Announcement

Collapse
No announcement yet.

The S3TC Patent Finally Expires Next Week - S3 Texture Compression

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

  • #11
    Originally posted by ElectricPrism View Post
    Someone please explain how this effects any current scenario real world situation for the better?
    This patent effects wine project Direct x as well. Reality here is opengl and direct x used the same compression. Opengl calls the compression s3tc and Direct x calls it dxtn but under the hood is the same thing.

    The current libtxc_dxtn is only software running in cpu. With patent expired this will open up for dxtn/s3tc to be implemented in gpu drivers and running on the gpu this can in fact for programs using this compression to perform better due to reduced memory transfer between system memory and gpu memory. Why able to transfer the textures compressed to the gpu then have them decompress on the gpu. Of course this advantage most likely will not appear straight away. But this difference is a performance different to some programs running under wine being slower.

    Comment


    • #12
      Originally posted by oiaohm View Post
      ...
      The current libtxc_dxtn is only software running in cpu. With patent expired this will open up for dxtn/s3tc to be implemented in gpu drivers and running on the gpu this can in fact for programs using this compression to perform better due to reduced memory transfer between system memory and gpu memory. Why able to transfer the textures compressed to the gpu then have them decompress on the gpu. Of course this advantage most likely will not appear straight away. But this difference is a performance different to some programs running under wine being slower.
      That is not correct. libtxc_dxtn is only needed to provide on-line compression if required (and decompression for gpus that don't support s3tc). Afaik Intel enabled s3tc support even without this lib years ago.

      Comment


      • #13
        Originally posted by log0 View Post
        That is not correct. libtxc_dxtn is only needed to provide on-line compression if required (and decompression for gpus that don't support s3tc). Afaik Intel enabled s3tc support even without this lib years ago.
        Its not that straight forwards about gpus that don't support s3tc. There are quite few GPU that can do s3tc decoding and encoding and its not a firmware/hardware feature so current they act like GPU that cannot do s3tc at all.

        http://fileadmin.cs.lth.se/graphics/...006/thesis.pdf
        I know this one is compression but there is another old paper using glsl for s3tc decompression. The reality is most gpu have the instructions to perform s3tc compression and decompression inside them. Both of these features were researched then stopped due to patents. So for gpu that don't have s3tc in chip implementing s3tc as a shader or equal for the gpu will come possible after the patent expires.

        Also do note I should have been clear that some games require both ways being compression and decompression as fast as possible. So game generates some textures compresses them them dumps them to disc so they don't need to regenerate them or just compress them leaves them in gpu until they have to move them. Now using the software compression of libtxc_dxtn where it has to come back to cpu memory hurts those games big time and if the card does not have s3tc compression as well welcome to even more pain.

        https://www.phoronix.com/scan.php?pa...tem&px=MTI3OTc
        If you read back here basically you need both compression and decompression of s3tc for all programs to work right. On-line compression need to be performable in the GPU and cpu if possible because the application may request it done when the data is either in cpu or gpu and transporting around to compress is not a great thing. Also do note that without on-line compression installed particular decompression modes of s3tc are also still disabled under intel.

        So intel did not enable full s3tc decompression in hardware without the lib to compress years ago. Only part s3tc decompression was enabled without the lib. So once the patent expires there are a lot of clean ups like implementing the compression better and getting to the point that full s3tc decompression is all by default.

        Comment


        • #14
          Get a patent to be used by mainstream and you get 20 fucking years to live like a parasite. The patent laws are such a clusterfuck.

          Comment


          • #15
            The sad thing is the S3TC patent probably wouldn't have stood up in court anyway. In a court, the "ordinary observer" test compares the design described in the patent, in the context of prior art. As noted in many places, the S3TC patent is strikingly similar to Color Cell Compression published in 1983. (http://ieeexplore.ieee.org/iel5/498/3989/00152671.pdf) which is not patented. From the context of prior art of the Color Cell Compressoin method, the patent on S3TC is not considered novel or innovative. It probably would have failed the ordinary observer test and been invalidated in a court if it was ever challenged.
            Last edited by Sidicas; 30 September 2017, 10:28 AM.

            Comment


            • #16
              Originally posted by cl333r View Post
              Get a patent to be used by mainstream and you get 20 fucking years to live like a parasite. The patent laws are such a clusterfuck.
              Well remember, the original idea makes sense - allow the independent small inventors protect their innovators from big companies that would steal their ideas.

              But in practice it reversed the problem - small inventors and innovators get buried in fake patent lawsuits and associated legal fees, and the big companies use patents to stifle competition.

              So generally speaking, I hate the idea of patents. However, the entire business model of the pharmaceutical industry is built around them. No company would spend billions researching a new drug for Alzheimer's disease or AIDS or Cystic fibrosis if every other drug company could just copy the result and sell it for pennies per pill. How do you remove patents without stifling drug research? I honestly don't know.

              Comment


              • #17
                Originally posted by Michael_S View Post

                Well remember, the original idea makes sense - allow the independent small inventors protect their innovators from big companies that would steal their ideas.

                But in practice it reversed the problem - small inventors and innovators get buried in fake patent lawsuits and associated legal fees, and the big companies use patents to stifle competition.

                So generally speaking, I hate the idea of patents. However, the entire business model of the pharmaceutical industry is built around them. No company would spend billions researching a new drug for Alzheimer's disease or AIDS or Cystic fibrosis if every other drug company could just copy the result and sell it for pennies per pill. How do you remove patents without stifling drug research? I honestly don't know.
                Well, let's just ignore the implications of artificial chemical drugs for a moment and just realize that software is not the same thing. There are valuable reasons why open source software movements happened and public domain medical drug movements didn't happen.

                Comment


                • #18
                  Originally posted by Michael_S View Post
                  Well remember, the original idea makes sense - allow the independent small inventors protect their innovators from big companies that would steal their ideas.

                  But in practice it reversed the problem - small inventors and innovators get buried in fake patent lawsuits and associated legal fees, and the big companies use patents to stifle competition.

                  So generally speaking, I hate the idea of patents. However, the entire business model of the pharmaceutical industry is built around them. No company would spend billions researching a new drug for Alzheimer's disease or AIDS or Cystic fibrosis if every other drug company could just copy the result and sell it for pennies per pill. How do you remove patents without stifling drug research? I honestly don't know.
                  Often times though the patents allow the pharmaceutical companies to jack up prices to such an extent that they become unaffordable for most people for those 20 year periods. And yes funneling billions in research for a quick result is respectable and deserving compensation however you have to wonder if other researchers could have made the same discovery at some point naturally within that 20 year period without the several billion price tag and could have released it in under that 20 year time frame at a much lower price in a patentless system. After all even without billions funneled in research for each novel new drug there will still be active researchers and advances in technology that allow for easier discoveries of new information. We can't really tell if that massive monetary infusion for quicker discoveries is actually drastically reducing the time to discoveries or only slightly because we've always had this patent system in modern medicine.

                  This exorbitant pricing of drugs is also taken into consideration with health insurance costs which, at least in the US, raises premiums substantially and limits access to good insurance for many people. And if more people had easier access to affordable health care you could argue they'd be more likely to take part in preventative health measures which could theoretically limit the need for more advanced medical treatments for many of the preventable diseases we're funneling billions of dollars into research for new drugs.

                  Comment


                  • #19
                    Patents are only there to prevent competition. If I come up with a drug made from fruits and this drug cures "cancer", my design is copyrighted and that is very fair. But patents go further and prohibit others from designing another "fruit based" drug that cures "cancer" even if their design was not remotely copied from mine. And that is completely stupid.

                    Comment


                    • #20
                      Originally posted by Michael_S View Post

                      ... No company would spend billions researching a new drug for Alzheimer's disease or AIDS or Cystic fibrosis if every other drug company could just copy the result and sell it for pennies per pill. How do you remove patents without stifling drug research? I honestly don't know.
                      No, drug companies don't spend "billions of dollars researching Alzheimers" and if they do it's only like 1 or 2 companies in the entire world who spends that much. Intel also spends billions of dollars on research and what they end up doing is almost always not worth a patent, just an improvement, but it's still worth it for them because of their huge market that pays it off and because they get ahead of the competition.

                      The idea that the world development would stop because of lack of patents is propaganda BS, in that there's a little truth to it but mostly BS and demagogy. Just like with petroleum where many (paid) experts said that if oil prices go down a lot - the world economy would crumble, but after the oil prices went down what really happened was that a few countries (like Russia) had issues but within a few years (except Venezuela) pretty much everybody is doing fine (including Russia) and their current issues are not because of low oil prices.

                      The whole idea that the world invents thanks to patents is BULL just like "fighting for democracy", in that it's a powerful tool to attack enemy countries under the pretext that they're not democratic and never attack your friends like Saudi Arabia who are very oppressive regimes.
                      Last edited by cl333r; 30 September 2017, 01:39 PM.

                      Comment

                      Working...
                      X