The International Journal of High Performance Computing Applications

The International Journal of High Performance Computing Applications is a bimonthly peer-reviewed journal from SAGE Publications in Thousand Oaks, California, carrying a title that dates from 1998 on a volume sequence that began in 1987, and devoted to the use of supercomputers and large parallel systems in scientific and engineering computation, with a 2025 Journal Impact Factor of 2.4.《国际高性能计算应用杂志》由 SAGE Publications 在加利福尼亚州千橡市出版，双月刊，同行评审，现名始于 1998 年，卷号可上溯到 1987 年，内容围绕超级计算机与大规模并行系统在科学与工程计算中的使用，2025 年期刊影响因子为 2.4。

Scope

• The remit is set out in plain terms, covering original peer-reviewed research papers and review articles on the use of supercomputers to solve complex modeling problems in a spectrum of disciplines.刊物的定位写得很直白，发表经同行评审的原创研究论文与综述文章，主题是使用超级计算机求解各学科中的复杂建模问题。

• A standing topic list names artificial intelligence, climate modeling, cryptographic analysis, geophysics, molecular biology, molecular dynamics, nuclear physics and physical oceanography as representative application fields.常设的选题清单把人工智能与气候模拟列在前列，密码分析与地球物理同在其内，分子生物学与分子动力学各自成题，核物理与物理海洋学也在名单之中。

• System architecture sits at the centre, covering interconnect design, memory hierarchies, heterogeneous nodes and the system-level choices that decide sustained performance at scale.体系结构处在中心位置，互连设计与存储层次都在其中，异构节点同样受理，决定大规模持续性能的系统级取舍是核心议题。

• Parallel algorithms and programming models follow, with MPI and OpenMP codes, task-based and asynchronous many-task runtimes, and the portability of programs across machines treated as first-class subjects.并行算法与编程模型紧随其后，MPI 与 OpenMP 程序在列，任务式与异步多任务运行时同样在内，程序跨机器迁移的可移植性被当作一等课题。

• Performance modeling and tuning receive steady attention, from roofline analysis and scaling studies to the benchmarking of complete applications on production machines.性能建模与调优得到持续关注，性能上限分析排在前面，扩展性研究紧随其后，生产系统上整应用的基准测试同样受理。

• Accelerators are treated as ordinary working hardware, with GPU-aware communication, CUDA and HIP ports, OpenACC directives and offload strategies appearing issue after issue.加速器被当作日常计算硬件对待，GPU 感知通信在列，CUDA 与 HIP 移植同样受理，OpenACC 指示语句与卸载策略在版面上并不少见。

• Compiler work belongs here too, including intermediate representations, code generation for sparse geometries, domain-specific languages and the automated translation of legacy Fortran and C++ codes.编译器工作同样属于这一范围，中间表示与代码生成都在其中，面向稀疏几何的生成方法也受理，领域专用语言与遗留 Fortran 代码的自动转换一并覆盖。

• Large-scale data handling and artificial intelligence form an expanding strand, with language models as research assistants, inference at scale, in-memory streaming and in-situ visualization all represented.大规模数据处理与人工智能构成不断扩张的一条线，语言模型作为科研助手进入选题，规模化推理同样受理，内存内流式传输与原位可视化一并覆盖。

• Application papers take up most of the page count, spanning computational fluid dynamics, combustion, weather and climate, ocean modeling, materials, fusion energy, astrophysics and subsurface flow.应用类论文占据多数版面，计算流体力学与燃烧模拟都在其中，天气与气候模式、海洋建模同样常见，材料与聚变能各成一支，天体物理与地下水流也有人在写。

• Energy and power are measured rather than assumed, with production codes instrumented to report joule counts alongside time to solution.能耗与功率以实测为准，生产代码被加上计量，焦耳数与求解时间一起报告。

• Fault tolerance and resilience complete the systems side, covering checkpointing strategies, silent data corruption and numerical algorithms that survive failure at scale.容错与韧性补全系统一侧，检查点策略在列，静默数据错误同样受理，能在大规模故障下继续求解的数值算法也在其中。

• Numerical kernels and libraries appear continuously, from sparse direct solvers and preconditioners to mixed-precision iterative refinement and the emulation of floating-point arithmetic in integers.数值内核与函数库持续出现，稀疏直接解法与预条件子都在其中，混合精度迭代精化同样受理，浮点运算的整数模拟也占一席。

• Benchmarks carry weight of their own, with HPL, HPCG and HPL-MxP results, benchmark taxonomies and cross-platform scaling comparisons published as substantive contributions.基准测试自有分量，HPL 与 HPCG 的成绩在列，混合精度基准的讨论同样受理，跨平台扩展性比较被当作实质贡献发表。

• Review coverage extends to hardware trends, including the memory systems and general-purpose processors that shape what scientific codes can achieve in the coming decade.综述覆盖硬件趋势，存储系统与通用处理器的发展都纳入讨论，这类内容决定了科学计算代码在下一个十年能做到什么。

• Newer execution settings have found room as well, with digital twins, cloud and serverless runs and mixed CPU-GPU workflows all appearing in recent volumes.新的运行环境同样找到了位置，数字孪生与云端无服务器作业都在其中，CPU 与 GPU 混合的工作流也见于近年各卷。

History

• The title descends from The International Journal of Supercomputer Applications, whose volume 1 number 1 appeared in spring 1987 under MIT Press in Cambridge, Massachusetts.本刊的前身是《国际超级计算机应用杂志》，第 1 卷第 1 期于 1987 年春季由麻省理工学院出版社（MIT Press）在马萨诸塞州剑桥出版。

• A second name arrived with volume 8 in spring 1994, when the journal became The International Journal of Supercomputer Applications and High Performance Computing, still from MIT Press and still quarterly.1994 年春季的第 8 卷换上了第二个刊名，改称《国际超级计算机应用与高性能计算杂志》，仍由麻省理工学院出版社出版，刊期依旧为季刊。

• Volume 8 numbers 3 and 4, a combined issue dated September 1994, printed the MPI message-passing standard chapter by chapter, an early sign of the journal's appetite for standards work.第 8 卷第 3-4 期合刊于 1994 年 9 月出版，逐章刊出 MPI 消息传递标准，刊物偏爱标准文本的口味在那时已经显露。^[1]

• SAGE took over near the end of that run, and the NLM record notes the imprint moving to Thousand Oaks, California, with Sage Science Press for the final issues of volume 11 in 1997.这段刊史临近结束时 SAGE 接手，美国国家医学图书馆的著录写明出版地转到加利福尼亚州千橡市，第 11 卷 1997 年的末几期由 Sage Science Press 印行。

• The present title opened with volume 12 numbers 1 and 2, a combined spring-summer issue dated 1998, so the volume count carried straight on from the predecessor.现名从第 12 卷第 1-2 期合刊开始，时在 1998 年春夏，卷号与前身刊物直接接续，没有重新起算。

• That opening issue was given over to MPI-2, printing ten chapters and an annex of the extended standard across pages 12 to 299.这一期整期交给 MPI-2，十章正文与一份附录排在 12 至 299 页，扩展标准的全文由此一次刊完。^[2]

• Volume numbers have tracked the calendar for most of the title's life, reaching volume 30 in 2016 and volume 40 in 2026, with one volume a year since 2020.在大部分年份里卷号与日历同步，2016 年为第 30 卷，2026 年为第 40 卷，2020 年起每年固定一卷。

• Quarterly publication held from 1987 to 2016, with issues dated February, May, August and November. From volume 31 in 2017 the journal moved to six issues a year in odd months.1987 年至 2016 年维持季刊，四期分别标在 2 月与 5 月，8 月与 11 月各一期，2017 年第 31 卷起改为每年六期，出刊月份落在单月。

• Combined issues appear when a slot needs merging, as with numbers 5-6 of volume 36 in November 2022 and numbers 3-4 of volume 37 in July 2023.需要合并时会出现合刊，2022 年 11 月第 36 卷第 5-6 期是一例，2023 年 7 月第 37 卷第 3-4 期又是一例。

• Special issues with distinctive titles are a long-standing habit, and the NLM record carries that practice as a general note on the serial.专刊另起题名是长期习惯，美国国家医学图书馆的著录把这一点写成了刊物的通则。

• SAGE has digitised the back files, so articles from 1987 onwards carry 10.1177 DOIs, and the predecessor volumes sit in Crossref under their own ISSNs.SAGE 把过刊数字化，1987 年以来的论文都带着 10.1177 前缀的 DOI，前身各卷在 Crossref 里另有自己的刊号。

• A total of 1,148 DOIs have been registered under the current ISSN, from the 1998 opening issue to volume 40 in 2026.现行刊号名下登记的 DOI 累计 1148 条，从 1998 年的创刊期一路排到 2026 年的第 40 卷。

• The publishing model is hybrid, and the 2025 Journal Citation Reports credit 19.42 percent of output to gold open access, with an open access option offered beside the subscription route.出版模式为混合开放获取，2025 年期刊引证报告把 19.42% 的产出记为金色开放获取，订阅之外另设开放获取选项。

• The 2025 metrics are a Journal Impact Factor of 2.4 and a five-year value of 2.9, with 1,416 citations, 57 articles and a cited half-life of 6.6 years.2025 年的指标是影响因子 2.4、五年影响因子 2.9，总被引 1416 次，年发文 57 篇，被引半衰期 6.6 年。

• Across three categories the journal stands at 38 of 60 in hardware and architecture, 120 of 185 in interdisciplinary applications and 59 of 146 in theory and methods.在三个类目中的位次分别是硬件与体系结构 60 种里的第 38 位，跨学科应用 185 种里的第 120 位，理论与方法 146 种里的第 59 位。

• Impact factors since 2022 read 2.82, 3.5, 2.5 and 2.4, easing back from the 2023 peak.2022 年以来的影响因子依次为 2.82 与 3.5，随后是 2.5 与 2.4，2023 年的高点之后逐步回落。

Editors

• The editor-in-chief posts are held jointly by Amir H Gandomi of the University of Technology Sydney and Ali Mirjalili of Torrens University, both in Australia, with no deputy or associate editors listed between them and the board.主编一职由悉尼科技大学（University of Technology Sydney）的（Amir H Gandomi）与托伦斯大学（Torrens University）的（Ali Mirjalili）共同担任，两人都在澳大利亚，主编与编委会之间不设副编辑或助理编辑。

• SAGE has opened a search for a new editor-in-chief, with applications due on 29 April 2026 and a three-year term beginning in the middle of that year.SAGE 已公开为新任主编征募人选，申请截止于 2026 年 4 月 29 日，任期三年，从当年年中起算。

• The call restates the remit and adds artificial intelligence and machine learning to modelling and simulation as fields the journal now seeks.征募启事复述了刊物范围，并在建模与模拟之外，把人工智能与机器学习列为当前重点约稿的方向。

• Sixteen editors sit on the board beside the two chiefs, distributed as ten in the United States, three in Australia, two in Japan, and one each in Italy, Britain and South Korea.两位主编之外，编委会列有 16 位编辑，其中美国 10 位，澳大利亚 3 位，日本 2 位，意大利、英国与韩国各 1 位。

• National laboratories carry much of the computing weight, through Ann Almgren and Zarija Lukic at Berkeley Lab, Nicola Ferrier at Argonne and Keita Teranishi at Oak Ridge.国家实验室承担了相当一部分计算分量，伯克利实验室的（Ann Almgren）与（Zarija Lukic）在列，阿贡实验室的（Nicola Ferrier）与橡树岭实验室的（Keita Teranishi）同样在内。

• Supercomputing centres and atmospheric computing appear through Sandra Gesing at the San Diego Supercomputer Center and Bryan Flynt at the University Corporation for Atmospheric Research.超算中心与大气计算由圣迭戈超级计算中心的（Sandra Gesing）与国家大气研究大学联盟（UCAR）的（Bryan Flynt）代表。

• Universities fill the rest, among them Charbel Farhat at Stanford, Tom Deakin at Bristol, Daisuke Takahashi at Tsukuba, John Taylor at the Australian National University, Sung-Eui Yoon at KAIST and Layne T Watson at Virginia Tech.其余席位来自高校，斯坦福的（Charbel Farhat）与布里斯托的（Tom Deakin）在列，筑波大学的（Daisuke Takahashi）与澳大利亚国立大学的（John Taylor）同样在内，韩国科学技术院的（Sung-Eui Yoon）与弗吉尼亚理工的（Layne T Watson）也在其中。

• Alice E. Koniges of the University of Hawaii, Ana Gainaru of Vanderbilt, Massimo Bernaschi of the National Research Council of Italy and Miwako Tsuji of RIKEN complete the list.夏威夷大学的（Alice E. Koniges）与范德堡大学的（Ana Gainaru）补位，意大利国家研究委员会的（Massimo Bernaschi）与理化学研究所的（Miwako Tsuji）收尾。

• Seven emeritus editors are listed below them, including William Gropp of Illinois, David Keyes of KAUST and A. J. G. Hey, now with Microsoft.名单之下另列七位荣休编辑，伊利诺伊大学的（William Gropp）与沙特阿卜杜拉国王科技大学的（David Keyes）在列，现任职于微软的（A. J. G. Hey）同样在内。

• The publisher reports a median of five days to a first decision, 118 days through peer review and 36 days from acceptance to publication, against an acceptance rate of 4.6 percent.出版方公布的中位时间是首次决定 5 天，整个同行评审 118 天，从录用到发表 36 天，录用率为 4.6%。

• Manuscripts arrive through ScholarOne at mc.manuscriptcentral.com/ijhpca, and every accepted paper has passed external referees.稿件经 ScholarOne 系统投稿，入口为 mc.manuscriptcentral.com/ijhpca，录用的论文都经过外部审稿人评议。

• Special issues are common and open with a guest editor's note, as with the 2026 collection on system-level innovations for performance and fairness at scale.专刊出现得频繁，每辑前面有一篇客座编辑引言，2026 年关于大规模性能与公平性的系统级创新专辑即是一例。

• Review articles carry the same weight as research papers, and recent examples include a two-part survey of hardware technology trends and a comprehensive survey of the conjugate gradient benchmark.综述文章与研究论文同等对待，近年有分两期刊出的硬件技术趋势综述，也有共轭梯度基准的全面调查。

• Papers report measured performance on real machines, so timing data, scaling curves and energy counts are routine parts of an issue.刊物上的论文报告真实机器上的实测性能，计时数据与扩展曲线是常规内容，能耗计数同样常见。

• The editorial office works with SAGE's production team in Thousand Oaks, and accepted articles appear online before they are assigned to a print issue.编辑部与 SAGE 在千橡市的生产团队协作，录用文章先在线发表，随后编入纸质期号。