The rapid growth of AI workloads at Meta has motivated our inhouse development of AI chips, aiming to significantly reduce the total cost of ownership and mitigate risks posed by unpredictable GPU supplies.At ISCA'23, we presented Meta's first-generation AI chip, MTIA 1.This paper describes its successor, MTIA 2i, now deployed at scale and serving billions of users.MTIA 2i significantly improves upon MTIA 1, reducing total cost of ownership by 44% compared to GPUs while delivering competitive performance per watt.A key differentiator is its memory hierarchy: instead of costly HBM, it uses large SRAM alongside LPDDR.Although there has been a proliferation of publications on AI chips, they often focus on architectural design and overlook three critical aspects:(1) co-designing and optimizing ML models to work effectively with the AI chip; (2) demonstrating sufficient flexibility to support a wide range of models; and (3) during the productionization process, addressing challenges unanticipated or decisions deferred at design time, such as dealing with memory errors, safe overclocking, reducing provisioned power, and implementing real-time firmware updates to mitigate silicon design defects.A key contribution of this paper is sharing our experience with these aspects, based on our journey of productionizing MTIA 2i at scale.

Training large AI models such as LLMs and DLRMs costs massive GPUs and computing time. The high training cost has become only affordable to big tech companies, meanwhile also causing increasing concerns about the environmental impact. This paper presents CoMERA, a **Co**mputing- and **M**emory-**E**fficient training method via **R**ank-**A**daptive tensor optimization. CoMERA achieves end-to-end rank-adaptive tensor-compressed training via a multi-objective optimization formulation, and improves the training to provide both a high compression ratio and excellent accuracy in the training process. Our optimized numerical computation (e.g., optimized tensorized embedding and tensor-vector contractions) and GPU implementation eliminate part of the run-time overhead in the tensorized training on GPU. This leads to, for the first time, $2-3\times$ speedup per training epoch compared with standard training. CoMERA also outperforms the recent GaLore in terms of both memory and computing efficiency. Specifically, CoMERA is $2\times$ faster per training epoch and $9\times$ more memory-efficient than GaLore on a tested six-encoder transformer with single-batch training. Our method also shows $\sim 2\times$ speedup than standard pre-training on a BERT-like code-generation LLM while achieving $4.23\times$ compression ratio in pre-training. With further HPC optimization, CoMERA may reduce the pre-training cost of many other LLMs. An implementation of CoMERA is available at <https://github.com/ziyangjoy/CoMERA>.

Meta has traditionally relied on using CPU-based servers for running inference workloads, specifically Deep Learning Recommendation Models (DLRM), but the increasing compute and memory requirements of these models have pushed the company towards using specialized solutions such as GPUs or other hardware accelerators. This paper describes the company's effort in constructing its first silicon specifically designed for recommendation systems; it describes the accelerator architecture and platform design, the software stack for enabling and optimizing PyTorch-based models and provides an initial performance evaluation. With our emerging software stack, we have made significant progress towards reaching the same or higher efficiency as the GPU: We averaged 0.9x perf/W across various DLRMs, and benchmarks show operators such as GEMMs reaching 2x perf/W. Finally, the paper describes the lessons we learned during this journey which can improve the performance and programmability of future generations of architecture.

The computation for today's intelligent personal assistants such as Apple Siri, Google Now, and Microsoft Cortana, is performed in the cloud. This cloud-only approach requires significant amounts of data to be sent to the cloud over the wireless network and puts significant computational pressure on the datacenter. However, as the computational resources in mobile devices become more powerful and energy efficient, questions arise as to whether this cloud-only processing is desirable moving forward, and what are the implications of pushing some or all of this compute to the mobile devices on the edge.

Mobile devices are becoming more powerful and versatile than ever, calling for better embedded processors. Following the trend in desktop CPUs, microprocessor vendors are trying to meet such needs by increasing the number of cores in mobile device SoCs. However, increasing the number does not translate proportionally into performance gain and power reduction. In the past, studies have shown that there exists little parallelism to be exploited by a multi-core processor in desktop platform applications, and many cores sit idle during runtime. In this paper, we investigate whether the same is true for current mobile applications. We analyze the behavior of a broad range of commonly used mobile applications on real devices. We measure their Thread Level Parallelism (TLP), which is the machine utilization over the non-idle runtime. Our results demonstrate that mobile applications are utilizing less than 2 cores on average, even with background applications running concurrently. We observe a diminishing return on TLP with increasing the number of cores, and low TLP even with heavy-load scenarios. These studies suggest that having many powerful cores is over-provisioning. Further analysis of TLP behavior and big-little core energy efficiency suggests that current mobile workloads can benefit from an architecture that has the flexibility to accommodate both high performance and good energy-efficiency for different application phases.

Mobile devices continue to increase the number of cores in an attempt to meet the needs of performance-demanding applications. However, the increasing number of cores does not necessarily translate into performance gain and/or power reduction. In this paper we investigate how multi-core mobile devices are utilized by applications. Our results demonstrate that mobile applications are utilizing less than 2 cores on average, which shows that multi-cores are generally underutilized by today's mobile applications. Unless application developers can significantly improve core utilization, further increasing core counts will result in little gain.