siiky
2024/01/26
2024/01/26
2024/01/26
whitepaper,petri_nets,programming,distributed
We believe that detailed processor-specific optimizations can only be achieved by optimizing code generators that have been highly optimized to a particular processor architecture. This is because most modern processors employ very sophisticated pipelining and superscalar execution schemes that differ from processor to processor. We take an intermediate approach. Our software synthesis method aims to produce, as intermediate output, plain C code that retains a high degree of parallelism so that the subsequent processor-specific code generation step can produce efficient executable machine code for the target processor.
p. 5, ยง4.2
So the end program really is a plain sequential program, nothing special about it. Bill Lin hopes the C compiler to take advantage of CPU features.
[4]
[6]