• How to compile .cl file that contains inline assembly for GCN cards?

    There are some examples of inline assembly inside .cl file: LLVM-AMDGPU-Assembler-Extra/s_memrealtime_inline.cl at master · ROCm-Developer-Tools/LLVM-AMDGPU-Assembler-Extra · GitHu… gatelessgate/equ...
    fancyix
    last modified by fancyix
  • OpenCL linker hangs & terminates application on R9 200

    After shipping our application, some users with AMD R9 200 series cards report the application hangs up and then quits. After studying log files and minidumps it seems the issue is the with OpenCL linker on those syst...
    george72
    last modified by george72
  • floating point precision

    Hi, In my current kernel I use floating point values, but I think I got precision problems... by example, I use an 'epsilon' defined as #define EPSILON 1e-4 and play with values like 500.f etc.... I'm not sure how...
    spectral
    last modified by spectral
  • Three cheers for anonymous AMD engineer that fixed OpenCL driver bug

    My app makes very heavy use of OpenCL events. For the longest time, the app was unstable - seemingly there was a race condition where event callbacks sometimes would not get called, causing my app to stop working. ...
    boxerab
    last modified by boxerab
  • Legacy OpenCL analysis not working with v18 drivers on Ubuntu

    Hi,   With any v18 amdgpu-pro driver, it is not possible to analyze Kernels for non-Vega ASICs   The error is always "Error: failed to disassemble binary output and produce textual ISA code for Hawaii (ke...
    greenstheorem
    last modified by greenstheorem
  • OpenCL compiler bug with big switches

    Hi. Recently I found an issue with AMD OpenCL compiler. I got OpenCL code that generated on CPU and results in big switch-case construction (about 4k case). Attempt to compile such kernel with AMD compiler leads to...
    kvalme
    last modified by kvalme
  • When will the AMDGPU-PRO driver support HD7900 series?

    My card is HD7990. And my latest amdgpu-pro driver 17.50 on Linux doesn't support my card, and it exports a lot of errors for all tensorflow tests. Also, the famous hello world opencl program doesn't work.   To ...
    karlcauchy
    last modified by karlcauchy
  • OpenCL & Linux on Raven Ridge

    Hello, This is more of a driver feature request but it is OpenCL specific. I've purchased a month ago a Riven Ridge APU (Ryzen 2400G) system and I'm eager to use OpenCL on its integrated Vega GPU. Unfortunately, it s...
    ekondis
    last modified by ekondis
  • OpenCL SDK for AMD EPYC and Hawaii architecture GPUs

    I am trying to setup a OpenCL environment for the following hardwares and operating system. CPU : AMD EPYC 7551P GPU : AMD FirePro S9150 (Hawaii) OS : CentOS 7.4   I've succesfully installed AMDGPU-PRO for th...
    ep-98d
    last modified by ep-98d
  • Various CL faults with vega on windows...

    I have a 290X and vega64 in the same system, the v64 is the primary card, however in the context of openCL it's device '1', with the 290X being '0'. This is all running windows 10 pro 64 with the latest 18.3.4 drivers...
    paul17041993
    last modified by paul17041993
  • Bug in AMD driver

    Hello community,   First, about me: I'm a student from Germany. I am studying Computational Engineering and I write GPU-accelerated numerical software.   With the current AMD Radeon 18.2.1 drivers, there ...
    robin.christ@gmx.de
    last modified by robin.christ@gmx.de
  • OpenCL kernel timeout on Ubuntu 14.04 (Crimson Edition 15.12) of 300 to 400 seconds

    Heyho,   i have some kernel timeout of about 300 to 400 seconds on AMD GCN devices (HD 7750 and R9 Fury X) with Ubuntu 14.04 and Crimson Edition 15.12 driver (fglrx-core). Does anybody know if there is an known...
    smato2018
    last modified by smato2018
  • vector data types on AMD GPUs

    First of all hi to everyone!, I am physicist and I am currently developing a Monte Carlo code for particle transport using OpenCL.   My question is regarding the convenience of using vector data types (like floa...
    endoerner
    last modified by endoerner
  • Mesa OpenCL with RX Vega support for OpenSUSE Tumbleweed

    Some days ago, I successfully ran the Mesa-OpenCL on my OpenSUSE Tumbleweed. Because, an OpenSUSE Tumbleweed still doesn't provide new LLVM 6.0 needed to support RX VEGA on Mesa-OpenCL, I wrote small tutorial to do it...
    matszpk
    last modified by matszpk
  • Pass value from device to host during kernel execution

    I would like to print a progress bar for my OpenCL code during the kernel execution. My CUDA equivalent of this code was able to achieve this using pinned memory, I was trying to implement the same using CL_MEM_ALLOC_...
    FangQ
    last modified by FangQ
  • Cannot Get OpenCL on Linux to Work At All

    I have a Vega 56. I downloaded the 17.50 driver, and installed it using: amdgpu-pro-install --compute   # rpm -qa | grep amdgpu | sort amdgpu-core-17.50-543815.el7.noarch amdgpu-dkms-17.50-543815.el7.noarch ...
    powerload
    last modified by powerload
  • Minimal OpenCL SDK in Vcpkg

    For those developing on Windows, I have submitted an OpenCL SDK package for Vcpkg, an invaluable tool for cross-platform developers living on Windows. The set of clMath libraries are also on the way. (clFFT and clBLAS...
    Meteorhead
    created by Meteorhead
  • OpenCL: How to allocate C++ vector in SVM memory?

    Hi,   How to allocate SVM memory and use it for a C++ vector? I guess vector.data() needs to be at SVM memory. Are atomic_svm_vector fine_svm_vector definitions in OpenCL C++ bindings something that helps to do ...
    sampie
    last modified by sampie
  • clBuildProgram crashes with LLVM error

    Hey, i am trying to build a simple kernel (see attachments) to test some implementation of a random number generator. However calling into clBuildProgram everything crashes with the following error message:   LL...
    pangea
    last modified by pangea
  • Double-precision atomic add support?

    Dear Forum, I'm wondering if currently (2018 March) any AMD GPU has hardware support for double-precision atomic addition in OpenCL. I'm not referring to the software solution where you can just emulate double-precis...
    biubiuty
    last modified by biubiuty