Simple and extended python examples do not work (v 0.7.0)

[yves-surrel]

In Line 28 of the "simple" example, OpAlgoSyncLocal does not exist. I tried replacing by OpTensorSyncLocal, but the test fails as tensor_out.data() remains stuck to (0, 0, 0).

The "extended" example does not compile at line 8

 tensor_in_a = Tensor([2, 2, 2])

with

TypeError: kp.Tensor: No constructor defined!

At line #12

mgr.eval_tensor_create_def([tensor_in_a, tensor_in_b, tensor_out])

refers to a non-existent attribute (eval_tensor_create_def) of Manager, etc.

(MacBookPro 2017, MacOs 11.2.1)

[axsaucedo]

Thanks @yves-surrel - it seems the examples are outdated, I will have a look at this. Are you abel to run the python test files to see if those work? It's under python/test

[yves-surrel]

Here they are:

https://github.com/EthicalML/vulkan-kompute/blob/master/python/test/test_kompute.py

In [16]: test_end_to_end()

In [17]: test_shader_str()

In [18]: test_sequence()
WARNING:kp:Kompute Sequence destroy called with null Device pointer
WARNING:kp:Kompute Tensor destructor reached with null Device pointer
WARNING:kp:Kompute Tensor destructor reached with null Device pointer
WARNING:kp:Kompute Tensor destructor reached with null Device pointer

In [19]: test_pushconsts()

In [20]: test_workgroup()
[ 0.  0.  0.  0.  0.  0.  0.  0.  1.  1.  1.  1.  1.  1.  1.  1.  2.  2.
  2.  2.  2.  2.  2.  2.  3.  3.  3.  3.  3.  3.  3.  3.  4.  4.  4.  4.
  4.  4.  4.  4.  5.  5.  5.  5.  5.  5.  5.  5.  6.  6.  6.  6.  6.  6.
  6.  6.  7.  7.  7.  7.  7.  7.  7.  7.  8.  8.  8.  8.  8.  8.  8.  8.
  9.  9.  9.  9.  9.  9.  9.  9. 10. 10. 10. 10. 10. 10. 10. 10. 11. 11.
 11. 11. 11. 11. 11. 11. 12. 12. 12. 12. 12. 12. 12. 12. 13. 13. 13. 13.
 13. 13. 13. 13. 14. 14. 14. 14. 14. 14. 14. 14. 15. 15. 15. 15. 15. 15.
 15. 15.]
[0. 1. 2. 3. 4. 5. 6. 7. 0. 1. 2. 3. 4. 5. 6. 7. 0. 1. 2. 3. 4. 5. 6. 7.
 0. 1. 2. 3. 4. 5. 6. 7. 0. 1. 2. 3. 4. 5. 6. 7. 0. 1. 2. 3. 4. 5. 6. 7.
 0. 1. 2. 3. 4. 5. 6. 7. 0. 1. 2. 3. 4. 5. 6. 7. 0. 1. 2. 3. 4. 5. 6. 7.
 0. 1. 2. 3. 4. 5. 6. 7. 0. 1. 2. 3. 4. 5. 6. 7. 0. 1. 2. 3. 4. 5. 6. 7.
 0. 1. 2. 3. 4. 5. 6. 7. 0. 1. 2. 3. 4. 5. 6. 7. 0. 1. 2. 3. 4. 5. 6. 7.
 0. 1. 2. 3. 4. 5. 6. 7.]

In [21]: test_mgr_utils()

In [22]: 

https://github.com/EthicalML/vulkan-kompute/blob/master/python/test/test_array_multiplication.py
pass

https://github.com/EthicalML/vulkan-kompute/blob/master/python/test/test_logistic_regression.py
pass

https://github.com/EthicalML/vulkan-kompute/blob/master/python/test/test_tensor_types.py

In [3]: test_type_float()

In [4]: test_type_float_double_incorrect()

In [5]: test_type_double()
[mvk-error] VK_ERROR_INVALID_SHADER_NV: Unable to convert SPIR-V to MSL:
MSL conversion error: double types are not supported in buffers in MSL.


[mvk-error] VK_ERROR_INVALID_SHADER_NV: Compute shader function could not be compiled into pipeline. See previous logged error.
Traceback (most recent call last):

  File "<ipython-input-5-2d088eb457a5>", line 1, in <module>
    test_type_double()

  File "/Volumes/FAT/Developpement/Python/core/untitled1.py", line 122, in test_type_double

RuntimeError: vk::Device::createComputePipeline: ErrorInvalidShaderNV


In [6]: test_type_int()
Dtype value int32

In [7]: test_type_unsigned_int()
Dtype value uint32

In [8]: test_tensor_numpy_ownership()
ERROR:kp:Kompute Manager destructor reached with null Device pointer

[yves-surrel]

Here is the working "simple" example:

from kp import Manager, Tensor, OpTensorSyncDevice, OpTensorSyncLocal, OpAlgoDispatch
from pyshader import python2shader, ivec3, f32, Array

mgr = Manager()

# Can be initialized with List[] or np.Array
tensor_in_a = mgr.tensor([2, 2, 2])
tensor_in_b = mgr.tensor([1, 2, 3])
tensor_out = mgr.tensor([0, 0, 0])

sq = mgr.sequence()

sq.eval(OpTensorSyncDevice([tensor_in_a, tensor_in_b, tensor_out]))

# Define the function via PyShader or directly as glsl string or spirv bytes
@python2shader
def compute_shader_multiply(index=("input", "GlobalInvocationId", ivec3),
                            data1=("buffer", 0, Array(f32)),
                            data2=("buffer", 1, Array(f32)),
                            data3=("buffer", 2, Array(f32))):
    i = index.x
    data3[i] = data1[i] * data2[i]

algo = mgr.algorithm([tensor_in_a, tensor_in_b, tensor_out], compute_shader_multiply.to_spirv())

# Run shader operation synchronously
sq.eval(OpAlgoDispatch(algo))

sq.eval(OpTensorSyncLocal([tensor_out]))

assert tensor_out.data().tolist() == [2.0, 4.0, 6.0]```

[axsaucedo]

Thank you @yves-surrel - would you be interested on contributing this as a PR? I can help testing that the documentation all works as expected

[axsaucedo]

I've updated the readme to align with the test to ensure that it correctly runs