RFC: RandomX/rx/0 low-level implementation details for custom application
Looking for technical docs/whitepapers on RandomX VM spec beyond the basic GitHub repo. Specifically need:
rx/0 dataset access patterns optimization
VM instruction set timing characteristics
Blake2b finalization implementation details
Cache-conscious programming approaches for scratchpad
x86-64 ASM optimizations for the superscalar pipeline
Building from scratch for research purposes. Not interested in ready-made miners or "how to mine" guides - need low-level architecture documentation and cycle-accurate analysis.
If you have experience with the execution units allocation or FPU pipeline utilization in RandomX, please share insights. Particularly interested in the 2080 instruction subset and register file usage patterns.
References to academic papers or deep technical discussions appreciated.
rx/0 dataset access patterns optimization
VM instruction set timing characteristics
Blake2b finalization implementation details
Cache-conscious programming approaches for scratchpad
x86-64 ASM optimizations for the superscalar pipeline
Building from scratch for research purposes. Not interested in ready-made miners or "how to mine" guides - need low-level architecture documentation and cycle-accurate analysis.
If you have experience with the execution units allocation or FPU pipeline utilization in RandomX, please share insights. Particularly interested in the 2080 instruction subset and register file usage patterns.
References to academic papers or deep technical discussions appreciated.