Announcement

Collapse
No announcement yet.

AMD Marketing Manager Mentions Linux & Mantle

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

  • #41
    Originally posted by Gulliver View Post
    Guys,

    What is HSLS?

    I tried to google it but found no answer.
    Proprietary shading language developed by Microsoft :

    Ladybug Mixer Vn: A Video Mixer Supports Real-time Composition of Multi-channel Video with HLSL Effects in GPU.

    "compatible with directx HLSH for simplified porting": http://images.anandtech.com/doci/7371/MantleHLSL.jpg

    Comment


    • #42
      Originally posted by Gulliver View Post
      Guys,

      What is HSLS?

      I tried to google it but found no answer.
      HLSL!

      Comment


      • #43
        Originally posted by bridgman View Post
        Good call. I can answer questions about HSA but not Mantle or future Catalyst plans ... which is a shame because the Mantle architect sits across the aisle from me
        Mantle, DX12, Metal (for obvious reason lets use Metal as example), all introduce somehow more flexible GPU pipeline. Execution do not have to be linear any more.

        Is HSA able to deliver such idea to "everything"?

        (Oh, now I will run OpenCL on this engine, this DMA will fetch some 2D textures for my next computation, here I will run some shaders, which output will be used for second pass of that OpenCL code, while when DMA finish I will run another shader....)

        How HSA fit into that flexibility trend?

        Comment


        • #44
          Originally posted by bridgman View Post
          Good call. I can answer questions about HSA but not Mantle or future Catalyst plans ... which is a shame because the Mantle architect sits across the aisle from me
          Mantle, DX12, Metal (for obvious reason lets use Metal as example), all introduce somehow more flexible GPU pipeline. Execution do not have to be linear any more.

          Is HSA able to deliver such idea to "everything"?

          (Oh, now I will run OpenCL on this engine, this DMA will fetch some 2D textures for my next computation, here I will run some shaders, which output will be used for second pass of that OpenCL code, while when DMA finish I will run another shader....)

          How HSA fit into that flexibility trend?

          Comment


          • #45
            Will the NVIDIA fanboys point me to the magical driver and hardware?
            Because my experience with nvidia hw and driver was crap and I had enough of the "buy NVIDIA, its magical" crap on this forums in AMD threads.
            I had a MSI notebook with GeForce Go 7600, installing drivers was a pain and when i needed multimonitor i had to hack xorg.conf
            In the end died due to the bumpgate problem.
            Then i moved back to win due to gaming.
            Then i moved back to linux due to advancements in wine.
            My card was AMD and it had its share of problems and some games would not run via wine, some due to blob other due to NVIDIA specific functions or hacks in wine.
            Then i wanted to add another gpu for XBMC seat. Bought GT 640.
            Well that didnt work as planed, after installing the blob and reboot i got only black screen.
            Problem was that if you had Intel driver installed and loaded you could not use the blob.
            Bought HD 6450 and all was working using blob or foss.
            Than started to play with virtualization and vga-passtrough.
            For NVIDIA you needed to hack the xen so when the time came for upgrade i went with AMD again.
            And the funny thing is that about 6 months ago my boss needed to connect projector to his linux nb with NVIDIA and he could not set it as second screen.
            Hope he was using old blob or else thats a big fail.

            Now i am running Intell HD 4600 for desktop, foss AMD HD 6450 for XBMC with vdpau acceleration and AMD R9 290 for my winconsole 8.1 via kvm and vfio vga-passtrough.
            My only complain is that i cannot reboot the winconsole8.1 as it would not start without restarting the whole PC.

            Comment


            • #46
              First of all OpenGL should make huge steps forward before its too late. Second, AMD FU! I can't even play a [email protected] single game with your open-source drivers on a 8 years old GPU!

              Fix your drivers first and then make my "future bright" with Mantle...

              Comment


              • #47
                Originally posted by Debcool View Post
                Now, try to do it with an 4850 (rv770)
                http://www.x.org/wiki/RadeonFeature/#4 Point 4 in "Footnotes"

                I lend my 7850 to a friend (win user) since I've got my GTX760, so I can't tell.
                I think that AMD never promised to deliver video decode on the older hardware. If you buy an AMD card today (except 290/290X which is not working reliably yet), it will run fine for 2D, 3D and video acceleration.

                Originally posted by Debcool View Post
                Now, I'd rather prefer to use only FOSS, I mean it. But can you seriously telling me that today, if you want to play big titles with most of the options "enable", you can do that with an only FOSS system? I doubt it.
                People reported that Metro: Last Light runs fine. I don't know if that is a big title for you, or if you thought of something bigger? The performance is not as good as with the proprietary drivers of course, but the game works. You cannot enable all options, because some OpenGL-4 features like tesselation are not yet supported by Mesa.
                Phoronix, Linux Hardware Reviews, Linux hardware benchmarks, Linux server benchmarks, Linux benchmarking, Desktop Linux, Linux performance, Open Source graphics, Linux How To, Ubuntu benchmarks, Ubuntu hardware, Phoronix Test Suite


                Originally posted by Debcool View Post
                Don't worry ( :-) ), the day I found that my hardware could run nicely with only FOSS, I'll do it without any regrets of pushing away blob's drivers.
                In other words, you don't really care and will only choose FOSS over binary blobs if there is no disadvantage in doing so

                Comment


                • #48
                  Originally posted by bridgman View Post
                  Good call. I can answer questions about HSA but not Mantle or future Catalyst plans ... which is a shame because the Mantle architect sits across the aisle from me
                  Please, just implement Mantle in an oss driver, there is no point in having it in Catalyst, since no legacy support needed. Cough, shure it is may be easyer to have just one implementation for it cross platform, but appart from that.

                  Comment


                  • #49
                    Originally posted by Debcool View Post
                    ...and I am still waiting to see. {...}
                    If that company want to get more customers or even just keep them, why doesn't she make it real? Nvidia, Intel, Valve... did and still do.
                    I am fed up waiting, wishing, hoping... Stop promessing, just do it and shut your mouth until then!
                    What? you're writing a post complaining about delays and long waiting, and you cite Valve as an example ?
                    Whoa...
                    Mind Blown level = over 9'000

                    :-D

                    Seriously, Catalyst driver have become better as of late, and the opensource support is decent (though not for the latest generation, as usual).
                    And at least you've getting better AMD driver faster than episodes of Half-life ;-)

                    Comment


                    • #50
                      Originally posted by DrYak View Post
                      Seriously, Catalyst driver have become better as of late...
                      Yes, Catalyst is by far not as bad as its reputation.

                      Then again I think AMD is lost when it comes to Catalyst and desktop/gaming.
                      The reputation has been burned into the heads of most of the Linux community.
                      Even if AMD would be able to pull an almost perfect blob with a snap - I'm sure the reputation would still remain for a very long time.
                      There seems to be a pretty unfair hysteresis connected to that matter...

                      For Nvidia, on the other hand, the situation seems to be reversed.

                      Comment

                      Working...
                      X