Hi,

my computer (Windows 10, FX-9590 and three ATI HD 4870 in crossfire) has problems with GPU work units. The work units start and need approximately three to four minutes to finish, but the result is a computation error. I tried to run a single work unit only and also disabled crossfire, but that didn't help.

Here is the link to a work unit I ran yesterday:

<core_client_version>7.16.20</core_client_version>
<![CDATA[
<message>
Unzul�ssige Funktion.
 (0x1) - exit code 1 (0x1)</message>
<stderr_txt>
<search_application> milkyway_separation 1.46 Windows x86 double OpenCL </search_application>
Reading preferences ended prematurely
BOINC GPU type suggests using OpenCL vendor 'Advanced Micro Devices, Inc.'
Error loading Lua script 'astronomy_parameters.txt': [string "number_parameters: 4..."]:1: '<name>' expected near '4' 
Switching to Parameter File 'astronomy_parameters.txt'
<number_WUs> 5 </number_WUs>
<number_params_per_WU> 20 </number_params_per_WU>
Using SSE4.1 path
Found 1 platform
Platform 0 information:
  Name:       AMD Accelerated Parallel Processing
  Version:    OpenCL 1.2 AMD-APP (937.2)
  Vendor:     Advanced Micro Devices, Inc.
  Extensions: cl_khr_icd cl_amd_event_callback cl_amd_offline_devices cl_khr_d3d10_sharing
  Profile:    FULL_PROFILE
Using device 0 on platform 0
Found 3 CL devices
Device 'ATI RV770' (Advanced Micro Devices, Inc.:0x1002) (CL_DEVICE_TYPE_GPU)
Board: ATI Radeon HD 4800 Series
Driver version:      CAL 1.4.1734
Version:             OpenCL 1.0 AMD-APP (937.2)
Compute capability:  0.0
Max compute units:   10
Clock frequency:     770 Mhz
Global mem size:     536870912
Local mem size:      16384
Max const buf size:  65536
Double extension:    cl_khr_fp64
Build log:
--------------------------------------------------------------------------------
LOOP UNROLL: pragma unroll (line 318)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 310)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 303)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 274)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 216)
    Unrolled as requested!

--------------------------------------------------------------------------------
Estimated AMD GPU GFLOP/s: 1232 SP GFLOP/s, 246 DP FLOP/s
Using a target frequency of 60.0
Using a block size of 2560 with 54 blocks/chunk
Using clWaitForEvents() for polling (mode -1)
Range:          { nu_steps = 320, mu_steps = 800, r_steps = 700 }
Iteration area: 560000
Chunk estimate: 4
Num chunks:     5
Chunk size:     138240
Added area:     131200
Effective area: 691200
Initial wait:   14 ms
Integration time: 42.083403 s. Average time per iteration = 131.510636 ms
Integral 0 time = 42.517566 s
Failed to calculate integral 0
Failed to calculate likelihood
Using SSE4.1 path
Found 1 platform
Platform 0 information:
  Name:       AMD Accelerated Parallel Processing
  Version:    OpenCL 1.2 AMD-APP (937.2)
  Vendor:     Advanced Micro Devices, Inc.
  Extensions: cl_khr_icd cl_amd_event_callback cl_amd_offline_devices cl_khr_d3d10_sharing
  Profile:    FULL_PROFILE
Using device 0 on platform 0
Found 3 CL devices
Device 'ATI RV770' (Advanced Micro Devices, Inc.:0x1002) (CL_DEVICE_TYPE_GPU)
Board: ATI Radeon HD 4800 Series
Driver version:      CAL 1.4.1734
Version:             OpenCL 1.0 AMD-APP (937.2)
Compute capability:  0.0
Max compute units:   10
Clock frequency:     770 Mhz
Global mem size:     536870912
Local mem size:      16384
Max const buf size:  65536
Double extension:    cl_khr_fp64
Build log:
--------------------------------------------------------------------------------
LOOP UNROLL: pragma unroll (line 318)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 310)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 303)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 274)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 216)
    Unrolled as requested!

--------------------------------------------------------------------------------
Estimated AMD GPU GFLOP/s: 1232 SP GFLOP/s, 246 DP FLOP/s
Using a target frequency of 60.0
Using a block size of 2560 with 54 blocks/chunk
Using clWaitForEvents() for polling (mode -1)
Range:          { nu_steps = 320, mu_steps = 800, r_steps = 700 }
Iteration area: 560000
Chunk estimate: 4
Num chunks:     5
Chunk size:     138240
Added area:     131200
Effective area: 691200
Initial wait:   14 ms
Integration time: 42.047935 s. Average time per iteration = 131.399796 ms
Integral 0 time = 42.481479 s
Failed to calculate integral 0
Failed to calculate likelihood
Using SSE4.1 path
Found 1 platform
Platform 0 information:
  Name:       AMD Accelerated Parallel Processing
  Version:    OpenCL 1.2 AMD-APP (937.2)
  Vendor:     Advanced Micro Devices, Inc.
  Extensions: cl_khr_icd cl_amd_event_callback cl_amd_offline_devices cl_khr_d3d10_sharing
  Profile:    FULL_PROFILE
Using device 0 on platform 0
Found 3 CL devices
Device 'ATI RV770' (Advanced Micro Devices, Inc.:0x1002) (CL_DEVICE_TYPE_GPU)
Board: ATI Radeon HD 4800 Series
Driver version:      CAL 1.4.1734
Version:             OpenCL 1.0 AMD-APP (937.2)
Compute capability:  0.0
Max compute units:   10
Clock frequency:     770 Mhz
Global mem size:     536870912
Local mem size:      16384
Max const buf size:  65536
Double extension:    cl_khr_fp64
Build log:
--------------------------------------------------------------------------------
LOOP UNROLL: pragma unroll (line 318)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 310)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 303)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 274)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 216)
    Unrolled as requested!

--------------------------------------------------------------------------------
Estimated AMD GPU GFLOP/s: 1232 SP GFLOP/s, 246 DP FLOP/s
Using a target frequency of 60.0
Using a block size of 2560 with 54 blocks/chunk
Using clWaitForEvents() for polling (mode -1)
Range:          { nu_steps = 320, mu_steps = 800, r_steps = 700 }
Iteration area: 560000
Chunk estimate: 4
Num chunks:     5
Chunk size:     138240
Added area:     131200
Effective area: 691200
Initial wait:   14 ms
Integration time: 42.034017 s. Average time per iteration = 131.356303 ms
Integral 0 time = 42.461353 s
Failed to calculate integral 0
Failed to calculate likelihood
Using SSE4.1 path
Found 1 platform
Platform 0 information:
  Name:       AMD Accelerated Parallel Processing
  Version:    OpenCL 1.2 AMD-APP (937.2)
  Vendor:     Advanced Micro Devices, Inc.
  Extensions: cl_khr_icd cl_amd_event_callback cl_amd_offline_devices cl_khr_d3d10_sharing
  Profile:    FULL_PROFILE
Using device 0 on platform 0
Found 3 CL devices
Device 'ATI RV770' (Advanced Micro Devices, Inc.:0x1002) (CL_DEVICE_TYPE_GPU)
Board: ATI Radeon HD 4800 Series
Driver version:      CAL 1.4.1734
Version:             OpenCL 1.0 AMD-APP (937.2)
Compute capability:  0.0
Max compute units:   10
Clock frequency:     770 Mhz
Global mem size:     536870912
Local mem size:      16384
Max const buf size:  65536
Double extension:    cl_khr_fp64
Build log:
--------------------------------------------------------------------------------
LOOP UNROLL: pragma unroll (line 318)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 310)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 303)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 274)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 216)
    Unrolled as requested!

--------------------------------------------------------------------------------
Estimated AMD GPU GFLOP/s: 1232 SP GFLOP/s, 246 DP FLOP/s
Using a target frequency of 60.0
Using a block size of 2560 with 54 blocks/chunk
Using clWaitForEvents() for polling (mode -1)
Range:          { nu_steps = 320, mu_steps = 800, r_steps = 700 }
Iteration area: 560000
Chunk estimate: 4
Num chunks:     5
Chunk size:     138240
Added area:     131200
Effective area: 691200
Initial wait:   14 ms
Integration time: 42.032976 s. Average time per iteration = 131.353051 ms
Integral 0 time = 42.447561 s
Failed to calculate integral 0
Failed to calculate likelihood
Using SSE4.1 path
Found 1 platform
Platform 0 information:
  Name:       AMD Accelerated Parallel Processing
  Version:    OpenCL 1.2 AMD-APP (937.2)
  Vendor:     Advanced Micro Devices, Inc.
  Extensions: cl_khr_icd cl_amd_event_callback cl_amd_offline_devices cl_khr_d3d10_sharing
  Profile:    FULL_PROFILE
Using device 0 on platform 0
Found 3 CL devices
Device 'ATI RV770' (Advanced Micro Devices, Inc.:0x1002) (CL_DEVICE_TYPE_GPU)
Board: ATI Radeon HD 4800 Series
Driver version:      CAL 1.4.1734
Version:             OpenCL 1.0 AMD-APP (937.2)
Compute capability:  0.0
Max compute units:   10
Clock frequency:     770 Mhz
Global mem size:     536870912
Local mem size:      16384
Max const buf size:  65536
Double extension:    cl_khr_fp64
Build log:
--------------------------------------------------------------------------------
LOOP UNROLL: pragma unroll (line 318)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 310)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 303)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 274)
    Unrolled as requested!
LOOP UNROLL: pragma unroll (line 216)
    Unrolled as requested!

--------------------------------------------------------------------------------
Estimated AMD GPU GFLOP/s: 1232 SP GFLOP/s, 246 DP FLOP/s
Using a target frequency of 60.0
Using a block size of 2560 with 54 blocks/chunk
Using clWaitForEvents() for polling (mode -1)
Range:          { nu_steps = 320, mu_steps = 800, r_steps = 700 }
Iteration area: 560000
Chunk estimate: 4
Num chunks:     5
Chunk size:     138240
Added area:     131200
Effective area: 691200
Initial wait:   14 ms
Integration time: 42.038542 s. Average time per iteration = 131.370445 ms
Integral 0 time = 42.463968 s
Failed to calculate integral 0
Failed to calculate likelihood
18:57:41 (1152): called boinc_finish(1)

</stderr_txt>
]]>

Does anybody know why these errors are happening?

Hi,

my computer (Windows 10, FX-9590 and three ATI HD 4870 in crossfire) has problems with GPU work units. The work units start and need approximately three to four minutes to finish, but the result is a computation error. I tried to run a single work unit only and also disabled crossfire, but that didn't help.

Does anybody know why these errors are happening?

First crossfire is useless to Boinc as you cannot use multiple gpu's to crunch a single gpu task.
Second your gpu's are OPEN CL 1.0 and I think Milky?Way has a minimum of 2.0 right now.

Second your gpu's are OPEN CL 1.0 and I think Milky?Way has a minimum of 2.0 right now.

No, Open CL 1.0 is fine, but Radeon HD 4000 series isn't supported because of missing "the full set of cl_khr_fp64" or because "the hardware only supports a half-baked OpenCL 1.0 implementation" or something like that. Not sure if there are any other projects supporting those GPUs than Moo! Wrapper, which still offers ATI CAL app.

---

Yes, I know that crossfire is useless to BOINC and all three cards got assigned a separate work unit that all had this error as result. I mentioned it, because that is the way the GPUs are configured right now and I was actually surprised that they even were recognized by BOINC/MilkyWay, because I had issues in the past.

Yes, it seems that the HD 4000 series is not supported anymore. When I got my first 4870 ten years ago it definitely worked and I remember that not too long ago some people claimed that their cards still can do work for MilkyWay, but apparently something has changed since then.

Thanks to both of you for your answers.