Compiling code into silicon

Re:Compiling code into silicon by pezzana on Tuesday December 07, 2021 @10:00AM (#29473113)

> ... Compiling simple programs into silicon should be like using llvm or gcc: fast, automated, and accessible.

Some benchmarks would have been helpful. What kind of performance gain are there to make?

[ Parent | on HN ]

Re:Compiling code into silicon by jawnv6 on Tuesday December 07, 2021 @11:36AM (#29474575)

The execution time is going to seem really zippy after the 8 week wait for the chips to be fabbed.

[ Parent | on HN ]

Re:Compiling code into silicon by nojs on Tuesday December 07, 2021 @02:58PM (#29477436)

I’d also like to know this. In general, what performance gains would one expect to see from an IC custom made for a very specific algorithm versus a modern CPU?

[ Parent | on HN ]

Re:Compiling code into silicon by artisanspam on Wednesday December 08, 2021 @12:04AM (#29481434)

This isn't a compiler. All this is is a build tool that calls compilers in the background. So benchmarks for this tool aren't really useful.

However, for benchmarks between actual FOSS SystemVerilog compiler front-ends (i.e. language parsing), I've found this[1] chart from CHIPS Alliance to be useful.

Some of the tools listed are parsing + synthesis (yosys) or front-end + simulator (icarus, verilator). Several are just parsers.

[1]: https://chipsalliance.github.io/sv-tests-results/

[ Parent | on HN ]