3 ms·
How does this stuff differ from the way GPU shader compilers extract ILP today? For example, IIRC, AMD GPUs compiler will already examine a given instruction st
by cromwellian 15y ago
How does this stuff differ from the way GPU shader compilers extract ILP today? For example, IIRC, AMD GPUs compiler will already examine a given instruction stream, split it up into homogenous instruction groups (called clauses) and pack them tightly into oodles of VLIW units based on dependency graph information.
Is this just another case of Apple giving existing practice a new name, and shepherding a patent through the USPTO?
- stephencanon 15y agoThe article would appear to answer your question: "It’s a high-level microarchitecture style: an alternative, rather than an extension, to today’s pervasive out-of-order superscalar designs, VLIW architectures, or vector machines."