being much less mature.
    If Microsoft could make D3D11 renderers perform like equally-competent Mantle renderers, they wouldn't have developed D3D12 like they did; ditto OpenGL 4.x vs Vulkan. Apple's reasons for making Metal are somewhat different, because of their strategy of isolating developers from standards, but one motivation for the design of Metal is shared with Mantle/Vulkan and D3D12.
    This is completely irrelevant to what we are talking about. of the games that support mantle I don't remember one that flat out fails on directx, maybe on some cards, but not across the board. He said that DXVK should be fine in those cases. maybe one or two games have a feature or two that doesn't work on dx11. but I can't think of any mantle games that did not work right on dx11, and I have played the majority of the mainstream ones anyways, maybe an indie title?

    but I can only assume that the issue of mantle working where dx11 failed is due to vendor implementation, as I don't recall a single game otherwise. if I am wrong on this ill eat my words.

    You don't appear to know why these things exist, or how they turned out, so why do you feel the need to comment on what "could" happen? The things that make DXVK a better D3D9 driver than the real deal are completely different, and that benefit only appears when comparing the same D3D9 renderer in both cases. Mantle/Vulkan/D3D12/Metal enable entirely different renderer architectures, and those renderer architectures (combined with more predictable API overhead) are a lot of work, work that was only done because it is a big win for performance.
    I don't see why them being completely different architectures is relevant, the truth of the matter is, vulkan is lower level than dx11, and will be supported for far longer, the idea that dxvk will neither become a more consistent, nor a more performant option than vendor implementations is quite ridiculous to me. the truth of the matter is Vulkan will still progress and develop further, and dx11 vendors will stop working on them long before vulkan dies if the past is any indication.

    this isn't about over coming the limitations of DX11, that's not what DXVK needs to do, it needs to be a very consistent and optimized translation layer. as vendor drivers for dx11 can sometimes be best described as, not great. you don't overcome the limitations of DX11, you overcome the limitations of a vendors shoddy implementation of it.

    Leave a comment:


  • microcode
    replied
    Originally posted by GinDiamond View Post
    I think the GlideAPI should be emulated by vulkan next
    nGlide has had Vulkan backend since 2017.

    Leave a comment:


  • microcode
    replied
    Originally posted by Quackdoc View Post
    This is only somewhat true. Vendor implementations vary greatly, and Vulkan is seemingly across the board better than DX11 for obvious reasons (I haven't tested everything so who knows if this is actually true.)

    DXVK could very well get to the point where it is implemented across the board better than it than native drivers.

    of course this doesn't help if its a flaw in the games coding. but in cases where driver implemtation is the issue. (OpenGL on AMD Windows is a notorious example) . It could very well be more stable and preformant.

    this is already often the case for D3D9 games. where DXVK can smoke native drivers even on windows. because no one cares about D3D9 drivers anymore. besides DXVK.
    The whole point of Mantle was to enable architectural changes in the renderers of these games, this concept was emerging at that time, and Mantle lit a fire under Apple and Microsoft's asses to release their versions of this; these renderers were only released because they performed better in practice, despite Mantle drivers being much less mature.
    If Microsoft could make D3D11 renderers perform like equally-competent Mantle renderers, they wouldn't have developed D3D12 like they did; ditto OpenGL 4.x vs Vulkan. Apple's reasons for making Metal are somewhat different, because of their strategy of isolating developers from standards, but one motivation for the design of Metal is shared with Mantle/Vulkan and D3D12.

    You don't appear to know why these things exist, or how they turned out, so why do you feel the need to comment on what "could" happen? The things that make DXVK a better D3D9 driver than the real deal are completely different, and that benefit only appears when comparing the same D3D9 renderer in both cases. Mantle/Vulkan/D3D12/Metal enable entirely different renderer architectures, and those renderer architectures (combined with more predictable API overhead) are a lot of work, work that was only done because it is a big win for performance.

    Leave a comment:


  • GinDiamond
    replied
    I think the GlideAPI should be emulated by vulkan next

    Leave a comment:


  • Quackdoc
    replied
    Originally posted by microcode View Post

    No. The limitations of D3D11 don't magically go away when you emulate it on Vulkán.

    ​​​​​​Why do so many ignorant people feel the need to spread their ignorance around here? Just read, nobody needs to hear your uninformed take.
    This is only somewhat true. Vendor implementations vary greatly, and Vulkan is seemingly across the board better than DX11 for obvious reasons (I haven't tested everything so who knows if this is actually true.)

    DXVK could very well get to the point where it is implemented across the board better than it than native drivers.

    of course this doesn't help if its a flaw in the games coding. but in cases where driver implemtation is the issue. (OpenGL on AMD Windows is a notorious example) . It could very well be more stable and preformant.

    this is already often the case for D3D9 games. where DXVK can smoke native drivers even on windows. because no one cares about D3D9 drivers anymore. besides DXVK.

    Leave a comment:


  • microcode
    replied
    Originally posted by loganj View Post
    probably using dxvk is enough already
    No. The limitations of D3D11 don't magically go away when you emulate it on Vulkán.

    ​​​​​​Why do so many ignorant people feel the need to spread their ignorance around here? Just read, nobody needs to hear your uninformed take.

    Leave a comment:


  • emblemparade
    replied
    Originally posted by ZuseZ4 View Post
    I'm surely glad that we had Mantle as it paved the way for Vulkan, but is it really worth the effort for those like .. 3(?) games?
    It's worth the effort for archival preservation. DOS is almost gone today, but DOSBox, as an open source project, will allow it to live on for centuries.

    This is obviously not a high-priority project by any means. But I'm very glad someone is working on it.

    Leave a comment:

Working...
X