Announcement

Collapse
No announcement yet.

CUDA 9 Release Candidate Arrives, Supports Volta & Faster Performance

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

  • CUDA 9 Release Candidate Arrives, Supports Volta & Faster Performance

    Phoronix: CUDA 9 Release Candidate Arrives, Supports Volta & Faster Performance

    NVIDIA is ending out the week with their first release candidate of CUDA 9...

    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

  • #2
    I can't find anything on the matter but I hope they add support for GCC versions > 5.x and clang > 3.8.

    ​​​​

    Comment


    • #3
      from Release Notes

      ‣ CUDA now supports the Clang 3.9 LLVM-based C and C++ compilers as host compilers on Linux operating systems.

      snip

      ‣ The Intel C++ compiler 17.0 is now supported.

      snip

      ‣ The CUDA compiler now supports C++14 features.

      snip

      ‣ The CUDA compiler now supports the -std=c++03 C++ dialect mode. This mode was added to enable existing code to be compiled with gcc 6.0 or later or a recent Clang compiler in c++03 mode if necessary. The default mode for these compilers is c++14. An application might have to be built in c++03 mode because it is linking to a third-party library compiled in that mode.

      snip

      ‣ Microsoft Visual Studio 2017 (Update 2) is now supported

      snip

      ‣ GCC 6.x is now supported.
      ‣ PGI pcg++ 17 is now supported.
      ‣ Xcode 8.3.3 is now supported.
      Last edited by sl1pkn07; 04 August 2017, 10:56 AM.

      Comment


      • #4
        let's see :

        llvm 3.9 : dropped from archlinux repos on 2017-04-29
        gcc 6.x : dropped on 2017-03-09

        Nope, still a lot of work for those AL users that need CUDA .
        Why do some people make it so easy to stay away from their products ?

        Comment


        • #5
          Do you think nvidia cares about arch users?????? You're funny.

          Comment


          • #6
            Originally posted by LoneVVolf View Post
            let's see :

            llvm 3.9 : dropped from archlinux repos on 2017-04-29
            gcc 6.x : dropped on 2017-03-09

            Nope, still a lot of work for those AL users that need CUDA .
            Why do some people make it so easy to stay away from their products ?
            How is it a lot of work? I just installed CUDA the other day on Manjaro and used the package manager. I'd assume there isn't much difference if any at all for vanilla Arch?

            Comment


            • #7
              Apart from having annoying restrictions on the compiler I do hope they've fixed the 10% speed drop they introduced with version 8.

              Comment


              • #8
                Originally posted by LoneVVolf View Post
                let's see :

                llvm 3.9 : dropped from archlinux repos on 2017-04-29
                gcc 6.x : dropped on 2017-03-09

                Nope, still a lot of work for those AL users that need CUDA .
                Why do some people make it so easy to stay away from their products ?
                use a really distro, a stable one for develop, arch is for blend edge, bugs and nighmares you have Debian, Ubuntu, Fedora, Linux Mint , Opensuse leap for stable developing

                Comment


                • #9
                  ah these standards... don't know which gpu to buy for this reason.. nvidia with cuda or amd with opencl...

                  Comment


                  • #10
                    Originally posted by smartalgorithm View Post
                    ah these standards... don't know which gpu to buy for this reason.. nvidia with cuda or amd with opencl...
                    Or AMD with OpenCL and HIP/HCC:



                    AMD, a strong proponent of open source and open standards, has created a new tool that will allow developers to convert CUDA code to common C++. The resulting C++ code can run through either CUDA NVCC or AMD HCC compilers. This new Heterogeneous-compute Interface for Portability, or HIP, is a tool that provides customers with more choice in hardware and development tools.
                    Test signature

                    Comment

                    Working...
                    X