Deep learning is finding its way into high energy physics by replacing traditional Monte Carlo simulations. However, deep learning still requires an excessive amount of computational resources. A promising approach to make deep learning more efficient is to quantize the parameters of the neural networks to reduced precision. Reduced precision computing is extensively used in modern deep learning and results to lower execution inference time, smaller memory footprint and less memory bandwidth. In this paper we analyse the effects of low precision inference on a complex deep generative adversarial network model. The use case which we are addressing is calorimeter detector simulations of subatomic particle interactions in accelerator based high energy physics. We employ the novel Intel low precision optimization tool (iLoT) for quantization and compare the results to the quantized model from TensorFlow Lite. In the performance benchmark we gain a speed-up of 1.73x on Intel hardware for the quantized iLoT model compared to the initial, not quantized, model. With different physics-inspired self-developed metrics, we validate that the quantized iLoT model shows a lower loss of physical accuracy in comparison to the TensorFlow Lite model.

Applications requiring high-speed TCP/IP processing can easily saturate a modern server. We and others have previously suggested alleviating this problem in multiprocessor environments by dedicating a subset of the processors to perform network packet processing. The remaining processors perform only application computation, thus eliminating contention between these functions for processor resources. Applications interact with packet processing engines (PPEs) using an asynchronous I/O (AIO) programming interface which bypasses the operating system. A key attraction of this overall approach is that it exploits the architectural trend toward greater thread-level parallelism in future systems based on multi-core processors. In this paper, we conduct a detailed experimental performance analysis comparing this approach to a best-practice configured Linux baseline system.

In recent years parallel computing on a fast network of high performance and low cost workstations has become a viable and an economic option, compared to that on an expensive high performance parallel supercomputer for solving large Grand Challenge scientific problems. This paper focuses on how to efficiently exploit the computing resources of a set of heterogeneous Unix workstations. We have further developed the CHARM parallel programming environment to allow programs written in the CHARM language execute adjacently on such a cluster. We have also developed a new scheme to schedule tasks statically and balance the load dynamically to achieve high effective utilization. Performance results for several applications programs including ray-tracing, all-pairs shortest path, and matrix multiply on a heterogeneous cluster of Sun Sparcs, IBM RS/6000s and HP-PA 7100s show a significant improvement in execution time over sequential execution.>

We preseni a general approach of self-scheduling a non-uniform parallel loop on a distributed-memory nzachine.The approach has two phases: a static schedu!tng phase and a dynamic scheduling phase.In addition to reduce scheduling overhead, using the static scheduling phase allows the data needed by the staticallyscheduled iterations to beprefetched.The dynamic scheduling phase balances the workload.Data dastrtbution methods for setf-schedulmg are also the focus of this paper.We classify the data distribution methods into four categories and present partial duplication, a method that allows the problem size to grow ltnearly in the number of processors.The experiments conducted on a 64-node NCUBE show that as much as 79% improvement is achieved over static scheduling on the generdion of a false-color image.