Blog Archives

20160707

Posted on 2016/07/07 by gumdaeng — Leave a comment

1. Moore’s law – it’s death and extension into another paradigm 세미나 내용 In the history of semiconductor, Moore’s law has been a guidance for the technology node transition for the industry. As the up-coming technology node’s dimension becomes sub-10nm,…

20151209

Posted on 2015/12/10 by gumdaeng — Leave a comment

1. Future Processor/Memory System Architecture, J. Thomas Pawlowski (slides) – MICRO 에서 발표된 세 개의 키노트 중에 가장 이해하지 못했음. 그런데 가장 중요했던 것 같다. 교수님께서는 이 키노트 발표가 중요하다고 말씀하셨음. 슬라이드 글자가 너무 작고, 발표자의 전달 방식이 나와 잘…

20151208

Posted on 2015/12/09 by gumdaeng — Leave a comment

1. Preparing for a Post Moore’s Law World, Prof. Todd Austin (slides) – C-FAR 연구 센터가 있음. 미래 아키텍쳐를 연구하기 위한 연구소. idle core가 늘어나고 있고, dennard scaling이 끝나가고 있으며, 실리콘의 결함이 늘어나고 있는 문제를 해결하기 위한 것. – dark…

Tagged with: 3D stacking, acceleration, accelerator, active interposer, approximate, approximation, border, border control, BP, cache, cache coherence, chip, CMOS scaling, cmp, CNN, column, comparison, Dark Silicon, data shuffling, database, Dennard Scaling, disintegration, DRAM, efficiency, energy, energy efficiency, gem5, gpu, imprecise, imprecision, instruction, integration, interposer, 정준기, knob, McPAT, micro, MICRO'12, MICRO'15, MICRO-48, migrate, migration, MLP, modularization, moore's law, MP, multi-chip module, multi-socket, neural, neural approximation, neural network, neural transformation, neuromorphic, neuroscience, NGPU, NoC, open source, parallelism, passive interposer, pattern, processor, quality, rank, sandbox, sandboxing, scaling, security, shuffling, silicon, SNN, stack, STDP, subarray, throughput, transformation
Posted in 1) Memo