In this paper, we investigate how to push the performance limits of serving Deep Neural Network (DNN) models on CPU-based servers. Specifically, we observe that while intra-operator parallelism across multiple threads is an effective way to reduce inference latency, it provides diminishing returns. Our primary insight is that instead of running a single instance of a model with all available threads on a server, running multiple instances each with smaller batch sizes and fewer threads for intra-op parallelism can provide lower inference latency. However, the right configuration is hard to determine manually since it is workload- (DNN model and batch size used by the serving system) and deployment-dependent (number of CPU cores on server). We present Packrat, a new serving system for online inference that given a model and batch size (𝐵) algorithmically picks the optimal number of instances (𝑖), the number of threads each should be allocated (𝑡), and the batch sizes each should operate on (𝑏) that minimizes latency. Packrat is built as an extension to TorchServe and supports online reconfigurations to avoid serving downtime. Averaged across a range of batch sizes, Packrat improves inference latency by 1.43× to 1.83× on a range of commonly used DNNs.

Building persistent memory (PM) data structures is difficult because crashes interrupt operations, leaving data structures in an inconsistent state. Solving this requires augmenting code that modifies PM state to ensure that interrupted operations can be completed or undone. Today, this is done using careful, hand-crafted code, a compiler pass, or page faults. We propose a new, easy way to transform volatile data structure code to work with PM that uses a cache-coherent accelerator to do this augmentation, and we show that it may outperform existing approaches for building PM structures.