Performance Analysis and Optimization of the FFTXlib on the Intel Knights Landing Architecture

Abstract

Abstract-In this paper, we address the decreasing performance of the FFTXlib, the Fast Fourier Transformation (FFT) kernel of Quantum ESPRESSO, when scaling to a full KNL node. An increased performance in the FFTXlib will likewise increase the performance of the entire Quantum ESPRESSO code one of the most used plane-wave DFT codes in the community of material science. Our approach focuses on, first, overlapping computation and communication and, second, decreasing resource contention for higher compute efficiency. In order to achieve this we use the OmpSs programming model based on task dependencies. We allow overlapping of computation and communication by converting all steps of the FFT into tasks following a flow dependency. In the same way, we decrease resource contention by converting each FFT into an individual task that can be scheduled asynchronously. In both cases, multiple FFTs can be computed in parallel. The task-based optimizations are implemented in the FFTXlib and show up to 10 % runtime reduction on the already highly optimized version. Since the task scheduling is done dynamically during execution by the parallel runtime, not statically by the user, it also frees the user from finding the ideal parallel configuration himself.

A survey on evaluating and optimizing performance of Intel Xeon Phi

“…Wagner et al 63 cache hit rate is already high. However, for MF-SGD, using both AVX-512 and MCDRAM boosts performance significantly, as AVX-512 instructions produce many memory accesses in each cycle which benefit from high bandwidth of MCDRAM.…”

“…Table 5 shows the parallel-programming languages used by different works. In OpenMP, both static 52,54,63,79 and dynamic 54,63,79 scheduling schemes have been used. Also, some works use OpenMP task pragmas for parallelization.…”

“…Also, some works use OpenMP task pragmas for parallelization. 33,52,63 TA B L E 5 Parallelization language/library used on Phi MPI 4,9,12,16,22,24,25,28,35,36,43,44,50,54,59,60,63,66,76,82,86 Others Pthreads, 11,23,76,84,91,95 Intel TBB, 8,48 Cilk Plus, 48 OpenCL 49,100 Chatzikonstantis et al 28 study inferior-olivary nucleus (InfOli) simulation which is used in brain modeling. They accelerate the simulation using (i) MPI, (ii) OpenMP, and (iii) hybrid MPI+OpenMP.…”

“…,[12][13][14][16][17][18]20,21,24,[27][28][29][30][33][34][35][36][37]39,[42][43][44][45][46][47][48]50,52,55,57,59,63,66,84,86,90,92,93,95,99 IntelMKL 2,17,19,31,32,40,93,99 …”

Improvements to Supercomputing Service Availability Based on Data Analysis

“…It consists of various infrastructure nodes connecte high-performance interconnect, including computing nodes, storage systems, nodes, management nodes, datemover (DM) nodes, and web servers. The computing nodes that perform parallel processes at high speeds consis 8305 Intel many-core processor Knights Landing Nodes (KNLs) [10,11] and a 132 server processor Skylake Nodes (SKLs) [12]. Each KNL node has 68 cores per sock 96 GB (16 GB × 6) memory, while each SKL node has two sockets, each of which cores and 192 GB (16 GB × 12) memory.…”