Summarizing CPU and GPU Design Trends with Product Data — mentioned by jmyeet

Thu, 31 Mar 2022 23:10:23 GMT

No. You can't easily equate human intelligence to required processing power but I've seen estimates of 10^15 to 10^18 FLOPS [1]. There's a huge variance there. Now consider the cost per FLOPS [2] and the energy cost of that [3]. Capital costs are still huge but let's focus on energy. You're talking in the megawatt energy range. That's... expensive. Particularly at scale. There have been and continue to be improvements in this but we're so far away from some superhuman AGI. Remember too that this would still require thousands of units and then you have to deal with the interconnects, cooling…

Paper and discussion history. A mention is not an endorsement.

Summarizing CPU and GPU Design Trends with Product Data — mentioned by egsmi

Wed, 07 Oct 2020 23:43:01 GMT

Not sure what “tame” means in this context. Performance per watt has been increasing every year for a long time and continues to do so [1]. So temperature is really just a design choice. One example, pick a performance get a power. Once you have a power, pick the mechanical system, get a temperature. It’s all just trade offs. If performance and enclosure was held constant over time then temperatures would go down, but few want that. [1] https://arxiv.org/pdf/1911.11313.pdf See figure 5

Paper and discussion history. A mention is not an endorsement.

Summarizing CPU and GPU Design Trends with Product Data · HackerPapers discussions

Summarizing CPU and GPU Design Trends with Product Data — mentioned by jmyeet

Summarizing CPU and GPU Design Trends with Product Data — mentioned by egsmi