
Takže ZEN4D = ZEN4c (Cloud)
Bergamo = 128 "ZEN4c" cores, density optimized cache , significantly improved power efficiency , the same ISA like ZEN4 , the same features like ZEN4 ( SMT, AVX, etc. )
Moderátoři: flanker, Eddward, Baneshee
na detaily budeme muset počkat, dense cache nemusi nutně znamenat menší ( viz. V-cache ), čekal bych že AMD obětuje AVX512 (fulls speed AVX512), frekvence budou jistě menší ( o kolik to se ukáže )
těch 2x density OK, 1.25 performance taky sedí s HPC variantou , ale 2x power efficiency ? To je nějaký divný údaj ( nebo se týká pouze Zen4c )TSMC 5nm Node uses EUV and improves logic density by 1.8x compared to 7nm. For SRAM, the density is 1.3x higher. Just like the 7nm Node, 5nm will have two variants with one optimized for Mobile and other for HPC. The mobile node will allow 15% higher performance or 30% lower power consumption. The HPC variant will instead offer 25% higher performance over 7nm.
ale potom je ten slajd mixem nesmyslů, protože míchá dohromady několi věcí. HPC proces, který je určený pro vysoké frekvence (optimized for high performance) s nějakou jinou variantou téhož procesu (mobile/HDL ?), která má mnohem vyšší hustotu ale zase za cenu nižších frekvencí a ta určitě nezvedá výkon o 25% (1.25 performance ) Takže jsou to možná nejvyšší/optimální dosažené hodnoty dvou odlišných produktů na dvou různých procesech....
HPCwire - AMD Launches Milan-X CPU with 3D V-Cache and Multichip Instinct MI200 GPULisa Su píše:“We’ve worked with TSMC to optimize 5nm for high performance computing,” said Su. “[The new process] offers twice the density, twice the power efficiency and 1.25x the performance of the 7nm process we’re using in today’s products.”
jj píše to i Anandtech
When asked if this was a specific statement about core performance, AMD said that it wasn’t, and just a comment on the process node technologies. It is worth noting that 2x efficiency is quite a substantial claim based on metrics provided by TSMC on its N7 -> N5 disclosures.
Dr Su reinforced that technology roadmaps are all about making the right choices and the right junctures, and explicitly stated that our 5nm technology is highly optimized for high-performance computing – it’s not necessarily the same as some other 5nm technologies out there.