Gradient functions are now registered in a global registry instead of being part of a transformation library. This means that you no longer need to pass around the library from the compiler in order to build a flow for learning.
SIMD generators for 32 and 64 bit integers. Although I have implemented support for 16-bit integers in the accumulate and reduce code in the macro assembler there is no 16 bit vector SIMD generator since 16 bit masking is not supported.
I have integrated the Myelin code profiling into the general learner task
Gradient functions are now registered in a global registry instead of being part of a transformation library. This means that you no longer need to pass around the library from the compiler in order to build a flow for learning.
SIMD generators for 32 and 64 bit integers. Although I have implemented support for 16-bit integers in the accumulate and reduce code in the macro assembler there is no 16 bit vector SIMD generator since 16 bit masking is not supported.
I have integrated the Myelin code profiling into the general learner task
Rotating register allocator.
Support for read/write file mappings.