Post by Axiomise
4,089 followers
ππ‘π² π¦π’π±ππ-π©π«πππ’π¬π’π¨π§ ππ₯π¨πππ’π§π -π©π¨π’π§π π―ππ«π’ππ’ππππ’π¨π§ π§ππππ¬ ππ¨π«π¦ππ₯ π¦πππ‘π¨ππ¬? Trustworthy AI Silicon needs mixed precision and transprecision models. Mixed-precision and transprecision computing introduce substantial verification challenges because correctness is no longer tied to a single, well-understood format, but to a tapestry of interacting precisions, formats, and rounding behaviours across the pipeline. The challenge is amplified further in configurable or transprecision FPUs, where the same hardware datapath may serve several formats and custom numerical types, making it easy for implementation shortcuts or shared logic to satisfy one format while violating the architectural intent of another. Effective verification of mixed-precision and transprecision designs requires more than numerical result checking: πit demands format-aware reference models π carefully targeted boundary-case stimulus πcross-format property checking, and πsystematic validation of rounding and exception behaviour under every supported precision configuration. This is where an app such as floatrix is really handy. πMinimal setup via a GUI πExhaustive proofs (thanks to CoreProve) πEdge and corner cases on pipelined implementations in no time. Our blog authored by Nicky Khodadad, Gia Nguyen Vu and Ashish Darbari shows how easy it is is to make mistakes in RTL implementations and thanks to the power of floatrix, easier to catch such bugs. https://lnkd.in/eBiGhDJZ Register your interest for product demos: https://lnkd.in/emDWPSbc #formalverification #aisilicon #axiomise #floatingpoint #precision
