Support half precision floating point types

[mairooni]

Description

This PR provides support for half-float types, i.e. float values represented by 16 bits instead of 32. Since Java does not currently offer a half-float type, in this implementation it is represented by a new class, HalfFloat. The HalfFloat class offers two constructors. The first receives a float value, which is internally converted to float-16, using the Float.floatToFloat16 function. The second directly receives a short value, which represents a half-float. The HalfFloat class exposes a set of operations that can be performed between half-float values, specifically addition, subtraction, multiplication, and division.
Finally, a new off-heap array type for half-float values (HalfFloatArray class) has been included.

Backend/s tested

Mark the backends affected by this PR.

• [x] OpenCL
• [x] PTX
• [x] SPIRV

OS tested

Mark the OS where this PR is tested.

• [x] Linux
• [ ] OSx
• [ ] Windows

Did you check on FPGAs?

If it is applicable, check your changes on FPGAs.

• [ ] Yes
• [x] No

How to test the new patch?

Tests have been added for the following scenarios:

1) Initialization of a new HalfFloat instance in the kernel: tornado-test -V uk.ac.manchester.tornado.unittests.arrays.TestArrays#testHalfFloatInitialization 2) Addition of two HalfFloatArrays: tornado-test -V uk.ac.manchester.tornado.unittests.arrays.TestArrays#testVectorAdditionHalfFloat 3) Subtraction of two HalfFloatArrays: tornado-test -V uk.ac.manchester.tornado.unittests.arrays.TestArrays#testVectorSubtractionHalfFloat 4) Multiplication of two HalfFloatArrays: tornado-test -V uk.ac.manchester.tornado.unittests.arrays.TestArrays#testVectorMultiplicationHalfFloat 5) Division between two HalfFloatArrays: tornado-test -V uk.ac.manchester.tornado.unittests.arrays.TestArrays#testVectorDivisionHalfFloat 6) Test the fromElements andfromArray functions of the HalfFloatArray class: tornado-test -V uk.ac.manchester.tornado.unittests.api.TestAPI#testSegmentsHalfFloats 7) Test the fromSegment function of the HalfFloatArray class: tornado-test -V uk.ac.manchester.tornado.unittests.api.TestAPI#testBuildWithSegmentsHalfFloat

[jjfumero]

Unit-test are passing for SPIR-V and PTX. For OpenCL, depending on the device, this passes.

In my case, using RTX 2060 and the NVIDIA Drivers 510.54 do not pass.

[mairooni]

I included a function that checks if FP16 is supported. However, since on my machine I do not have this capability but I can pass the tests for OpenCL, we decided with @jjfumero to keep the printing of the OpenCL pragma for the time being, regardless of whether it is supported, until we figure out why this is happening.

[jjfumero]

Yes, we can query the capability. The pragma can always be generated. We need to investigate how to run OpenCL FP16 on modern NVIDIA GPUs.

[mikepapadim]

Ok, lets merge it then and we can iterate for a fix in codegen when they add the extension in the driver

[jjfumero]

It appears a new conflict. @mairooni, can you take a look?