Architect Labs Unveils AI-Designed Frostbite CPU Architecture
Architect Labs introduced Frostbite, a general-purpose CPU architecture with a new instruction set based on interaction nets and a mesh of small computing cells. The company says its AI systems autonomously iterated across the compiler, ISA, architecture, microarchitecture, RTL, verification and final GDS. On 25 workloads, including 15 held out for final evaluation, Architect Labs reported 4.59 times the performance per watt of an undisclosed, silicon-proven reference CPU core and an 8.77 times advantage in performance per area. These are vendor-reported pre-silicon simulation results, not measurements from a physical processor. Frostbite remains experimental, and no production or availability announcement has been made.
Frostbite is a substantive experiment in autonomous, cross-stack CPU design, showing how AI systems can search across software and hardware layers. However, its claimed efficiency gains remain unverified on physical silicon.
Abstract
In our previous post,Redwood, we showed that given an architecture, our AI systems could implement, verify, and deploy an inference accelerator in record time. However, similar to other scientific domains, we believe that AI systems, given the correct freedom and guardrails, can co-design software and hardware together, making optimizations across the stack and opening a design space so large it was previously inaccessible. We set out to answer the question: “Could our AI systems & tools lead to a novel hardware discovery that radically outperforms human-made designs?”
Today, we're introducing Frostbite, a general-purpose CPU with a new instruction set built from the ground up. Frostbite exploits the properties of interaction nets, namely that every step is local, every dependency is explicit, and every value has exactly one owner, to reach a step-function gain in CPU performance that has long stagnated. On a set of 25 initially chosen workloads, including 15 held out until the final evaluation, we achieved a 4.59x performance-per-watt multiple when compared to a modern, silicon-proven superscalar (dual-issue, in-order) CPU core.
Our AI systems autonomously iterated over the compiler, ISA, architecture, microarchitecture, RTL, verification, and final GDS. When we took a closer look at the discovered architectural primitives, we were surprised to find novel hardware implementations and runtime features that would have otherwise been difficult to architect. In spite of stagnation in CPU core improvement, we were able to take a workload concept all the way to a final GDS, and prove a step-function increase in perf/W and perf/area. We believe that as AI systems become more capable, every class of chip we know today, from datacenter-grade CPUs, to AI accelerators, to networking switches and NICs, will go through similar fundamental architectural shifts.