AMD's ROCm 1.4 Now Available With OpenCL Support

[phoronix]

Phoronix: AMD's ROCm 1.4 Now Available With OpenCL Support

The Radeon Open Compute platform has been updated and quietly released prior to the weekend. The ROCm 1.4 release comes with preliminary OpenCL support...

http://www.phoronix.com/scan.php?page=news_item&px=ROCm-1.4-Available

[darkbasic]

It still needs closed source components ---> useless.

[Min1123]

So, leave that alone for a moment, it requires PCI-e atomics, which apparently only exist in certain Xeons, certain v3 Core series, and only Ryzen and Naples AMD CPUs, which haven't been released yet at the time of this post. So, if you don't have a Broadwell-based server chip, or aren't in AMD or Cavium's inner circle testing the chips, this isn't going to work.

[RussianNeuroMancer]

Hopefully AMD will make amdgpu-pro available via packages.amd.com someday.

[bridgman]

It still needs closed source components ---> useless.

It was advertised as a closed-source developer preview running in parallel with the remaining open sourcing effort. How does that make it useless ?

[Marc Driftmeyer]

It was advertised as a closed-source developer preview running in parallel with the remaining open sourcing effort. How does that make it useless ?

It's useless unless you use the sanctioned Ubuntu 4.7 blessed Kernel and a few other tag along options which I know no one who is going to rebuild a system [outside of Phoronix whose livelihood seems to be a perpetual state of testing] to use AMDGPU-Pro.

Seeing as OpenCL is broken on the Mesa stack and libclc [having tested LLVM 3.9.1 and Debian rebuilt Mesa] against Blender 2.78a and Luxrender 1.6 with the following erroneous error that Debian refuses to test further. The work Stellar did to add those 2 lousy lines of Code to recognize Polaris based GPGPUs was for naught with libclc throwing errors: a project Mesa depends on and has been languishing since mid September 2016.

For reference the version of libclc in use: libclc-amdgcn (0.2.0+git20160907-3)

Pardon the long dump of info:

evice init success
Compiling OpenCL program base
OpenCL build failed with error CL_BUILD_PROGRAM_FAILURE, errors in console.
OpenCL program base build output: /usr/lib/blender/2.78/scripts/addons/cycles/kernel/kernels/opencl/../../kernel_compat_opencl.h:134:9: warning: 'NULL' macro redefined
/usr/include/clc/clctypes.h:89:9: note: previous definition is here
/usr/lib/blender/2.78/scripts/addons/cycles/kernel/kernels/opencl/../../util_math.h:1426:14: warning: implicit declaration of function 'lgamma' is invalid in C99
/usr/lib/blender/2.78/scripts/addons/cycles/kernel/kernels/opencl/../../kernel_montecarlo.h:41:14: warning: double precision constant requires cl_khr_fp64, casting to single precision
/usr/lib/blender/2.78/scripts/addons/cycles/kernel/kernels/opencl/../../util_math.h:56:26: note: expanded from macro 'M_2PI_F'
/usr/lib/blender/2.78/scripts/addons/cycles/kernel/kernels/opencl/../../kernel_montecarlo.h:75:14: warning: double precision constant requires cl_khr_fp64, casting to single precision
/usr/lib/blender/2.78/scripts/addons/cycles/kernel/kernels/opencl/../../util_math.h:56:26: note: expanded from macro 'M_2PI_F'
/usr/lib/blender/2.78/scripts/addons/cycles/kernel/kernels/opencl/../../kernel_montecarlo.h:92:14: warning: double precision constant requires cl_khr_fp64, casting to single precision
/usr/lib/blender/2.78/scripts/addons/cycles/kernel/kernels/opencl/../../util_math.h:56:26: note: expanded from macro 'M_2PI_F'
/usr/lib/blender/2.78/scripts/addons/cycles/kernel/kernels/opencl/../../kernel_montecarlo.h:107:14: warning: double precision constant requires cl_khr_fp64, casting to single precision
/usr/lib/blender/2.78/scripts/addons/cycles/kernel/kernels/opencl/../../util_math.h:56:26: note: expanded from macro 'M_2PI_F'
/usr/lib/blender/2.78/scripts/addons/cycles/kernel/kernels/opencl/../../kernel_projection.h:78:62: warning: double precision constant requires cl_khr_fp64, casting to single precision
/usr/lib/blender/2.78/scripts/addons/cycles/kernel/kernels/opencl/../../util_math.h:56:26: note: expanded from macro 'M_2PI_F'
/usr/lib/blender/2.78/scripts/addons/cycles/kernel/kernels/opencl/../../kernel_projection.h:83:63: warning: double precision constant requires cl_khr_fp64, casting to single precision
/usr/lib/blender/2.78/scripts/addons/cycles/kernel/kernels/opencl/../../util_math.h:56:26: note: expanded from macro 'M_2PI_F'
/usr/lib/blender/2.78/scripts/addons/cycles/kernel/kernels/opencl/../../closure/alloc.h:65:5: error: invalid argument type 'ShaderClosure *' (aka 'struct ShaderClosure *') to unary expression
/usr/lib/blender/2.78/scripts/addons/cycles/kernel/kernels/opencl/../../closure/../closure/bsdf_microfacet.h:62:21: warning: double precision constant requires cl_khr_fp64, casting to single precision
/usr/lib/blender/2.78/scripts/addons/cycles/kernel/kernels/opencl/../../util_math.h:56:26: note: expanded from macro 'M_2PI_F'
/usr/lib/blender/2.78/scripts/addons/cycles/kernel/kernels/opencl/../../closure/../closure/bsdf_microfacet.h:145:21: warning: double precision constant requires cl_khr_fp64, casting to single precision
/usr/lib/blender/2.78/scripts/addons/cycles/kernel/kernels/opencl/../../util_math.h:56:26: note: expanded from macro 'M_2PI_F'
/usr/lib/blender/2.78/scripts/addons/cycles/kernel/kernels/opencl/../../closure/../closure/bsdf_microfacet.h:269:11: error: invalid argument type 'MicrofacetExtra *' (aka 'struct MicrofacetExtra *') to unary expression
/usr/lib/blender/2.78/scripts/addons/cycles/kernel/kernels/opencl/../../closure/../closure/bsdf_microfacet.h:269:29: error: invalid argument type 'MicrofacetExtra *' (aka 'struct MicrofacetExtra *') to unary expression
/usr/lib/blender/2.78/scripts/addons/cycles/kernel/kernels/opencl/../../closure/../closure/bsdf_microfacet_multi.h:47:21: warning: double precision constant requires cl_khr_fp64, casting to single precision
/usr/lib/blender/2.78/scripts/addons/cycles/kernel/kernels/opencl/../../util_math.h:56:26: note: expanded from macro 'M_2PI_F'
/usr/lib/blender/2.78/scripts/addons/cycles/kernel/kernels/opencl/../../closure/../closure/bsdf_ashikhmin_shirley.h:136:52: warning: implicit declaration of function 'native_tan' is invalid in C99
/usr/lib/blender/2.78/scripts/addons/cycles/kernel/kernels/opencl/../../kernel_compat_opencl.h:116:19: note: expanded from macro 'tanf'
/usr/lib/blender/2.78/scripts/addons/cycles/kernel/kernels/opencl/../../closure/../closure/bsdf_ashikhmin_shirley.h:167:10: warning: double precision constant requires cl_khr_fp64, casting to single precision
/usr/lib/blender/2.78/scripts/addons/cycles/kernel/kernels/opencl/../../util_math.h:56:26: note: expanded from macro 'M_2PI_F'
/usr/lib/blender/2.78/scripts/addons/cycles/kernel/kernels/opencl/../../svm/svm_gradient.h:46:25: warning: double precision constant requires cl_khr_fp64, casting to single precision
/usr/lib/blender/2.78/scripts/addons/cycles/kernel/kernels/opencl/../../util_math.h:56:26: note: expanded from macro 'M_2PI_F'
/usr/lib/blender/2.78/scripts/addons/cycles/kernel/kernels/opencl/../../svm/svm_closure.h:295:64: warning: double precision constant requires cl_khr_fp64, casting to single precision
/usr/lib/blender/2.78/scripts/addons/cycles/kernel/kernels/opencl/../../util_math.h:56:26: note: expanded from macro 'M_2PI_F'
/usr/lib/blender/2.78/scripts/addons/cycles/kernel/kernels/opencl/../../svm/svm_wave.h:37:8: warning: double precision constant requires cl_khr_fp64, casting to single precision
/usr/lib/blender/2.78/scripts/addons/cycles/kernel/kernels/opencl/../../util_math.h:56:26: note: expanded from macro 'M_2PI_F'
/usr/lib/blender/2.78/scripts/addons/cycles/kernel/kernels/opencl/../../svm/svm_sky.h:105:32: warning: double precision constant requires cl_khr_fp64, casting to single precision
/usr/lib/blender/2.78/scripts/addons/cycles/kernel/kernels/opencl/../../util_math.h:56:26: note: expanded from macro 'M_2PI_F'

Caching exr file, 1920x1080, /home/mdriftmeyer/Blender/Render-Cache/cached_RR_showerGlass_Scene_6977eba0dfffb5ff4cc8fd123d14ac62.exr
Error: Failed loading render kernel, see console for errors


CLINFO:

mdriftmeyer@horus:~/Downloads$ clinfo
Number of platforms                               1
  Platform Name                                   Clover
  Platform Vendor                                 Mesa
  Platform Version                                OpenCL 1.1 Mesa 13.0.2
  Platform Profile                                FULL_PROFILE
  Platform Extensions                             cl_khr_icd
  Platform Extensions function suffix             MESA

  Platform Name                                   Clover
Number of devices                                 1
  Device Name                                     AMD POLARIS10 (DRM 3.3.0 / 4.8.0-2-amd64, LLVM 3.9.1)
  Device Vendor                                   AMD
  Device Vendor ID                                0x1002
  Device Version                                  OpenCL 1.1 Mesa 13.0.2
  Driver Version                                  13.0.2
  Device OpenCL C Version                         OpenCL C 1.1
  Device Type                                     GPU
  Device Profile                                  FULL_PROFILE
  Max compute units                               36
  Max clock frequency                             0MHz
  Max work item dimensions                        3
  Max work item sizes                             256x256x256
  Max work group size                             256
[B]=== CL_PROGRAM_BUILD_LOG ===
./generic/lib/workitem/get_global_id.cl:4:30: in function sum void (float addrspace(1)*, float addrspace(1)*, float addrspace(1)*): unsupported call to function get_local_size
  Preferred work group size multiple              ./generic/lib/workitem/get_global_id.cl:4:30: in function sum void (float addrspace(1)*, float addrspace(1)*, float addrspace(1)*): unsupported call to function get_local_size[/B]

  Preferred / native vector sizes                 
    char                                                16 / 16      
    short                                                8 / 8       
    int                                                  4 / 4       
    long                                                 2 / 2       
    half                                                 0 / 0        (n/a)
    float                                                4 / 4       
    double                                               2 / 2        (cl_khr_fp64)
  Half-precision Floating-point support           (n/a)
  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                              8587284480 (7.998GiB)
  Error Correction support                        No
  Max memory allocation                           7728556032 (7.198GiB)
  Unified memory for Host and Device              Yes
  Minimum alignment for any data type             128 bytes
  Alignment of base address                       1024 bits (128 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_global_int32_base_atomics cl_khr_global_int32_extended_atomics cl_khr_local_int32_base_atomics cl_khr_local_int32_extended_atomics cl_khr_byte_addressable_store cl_khr_fp64

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                                   AMD POLARIS10 (DRM 3.3.0 / 4.8.0-2-amd64, LLVM 3.9.1)
  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                                   AMD POLARIS10 (DRM 3.3.0 / 4.8.0-2-amd64, LLVM 3.9.1)

ICD loader properties
  ICD loader Name                                 OpenCL ICD Loader
  ICD loader Vendor                               OCL Icd free software
  ICD loader Version                              2.2.10
  ICD loader Profile                              OpenCL 2.1
mdriftmeyer@horus:~/Downloads$

I am hardly convinced that ROCm will `just work' with the FOSS status currently hosed.

[finalzone]

Does the new version of rocm handle newer Linux kernel?
Has it been tested on hybrid laptop i.e. AMD Radeon M265DX?
Edit: I decide to install rocm anyway and encounter this issue while following the instruction

$ sudo dnf install rocm --allowerasing
Last metadata expiration check: 0:00:42 ago on Mon Dec 19 09:32:16 2016.
Error: package rocm-1.4.0-0.x86_64 requires rocm-kernel = 1.4.0, but none of the providers can be installed

It looks like rocm-kernel is missing.

[whitecat]

which apparently only exist in certain Xeons, certain v3 Core series

They said this is available on all Haswell+ CPU (except lower-end, like Pentium and Celeron).

Anyway, this is useless at least for one case: crypto currency miner. In this use case, people are using mostly PCI-Ex 1x riser so they do not plug to the CPU PCI-Ex root port.

[bridgman]

Does the new version of rocm handle newer Linux kernel?
Has it been tested on hybrid laptop i.e. AMD Radeon M265DX?
Edit: I decide to install rocm anyway and encounter this issue while following the instruction

$ sudo dnf install rocm --allowerasing
Last metadata expiration check: 0:00:42 ago on Mon Dec 19 09:32:16 2016.
Error: package rocm-1.4.0-0.x86_64 requires rocm-kernel = 1.4.0, but none of the providers can be installed

It looks like rocm-kernel is missing.

It supports the distros listed in the README with their stock kernels - IIRC that is Ubuntu 16.04, 16.10 plus Fedora 23.

The M265 is an SI chip with no MEC block, so ROCM is not going to work on it.

[finalzone]

It supports the distros listed in the README with their stock kernels - IIRC that is Ubuntu 16.04, 16.10 plus Fedora 23.

The M265 is an SI chip with no MEC block, so ROCM is not going to work on it.

Thanks for the reply. Just to let you know Fedora 23 is reaching EOL on December 20.