Created attachment 126862 [details]
OpenCLTest_02, some result are zero after a cell copyl

On the second sample file some results are zero without any apparent reason.

Copy E14 in E13
Columns M:O results are zero
Hard recalc
Results are calculated again

Maybe in relation with issue in comment#1, fails with the same version and works with the same versions.

I can't reproduce in a fresh build from the 5.2 branch.

I follow your steps:

1. Load the first attached document. M7:R250 are all zero. (I.e. the values present in the document file.)

2. Delete M250:R250 (Select that range, press the "Delete" key, keep the default tick boxes selected, i.e. Text, Date&Time, Numbers, and Formulas), press OK.

3. Hard recalc, i.e. press Ctrl+Shift+F9. Non-zero numbers appear here and there in M7:R249. Also some #VALUE! errors. This is apparently as expected? At least it matches what I get if I turn off OpenCL.

4. Copy M249:R249 to M250(:R250). M250:R250 also shows zeros, like M249:R249.

5. Hard recalc. ==> Nothing changes. <== This is different from what you claim.

In fact, I don't see what impact the deletion and copying would have on this; all that is needed to get correct results is a hard recalc after loading the document, as the values stored in the document are incorrect.

I did not look at comment #1 yet.

As for the document in comment #1, I can't reproduce the issue at all. Even if I turn off both OpenCL and the software interpreter, I still see just zeros in M13:O14 upon a hard recalc.

My opencl_devices.log below. What's noticeable is that I use a newer version of AMD's driver. Maybe the problem you are seeing has been fixed in a newer driver?

Device Index: 0
  Selected: true
  Device Name: Spectre
  Device Vendor: Advanced Micro Devices, Inc.
  Device Version: OpenCL 1.2 AMD-APP (2117.9)
  Driver Version: 2117.9 (VM)
  Device Type: gpu 
  Device Extensions: cl_khr_fp64 cl_amd_fp64 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_3d_image_writes cl_khr_byte_addressable_store cl_khr_gl_sharing cl_amd_device_attribute_query cl_amd_vec3 cl_amd_printf cl_amd_media_ops cl_amd_media_ops2 cl_amd_popcnt cl_khr_d3d10_sharing cl_khr_d3d11_sharing cl_khr_dx9_media_sharing cl_khr_image2d_from_buffer cl_khr_spir cl_khr_gl_event cl_amd_liquid_flash 
  Device OpenCL C Version: OpenCL C 1.2 
  Device Available: true
  Device Compiler Available: true
  Device Linker Available: true
  Platform Name: AMD Accelerated Parallel Processing
  Platform Vendor: Advanced Micro Devices, Inc.
  Platform Version: OpenCL 2.0 AMD-APP (2117.9)
  Platform Profile: FULL_PROFILE
  Platform Extensions: cl_khr_icd cl_khr_d3d10_sharing cl_khr_d3d11_sharing cl_khr_dx9_media_sharing cl_amd_event_callback cl_amd_offline_devices 

Device Index: 1
  Selected: false
  Device Name: AMD A10-7800 Radeon R7, 12 Compute Cores 4C+8G 
  Device Vendor: AuthenticAMD
  Device Version: OpenCL 1.2 AMD-APP (2117.9)
  Driver Version: 2117.9 (sse2,avx,fma4)
  Device Type: cpu 
  Device Extensions: cl_khr_fp64 cl_amd_fp64 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_3d_image_writes cl_khr_byte_addressable_store cl_khr_gl_sharing cl_ext_device_fission cl_amd_device_attribute_query cl_amd_vec3 cl_amd_printf cl_amd_media_ops cl_amd_media_ops2 cl_amd_popcnt cl_khr_d3d10_sharing cl_khr_spir cl_khr_gl_event 
  Device OpenCL C Version: OpenCL C 1.2 
  Device Available: true
  Device Compiler Available: true
  Device Linker Available: true
  Platform Name: AMD Accelerated Parallel Processing
  Platform Vendor: Advanced Micro Devices, Inc.
  Platform Version: OpenCL 2.0 AMD-APP (2117.9)
  Platform Profile: FULL_PROFILE
  Platform Extensions: cl_khr_icd cl_khr_d3d10_sharing cl_khr_d3d11_sharing cl_khr_dx9_media_sharing cl_amd_event_callback cl_amd_offline_devices

Sorry I'm a bit busy on the job, I'll test again this night, It's strange only 5.2.1 has both issues for me, working fine with 5.1.5 and master.

I think I have found the origin of the issue, after remember to test a clean profile. With so many test sometimes one forget the basics like to try a clean profile, what so often I recommend discard it.

So sorry for your time.

Doing the test with the clean profile 5.2.1 works fine again.

Looking on the expert configuration for "opencl" entries, the entry:
"org.openoffice.Office.Calc.Formula.Calculation.OpenCLSubsetOnly"
is 'false' in the bad profile, changing it to 'true' seems to solve the issue (after restart).

Sure I have it as false from a version where the option was available in the Calc/Formula/Custom-Details, not the case now with 5.2.1

Even with 5.1.5.2 the issue is there with the option disable.

Maybe it should be set up as 'true' always when installing of 5.2.1 (5.1.5?) and next versions to avoid inherit this option as false, and avoid some headache any others.

I find OpenCL a big improvement for calc, specially with large and complex spreadsheets, so what can be done to have it working, always nice.

Yep, that option was removed from the UI exactly because it was extremely unsafe;) Can we resolve this bug then, as WORKSFORME or NOTABUG?

I think better as not a bug.

OK, thanks.