cancel
Showing results for 
Search instead for 
Did you mean: 

Drivers & Software

devlista
Adept I

Install pro driver with opencl on Debian Stretch? Also compiler flag quest.

Hello, I am the proud owner of a Rx 580 card and I'm on Debian 9 (stretch)

I've been using AMD hardware for my desktops for >20 years - going back to the Rage 128 on my power mac

I'm currently using the open source amdgpu stack via Debian 9 (stable) with backports enabled.

However I am getting errors in some programs, for example, Blender, LuxCoreRender, etc.. (output from clinfo is below)

I was unable to install the amdgpu-pro driver either, using the supplied script.

Since Ubuntu is based on Debian I figured that this couldn't possibly be that difficult but rather than trying to figure it out myself with possible impact on stability (my machine is more stable now than it had been in a long time) I thought I should ask you first..

I'd like to do whatever it takes so that the pro driver is installed, without problems, and openCL and the Blender export tools work.

I want to be able to use openCL accelerated rendering to render optically complex materials like glass, clear plastic and water.

Can that be done without "breaking Debian" as they say? I am an intermediate level user and can compile software, edit scripts, etc.

Also, unrelated, I wanted to ask what the best compiler optimization flag string was for compiling software for Bulldozer. I am still on a Bulldozer machine.

Here (below) is the clinfo output as it is returned now.

Thank you!

I am very happy with my purchases, generally. Your products offer excellent value.

-Chris

~$ clinfo
/usr/share/libdrm/amdgpu.ids version: 1.0.0
Number of platforms                               1
  Platform Name                                   Clover
  Platform Vendor                                 Mesa
  Platform Version                                OpenCL 1.1 Mesa 18.1.6
  Platform Profile                                FULL_PROFILE
  Platform Extensions                             cl_khr_icd
  Platform Extensions function suffix             MESA

  Platform Name                                   Clover
Number of devices                                 1
  Device Name                                     Radeon RX 580 Series (POLARIS10, DRM 3.25.0, 4.17.0-0.bpo.3-amd64, LLVM 6.0.0)
  Device Vendor                                   AMD
  Device Vendor ID                                0x1002
  Device Version                                  OpenCL 1.1 Mesa 18.1.6
  Driver Version                                  18.1.6
  Device OpenCL C Version                         OpenCL C 1.1
  Device Type                                     GPU
  Device Profile                                  FULL_PROFILE
  Max compute units                               36
  Max clock frequency                             1366MHz
  Max work item dimensions                        3
  Max work item sizes                             256x256x256
  Max work group size                             256
  Preferred work group size multiple              64
  Preferred / native vector sizes                
    char                                                16 / 16     
    short                                                8 / 8      
    int                                                  4 / 4      
    long                                                 2 / 2      
    half                                                 8 / 8        (cl_khr_fp16)
    float                                                4 / 4      
    double                                               2 / 2        (cl_khr_fp64)
  Half-precision Floating-point support           (cl_khr_fp16)
    Denormals                                     No
    Infinity and NANs                             Yes
    Round to nearest                              Yes
    Round to zero                                 No
    Round to infinity                             No
    IEEE754-2008 fused multiply-add               No
    Support is emulated in software               No
    Correctly-rounded divide and sqrt operations  No
  Single-precision Floating-point support         (core)
    Denormals                                     No
    Infinity and NANs                             Yes
    Round to nearest                              Yes
    Round to zero                                 No
    Round to infinity                             No
    IEEE754-2008 fused multiply-add               No
    Support is emulated in software               No
    Correctly-rounded divide and sqrt operations  No
  Double-precision Floating-point support         (cl_khr_fp64)
    Denormals                                     Yes
    Infinity and NANs                             Yes
    Round to nearest                              Yes
    Round to zero                                 Yes
    Round to infinity                             Yes
    IEEE754-2008 fused multiply-add               Yes
    Support is emulated in software               No
    Correctly-rounded divide and sqrt operations  No
  Address bits                                    64, Little-Endian
  Global memory size                              8589934592 (8GiB)
  Error Correction support                        No
  Max memory allocation                           6440266752 (5.998GiB)
  Unified memory for Host and Device              No
  Minimum alignment for any data type             128 bytes
  Alignment of base address                       32768 bits (4096 bytes)
  Global Memory cache type                        None
  Image support                                   No
  Local memory type                               Local
  Local memory size                               32768 (32KiB)
  Max constant buffer size                        2147483647 (2GiB)
  Max number of constant args                     16
  Max size of kernel argument                     1024
  Queue properties                               
    Out-of-order execution                        No
    Profiling                                     Yes
  Profiling timer resolution                      0ns
  Execution capabilities                         
    Run OpenCL kernels                            Yes
    Run native kernels                            No
  Device Available                                Yes
  Compiler Available                              Yes
  Device Extensions                               cl_khr_byte_addressable_store cl_khr_global_int32_base_atomics cl_khr_global_int32_extended_atomics cl_khr_local_int32_base_atomics cl_khr_local_int32_extended_atomics cl_khr_int64_base_atomics cl_khr_int64_extended_atomics cl_khr_fp64 cl_khr_fp16

NULL platform behavior
  clGetPlatformInfo(NULL, CL_PLATFORM_NAME, ...)  Clover
  clGetDeviceIDs(NULL, CL_DEVICE_TYPE_ALL, ...)   Success [MESA]
  clCreateContext(NULL, ...) [default]            Success [MESA]
  clCreateContextFromType(NULL, CL_DEVICE_TYPE_CPU)  No devices found in platform
  clCreateContextFromType(NULL, CL_DEVICE_TYPE_GPU)  Success (1)
    Platform Name                                 Clover
    Device Name                                   Radeon RX 580 Series (POLARIS10, DRM 3.25.0, 4.17.0-0.bpo.3-amd64, LLVM 6.0.0)
  clCreateContextFromType(NULL, CL_DEVICE_TYPE_ACCELERATOR)  No devices found in platform
  clCreateContextFromType(NULL, CL_DEVICE_TYPE_CUSTOM)  No devices found in platform
  clCreateContextFromType(NULL, CL_DEVICE_TYPE_ALL)  Success (1)
    Platform Name                                 Clover
    Device Name                                   Radeon RX 580 Series (POLARIS10, DRM 3.25.0, 4.17.0-0.bpo.3-amd64, LLVM 6.0.0)

ICD loader properties
  ICD loader Name                                 OpenCL ICD Loader
  ICD loader Vendor                               OCL Icd free software
  ICD loader Version                              2.2.11
  ICD loader Profile                              OpenCL 2.1

0 Likes
3 Replies
devlista
Adept I

This is a different problem it seems but quite similar to what's happening to me when I try to use the open source driver. Invariably Blender, LuxRender, etc, hang when trying to build the opencl kernel.

You could make a script available to properly set up the opencl drivers on other distros, then the packagers can use it as a starting point for their packagae

https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=857718

https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=848258https://developer.blender.org/T50522
0 Likes

I realized that an old version of AMDAPPSDK v 3 was installed and just removed it  That may help. Rebooting now.

If I can get Blender to work with the open source stack that would be enough for me right now.

I'm going to see if I can extract some debugging info from the crashes as described here:

https://bugs.freedesktop.org/show_bug.cgi?id=102009

0 Likes
devlista
Adept I

I just wanted to add here that I managed to resolve this problem by some basic stuff with the installer (force quitting it in order to do a partial install) - to get OpenCL 1.2 and 2 support and my graphics apps which depend on it working well on my Debian box for now,

But its ugly.

0 Likes