def test_functions(self):
v = Metal.MTLOriginMake(1, 2, 3)
self.assertIsInstance(v, Metal.MTLOrigin)
self.assertEqual(v, (1, 2, 3))
v = Metal.MTLSizeMake(1, 2, 3)
self.assertIsInstance(v, Metal.MTLSize)
self.assertEqual(v, (1, 2, 3))
v = Metal.MTLRegionMake1D(1, 2)
self.assertIsInstance(v, Metal.MTLRegion)
v = Metal.MTLRegionMake2D(1, 2, 3, 4)
self.assertIsInstance(v, Metal.MTLRegion)
v = Metal.MTLRegionMake3D(1, 2, 3, 4, 5, 6)
self.assertIsInstance(v, Metal.MTLRegion)
v = Metal.MTLSamplePositionMake(0.5, 1.5)
self.assertIsInstance(v, Metal.MTLSamplePosition)
self.assertEqual(v, (0.5, 1.5))
m[:] = bytes(vA)
vB = np.random.randn(arrayLength).astype(np.float32)
t = bufferB.contents()
m = t.as_buffer(bufferSize)
m[:] = bytes(vB)
commandBuffer = commandQueue.commandBuffer()
computeEncoder = commandBuffer.computeCommandEncoder()
computeEncoder.setComputePipelineState_(addFunctionPSO)
computeEncoder.setBuffer_offset_atIndex_(bufferA, 0, 0)
computeEncoder.setBuffer_offset_atIndex_(bufferB, 0, 1)
computeEncoder.setBuffer_offset_atIndex_(bufferResult, 0, 2)
gridSize = Metal.MTLSizeMake(arrayLength, 1, 1)
threadGroupSize = addFunctionPSO.maxTotalThreadsPerThreadgroup()
if threadGroupSize > arrayLength:
threadGroupSize = arrayLength
threadgroupSize = Metal.MTLSizeMake(threadGroupSize, 1, 1)
computeEncoder.dispatchThreads_threadsPerThreadgroup_(gridSize,
threadgroupSize)
computeEncoder.endEncoding()
commandBuffer.commit()
commandBuffer.waitUntilCompleted()
t = bufferResult.contents()
result = np.frombuffer(t.as_buffer(bufferSize), dtype=np.float32)
