Not working with ARM/Adreno GPUs

[hvc3000]

Hi Folks,

i tried to compile nws (nano-work-server) for ARM, this mostly works without a problem. The integrated NEON SIMD of Cortex CPUs with ISA ARMv8-A and above works well with nws and computes work in an appropiate time. In my oppinion this is a great part reducing electrical energy consuption and wasting for generating PoW.

I was wondering while ARM Mali GPUs like T- and G-family or Quallcom Adreno GPUs are OpenCL 1.2 and 2.0 compatible but can't compute nws work even getting loaded/recognized as an OpenCL device.

As an Example from a Mali-T830 GPU at termux, where nws always works with CPU:

$ clinfo
Number of platforms                               1
  Platform Name                                   ARM Platform
  Platform Vendor                                 ARM
  Platform Version                                OpenCL 1.2 v1.r12p1-04bet0.1bb9662be2ebee934dcbd7265c794a91
  Platform Profile                                FULL_PROFILE
  Platform 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_3d_image_writes cl_khr_fp64 cl_khr_int64_base_atomics cl_khr_int64_extended_atomics cl_khr_fp16 cl_khr_gl_sharing cl_khr_icd cl_khr_egl_event cl_khr_egl_image cl_arm_core_id cl_arm_printf cl_arm_thread_limit_hint cl_arm_non_uniform_work_group_size cl_arm_import_memory
  Platform Extensions function suffix             ARM

  Platform Name                                   ARM Platform
Number of devices                                 1
  Device Name                                     Mali-T830
  Device Vendor                                   ARM
  Device Vendor ID                                0x8301000
  Device Version                                  OpenCL 1.2 v1.r12p1-04bet0.1bb9662be2ebee934dcbd7265c794a91
  Driver Version                                  1.2
  Device OpenCL C Version                         OpenCL C 1.2 v1.r12p1-04bet0.1bb9662be2ebee934dcbd7265c794a91
  Device Type                                     GPU
  Device Available                                Yes
  Device Profile                                  FULL_PROFILE
  Max compute units                               2
  Max clock frequency                             360MHz
  Device Partition                                (core)
    Max number of sub-devices                     0
    Supported partition types                     None
  Max work item dimensions                        3
  Max work item sizes                             256x256x256
  Max work group size                             256
  Compiler Available                              Yes
  Linker Available                                Yes
  Preferred work group size multiple              4
  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)
    
(...and.so.on...)

So if you want to start nws (with small values):

./nano-work-server --gpu 0:0:256 --gpu-local-work-size 16 (in any combination of other flags and values)

it returns:

thread 'main' panicked at 'Platform::list: Error retrieving platform list: Unable to get platform id list after 10 seconds of waiting.', /data/data/com.termux/files/home/.cargo/registry/src/github.com-1ecc6299db9ec823/ocl-0.19.3/src/standard/platform.rs:50:14

For me this looks like ocl-0.19.3 does not support Mali/Adreno GPUs or others like Raspberry Pi 3 GPU with VC4CL despite of there OpenCL support.

Is it possible to add/change/modify/motherofgod the ocl-0.19.3 crate (or sth like it's platform identifier) to sth which can support GPUs from above?

Thanks for help and advise 💯