All of us know that the major PC-hardware companies (AMD&ATI, Intel, nVidia) release lots of software tools in order to aid to a software developers. This is good, but not enough, IMHO. In fact, quiet not every software developer is interested to optimize its program to a new hardware, because this takes lots of various resources. As result, we can see that almost all software developers do ignore optimizations, which can be achieved by using a modern technologies like last (and even first!) SSE's, multi-core CPU, GPU-calculations and so on. Of course, there are several tools exist that can help software developers to optimize its program: very efficient compilers, CUDA etc.
But let me to expose yet another point of view. What if to negotiate with software developers to publish the sources of most calculation-intensive parts of their programs? You may ask me: what for? To optimize it by call upon the huge world community of enthusiasts or professionals. Let's imagine that some software company would publish such a part of program, then some enthusiasts or professionals could optimize it and release their work as libraries to integration into the program by using, for example, not yet existed Open Application Accelerator
. When the program runs, it can call the OpenAA and ask it for optimized libraries. If ones exist, the program will use them, if not, the program will use her own libraries.
This method could be the panacea for a software developers. They need just to publish a piece of code, that's all. All the work will be done by enthusiasts or professionals, whose will be free to use any technology they do like.
But let me to expose yet another point of view. What if to negotiate with software developers to publish the sources of most calculation-intensive parts of their programs? You may ask me: what for? To optimize it by call upon the huge world community of enthusiasts or professionals. Let's imagine that some software company would publish such a part of program, then some enthusiasts or professionals could optimize it and release their work as libraries to integration into the program by using, for example, not yet existed Open Application Accelerator
. When the program runs, it can call the OpenAA and ask it for optimized libraries. If ones exist, the program will use them, if not, the program will use her own libraries. This method could be the panacea for a software developers. They need just to publish a piece of code, that's all. All the work will be done by enthusiasts or professionals, whose will be free to use any technology they do like.
No, I meant something... else . Of course, it is very good - to release several software libraries and tools in order to help for software developers. The problem is in last ones. Almost all of them just do not want to do anything with software optimization! It is very time- and labor-intensive work, so it is obvious that no one wants to do this. My point is to allow to do this hard work by many single enthusiasts all over the world, but this could not be possible without an access to some parts of source code.
Let's imagine, for example, that one enthusiast claims that he is capable to optimize some kind of program. Even he is capable to do this, he couldn't prove it, until software developer will open an access to several part(s) of its program. I do summon for software developers to open CPU-intensive parts of their programs, because I do believe that many enthusiasts could optimize them.
. Of course, it is very good - to release several software libraries and tools in order to help for software developers. The problem is in last ones. Almost all of them just do not want to do anything with software optimization! It is very time- and labor-intensive work, so it is obvious that no one wants to do this. My point is to allow to do this hard work by many single enthusiasts all over the world, but this could not be possible without an access to some parts of source code. Let's imagine, for example, that one enthusiast claims that he is capable to optimize some kind of program. Even he is capable to do this, he couldn't prove it, until software developer will open an access to several part(s) of its program. I do summon for software developers to open CPU-intensive parts of their programs, because I do believe that many enthusiasts could optimize them.
