From: Luke Kenneth Casson Leighton Date: Tue, 27 Aug 2019 17:15:21 +0000 (+0100) Subject: add PDF version, change SV/RVV comparison slide X-Git-Tag: convert-csv-opcode-to-binary~4195 X-Git-Url: https://git.libre-soc.org/?a=commitdiff_plain;h=a40fbd36b9ac3bb432f0d47ab9c2272bcb2d1a7d;p=libreriscv.git add PDF version, change SV/RVV comparison slide --- diff --git a/simple_v_extension/open_3d_alliance_2019aug26.pdf b/simple_v_extension/open_3d_alliance_2019aug26.pdf new file mode 100644 index 000000000..48296cda0 Binary files /dev/null and b/simple_v_extension/open_3d_alliance_2019aug26.pdf differ diff --git a/simple_v_extension/open_3d_alliance_2019aug26.tex b/simple_v_extension/open_3d_alliance_2019aug26.tex index 78f5a9605..338238cb4 100644 --- a/simple_v_extension/open_3d_alliance_2019aug26.tex +++ b/simple_v_extension/open_3d_alliance_2019aug26.tex @@ -139,16 +139,19 @@ \begin{itemize} \item RVV is designed exclusively for supercomputing\\ - (RVV simply has not been designed with 3D in mind).\vspace{6pt} + (RVV simply has not been designed with 3D in mind). \item Like SIMD, RVV uses dedicated opcodes\\ - (google "SIMD considered harmful")\vspace{6pt} + (google "SIMD considered harmful") \item 98\% of FP opcodes are duplicated in RVV. Large portion\\ - of BitManip opcodes duplicated in predicate Masks\vspace{6pt} + of BitManip opcodes duplicated in predicate Masks \item OP32 space is extremely precious: 48 and 64 bit opcode space\\ - comes with an inherent I-Cache power consumption penalty\vspace{6pt} + comes with an inherent I-Cache power consumption penalty \item Simple-V "prefixes" scalar opcodes (all of them)\\ No need for any new "vector" opcodes (at all).\\ - Can therefore use the RVV major opcode for 3D\vspace{6pt} + Can therefore use the RVV major opcode for 3D + \item SV augments "scalar" opcodes. Implications: it is relatively\\ + straightforward to convert an \textit{existing design} to SV.\\ + SV "slots in" between instruction decode and the ALU. \end{itemize} }