Charles E. Leiserson, Neil C. Thompson, Joel S. Emer
Bradley C. Kuszmaul, Butler W. Lampson, Daniel Sanchez, Tao B. Schardl
2020 · Science
From bottom to topThe doubling of the number of transistors on a chip every 2 years, a seemly inevitable trend that has been called Moore's law, has contributed immensely to improvements in computer performance. However, silicon-based transistors cannot get much smaller than they are today, and other approaches should be explored to keep performance growing. Leisersonet al.review recent examples and argue that the most promising place to look is at the top of the computing stack, where improvements in software, algorithms, and hardware architecture can bring the much-needed boost.Science, this issue p.eaam9744
Metadata: Crossref. Abstracts retain their original authors’ rights.
Original HN comments, with labels from simple text rules. A label describes the whole comment and may miss context or irony. Read the discussion to make up your own mind. About the labels
3 selected excerpts · 3 total mentions
Most common coding patterns leave a lot of performance unclaimed, by not fully specializing to the hardware. This article is an interesting example. For another interesting demonstration, see this CS classic "There's plenty of room at the top" https://www.science.org/doi/10.1126/science.aam9744
You're correct that specialised analog companies have not done well historically. However, we don't find ourselves in exactly the same position in computer architecture/performance as we've been decades before. There's some (relatively) new ideas that now the performance of computers will be pushed more by dedicated silicon for a dedicated purpose, and tools. See for example there's plenty of room at the top [1], or Hennesy's talk at Google [2]. This of course does not mean that analog computers are suddenly viable, but it does mean that they could potentially fill a niche where they failed previously. Anecdotally, when looking at jobs for hardware design by the likes of…
Infineon, STM, Cyient etc. there seems to be a relatively high ask for (senior) analog designers, and a new focus on mixed-technology chips. It might turn out to be a dud still, but it isn't the same situation as decades before. 1: https://www.science.org/doi/abs/10.1126/science.aam9744 , or the IEEE article https://ieeexplore.ieee.org/abstract/document/7863324 2: https://www.youtube.com/watch?v=Azt8Nc-mtKM , around the 12:15 mark, but the entire video is relevant.
Here is a (semi)recent paper in Science about the end of Moore's law. As I understand it (but I'm not an expert), Figure 2 seems to give pretty compelling evidence that Dennard scaling (i.e, the phenomenon that historically allowed for smaller chips to run at higher clock speeds without an increase in power usage) seems to have stopped around 2005, and that subsequent speed ups have largely come from in-chip parallelism. https://www.science.org/doi/10.1126/science.aam9744
More from the same topics.
Haoran Xu, Fredrik Kjolstad
Fast compilation is important when compilation occurs at runtime, such as query compilers in modern database systems and WebAssembly virtual machines in modern browsers.
David Manheim, Scott Garrabrant
There are several distinct failure modes for overoptimization of systems on the basis of metrics.
Donald E. Knuth
The author presents two tricks to accelerate depth-first search algorithms for a class of combinatorial puzzle problems, such as tiling a tray by a fixed set of polyominoes.
Jon Kleinberg, Sendhil Mullainathan, Manish Raghavan
Recent discussion in the public sphere about algorithmic classification has involved tension between competing notions of what it means for a probabilistic classification to be fair to different groups.