The best we have nowadays is using compute shaders for the same purpose.
Just like when using a TMS34010 with its C SDK.