{"id":9139,"date":"2025-09-09T04:50:13","date_gmt":"2025-09-09T04:50:13","guid":{"rendered":"https:\/\/www.newsbeep.com\/ie\/9139\/"},"modified":"2025-09-09T04:50:13","modified_gmt":"2025-09-09T04:50:13","slug":"maximizing-processor-efficiency-with-the-lean-metric","status":"publish","type":"post","link":"https:\/\/www.newsbeep.com\/ie\/9139\/","title":{"rendered":"Maximizing Processor Efficiency With the LEAN Metric"},"content":{"rendered":"

In July, a University of Michigan computer engineering<\/a> professor put out a new idea for measuring the efficiency of a processor design<\/a>. Todd Austin\u2019s<\/a> LEAN metric received both praise and skepticism, but even the critics understood the rationale: A lot of silicon is devoted to things that are not actually doing computing. For example, more than 95 percent of an Nvidia Blackwell<\/a> GPU is designated for other tasks, Austin told IEEE Spectrum. It\u2019s not like these parts aren\u2019t doing important things, such as choosing the next instruction to execute, but Austin believes processor architectures can and should move toward designs that maximize computing and minimize everything else.<\/p>\n

Todd Austin<\/p>\n

Todd Austin is a professor of electrical engineering<\/a> and computer science at the University of Michigan in Ann Arbor.<\/p>\n

What does the LEAN score measure?<\/p>\n

Todd Austin: LEAN stands for Logic Executing Actual Numbers. A score of 100 percent\u2014an admittedly unreachable goal\u2014would mean that every transistor is computing a number that contributes to the final results of a program. Less than 100 percent means that the design devotes silicon and power to inefficient computing and to logic that doesn\u2019t do computing.<\/p>\n

What\u2019s this other logic doing?<\/p>\n

Austin: If you look at how high-end architectures have been evolving, you can divide the design into two parts: the part that actually does the computation of the program and the part that decides what computation to do. The most successful designs are squeezing that \u201cdeciding what to do\u201d part down as much as possible.<\/p>\n

Where is computing efficiency lost in today\u2019s designs?<\/p>\n

Austin: The two losses that we experience in computation are precision loss and speculation loss. Precision loss means you\u2019re using too many bits to do your computation. You see this trend in the GPU world. They\u2019ve gone from 32-bit floating-point precision to 16-bit to 8-bit to even smaller. These are all trying to minimize precision loss in the computation.<\/p>\n

Speculation loss comes when instructions are hard to predict. [Speculative execution<\/a> is when the computer guesses what instruction will come next and starts working even before the instruction arrives.] Routinely, in a high-end CPU, you\u2019ll see two [speculative] instruction results thrown away for every one that is usable.<\/p>\n

You\u2019ve applied the metric to an Intel<\/a> CPU, an Nvidia<\/a> GPU, and Groq<\/a>\u2019s AI inference chip. Find anything surprising?<\/p>\n

Austin: Yeah! The gap between the CPU and the GPU was a lot less than I thought it would be. The GPU was more than three times better than the CPU. But that was only 4.64 percent [devoted to efficient computing] versus 1.35 percent. For the Groq<\/a> chip, it was 15.24 percent. There\u2019s so much of these chips that\u2019s not directly doing compute.<\/p>\n

What\u2019s wrong with computing today that you felt like you needed to come up with this metric?<\/p>\n

Austin: I think we\u2019re actually in a very good state. But it\u2019s very apparent when you look at AI scaling trends that we need more compute, bigger access to memory, more memory bandwidth. And this comes around at the end of Moore\u2019s Law<\/a>. As a computer architect, if you want to create a better computer, you need to take the same 20 billion transistors<\/a> and rearrange them in a way that is more valuable than the previous arrangement. I think that means we\u2019re going to need leaner and leaner designs.<\/p>\n

This article appears in the September 2025 print issue as \u201cTodd Austin.\u201d<\/p>\n

From Your Site Articles<\/p>\n

Related Articles Around the Web<\/p>\n","protected":false},"excerpt":{"rendered":"In July, a University of Michigan computer engineering professor put out a new idea for measuring the efficiency…\n","protected":false},"author":2,"featured_media":9140,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[21],"tags":[9596,241,61,60,9598,9599,80,9597],"class_list":{"0":"post-9139","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"has-post-thumbnail","7":"category-computing","8":"tag-5-questions","9":"tag-computing","10":"tag-ie","11":"tag-ireland","12":"tag-processor-design","13":"tag-silicon","14":"tag-technology","15":"tag-typedepartments"},"_links":{"self":[{"href":"https:\/\/www.newsbeep.com\/ie\/wp-json\/wp\/v2\/posts\/9139","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.newsbeep.com\/ie\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.newsbeep.com\/ie\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/ie\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/ie\/wp-json\/wp\/v2\/comments?post=9139"}],"version-history":[{"count":0,"href":"https:\/\/www.newsbeep.com\/ie\/wp-json\/wp\/v2\/posts\/9139\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/ie\/wp-json\/wp\/v2\/media\/9140"}],"wp:attachment":[{"href":"https:\/\/www.newsbeep.com\/ie\/wp-json\/wp\/v2\/media?parent=9139"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.newsbeep.com\/ie\/wp-json\/wp\/v2\/categories?post=9139"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.newsbeep.com\/ie\/wp-json\/wp\/v2\/tags?post=9139"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}