4 ms·
> it has got the right top eight instructions that you actually need to optimize for I wonder which ones he considers those to be -- the interview doesn't say,
by remexre 2y ago
> it has got the right top eight instructions that you actually need to optimize for
I wonder which ones he considers those to be -- the interview doesn't say, sadly...
- anonymoushn 2y agoI'm hoping we at least get pshufb, aesenc, rep movsb, some sort of integer multiply, and a floating point fma
- janice1999 2y agoThey've made some details public about their architecture [1] and even their RISC-V test suite [2] but I'd also love to know what it was referring to. [1] https://www.theregister.com/2024/08/27/tenstorrent_ai_blackhole/ https://www.theregister.com/2024/08/27/tenstorrent_ai_blackh... [2] https://github.com/tenstorrent/riscv_arch_tests https://github.com/tenstorrent/riscv_arch_tests
- rowanG077 2y agoI would say it's load, store, compare, add, subtract, call, return and branch. If you have that you cover almost all code.
- sgerenser 2y agoImplementing multiplies as a bunch of adds has got to suck for performance though.
- colejohnson66 2y agoWell, if you decompose it to a series of shifts and adds, it’s a bit quicker. That’s actually how the 8086 implemented it.
- rowanG077 2y agoSure. Multiplication is exceedingly rare compared to any of those 8 instructions though. RISC-V base instruction set doesn't include multiplication for example. If you had room to add more then 8 I would rather add the bitops first. This will allow you to also write a much more efficient soft mul in addition to many more things.