BEGIN:VCALENDAR VERSION:2.0 PRODID:Linklings LLC BEGIN:VTIMEZONE TZID:America/Chicago X-LIC-LOCATION:America/Chicago BEGIN:DAYLIGHT TZOFFSETFROM:-0600 TZOFFSETTO:-0500 TZNAME:CDT DTSTART:19700308T020000 RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU END:DAYLIGHT BEGIN:STANDARD TZOFFSETFROM:-0500 TZOFFSETTO:-0600 TZNAME:CST DTSTART:19701101T020000 RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU END:STANDARD END:VTIMEZONE BEGIN:VEVENT DTSTAMP:20260522T150153Z LOCATION:D175 DTSTART;TZID=America/Chicago:20181112T090000 DTEND;TZID=America/Chicago:20181112T173000 UID:submissions.supercomputing.org_SC18_sess176@linklings.com SUMMARY:LLVM-HPC2018: The Fifth Workshop on the LLVM Compiler Infrastructu re in HPC DESCRIPTION:OpenMP GPU Offload in Flang and LLVM\n\nGraphics Processing Un its (GPUs) have been widely adopted to accelerate the execution of High Pe rformance Computing (HPC) workloads due to their enormous computational th roughput, ability to execute a large number of threads inside SIMD groups in parallel, and their use of multithreaded hardware to ...\n\n\nGuray Oze n, Simone Atzeni, Michael Wolfe, Annemarie Southwell, and Gary Klimowicz ( Nvidia Corporation)\n---------------------\nLLVM-HPC2018: Final Discussion \n\nHal Finkel (Argonne National Laboratory)\n---------------------\nClacc : Translating OpenACC to OpenMP in Clang\n\nOpenACC was launched in 2010 a s a portable programming model for heterogeneous accelerators. Although v arious implementations already exist, no extensible, open-source, producti on-quality compiler support is available to the community. This deficienc y poses a serious risk for HPC application devel...\n\n\nJoel Denny, Seyon g Lee, and Jeffrey Vetter (Oak Ridge National Laboratory)\n--------------- ------\nA Study of OpenMP Device Offloading in LLVM: Correctness and Consi stency\n\nTo leverage widely available accelerators, OpenMP has introduced device constructs. Device constructs simplify the development of heteroge neous parallel programs and improve the performance. Many compilers includ ing Clang already have support for device constructs, but there exist few documentations...\n\n\nLechen Yu (Georgia Institute of Technology)\n------ ---------------\nPointers Inside Lambda Closure Objects in OpenMP Target O ffload Regions\n\nWith the diversification of HPC architectures beyond tra ditional CPU-based clusters, a number of new frameworks for performance po rtability across architectures have arisen. One way of implementing such f rameworks is to use C++ templates and lambda expressions to design loop-li ke functions. However,...\n\n\nDavid Truby (University of Warwick) and Ste ven Wright (University of York)\n---------------------\nUser-Directed Loop -Transformations in Clang\n\nDirectives for the compiler such as pragmas c an help programmers to separate an algorithm's semantics from its optimiza tion. This keeps the code understandable and easier to optimize for differ ent platforms. Simple transformations such as loop unrolling are already i mplemented in most mainstream com...\n\n\nMichael Kruse and Hal Finkel (Ar gonne National Laboratory, Argonne Leadership Computing Facility)\n------- --------------\nPInT: Pattern Instrumentation Tool for Analyzing and Class ifying HPC Applications\n\nThe relationship of application performance to its required development effort plays an important role in today’s budget- oriented HPC environment. This effort-performance relationship is especial ly affected by the structure and characterization of an HPC application. W e aim at a classification of HP...\n\n\nFabian Schlebusch, Yannik Müller, Sandra Wienke, Julian Miller, and Matthias Müller (RWTH Aachen University) \n---------------------\nOP2-Clang: A Source-to-Source Translator Using Cl ang/LLVM LibTooling\n\nDomain Specific Languages or Active Library framewo rks have recently emerged as an important method for gaining performance p ortability, where an application can be efficiently executed on a wide ran ge of HPC architectures without significant manual modifications. Embedded DSLs such as OP2, provides...\n\n\nGábor Dániel Balogh (Pázmány Péter Cat holic University, Hungary); Gihan Ravideva Mudalige (University of Warwick ); Istvan Z. Reguly (Pázmány Péter Catholic University, Hungary); and Samu el Antao and Carlo Bertolli (IBM)\n---------------------\nWorkshop Afterno on Break\n\nHal Finkel (Argonne National Laboratory)\n-------------------- -\nCompiler Optimization for Heterogeneous Locality and Homogeneous Parall elism in OpenCL and LLVM\n\nHeterogeneous platforms may include accelerato rs such as Digital Signal Processors (DSP’s) that employ SW-controlled scr atch-pad memories instead of, or in addition to standard HW-cached memory. Controlling scratch-pads efficiently typically requires tiling and pipeli ning loops, thereby optimizing f...\n\n\nDorit Nuzman (Intel Corporation); Michael Zuckerman (Mobileye, an Intel Company); and Ayal Zaks (Intel Corp oration)\n---------------------\nFunction/Kernel Vectorization via Loop Ve ctorizer\n\nCurrently, there are three vectorizers in the LLVM trunk: Loop Vectorizer, SLP Vectorizer, and Load-Store Vectorizer. There is a need fo r vectorizing functions/kernels: 1) Function calls are an integral part of programming real world application code and we cannot always rely on full y inlining them....\n\n\nMatt Masten, Evgeniy Tyurin, Konstantina Mitropou lou, Hideki Saito, and Eric Garcia (Intel Corporation)\n------------------ ---\nKeynote: Glow: An Optimizing Compiler for High-Performance Machine Le arning\n\nMachine learning is an increasingly large fraction of datacenter workloads, making efficient execution of ML models a priority for industr y. At the same time, the slow down of Moore's Law has created space for a plethora of innovative hardware designs to wring maximum performance from each transisto...\n\n\nBert Maher (Facebook)\n---------------------\nLLVM and the Automatic Vectorization of Loops Invoking Math Routines: -fsimdmat h\n\nThe vectorization of loops invoking math function is an important opt imization that is available in most commercial compilers. This paper descr ibes a new command line option, -fsimdmath, available in Arm Compiler for HPC, that enables auto-vectorization of math functions in C and C++ code, and that ...\n\n\nFrancesco Petrogalli and Paul Walker (ARM Ltd)\n-------- -------------\nWorkshop Lunch (on your own)\n\nHal Finkel (Argonne Nationa l Laboratory)\n---------------------\nWorkshop Morning Break\n\nHal Finkel (Argonne National Laboratory)\n---------------------\nIntroduction - LLVM -HPC2018: The Fifth Workshop on the LLVM Compiler Infrastructure in HPC\n\ nLLVM, winner of the 2012 ACM Software System Award, has become an integra l part of the software-development ecosystem for optimizing compilers, dyn amic-language execution engines, source-code analysis and transformation t ools, debuggers and linkers, and a whole host of programming-language and toolc...\n\n\nHal Finkel (Argonne National Laboratory)\n------------------ ---\nAIWC: OpenCL-Based Architecture Independent Workload Characterization \n\nMeasuring performance-critical characteristics of application workload s is important both for developers, who must understand and optimize the p erformance of codes, as well as designers and integrators of HPC systems, who must ensure that compute architectures are suitable for the intended w orkloads...\n\n\nBeau Johnston and Josh Milthorpe (Australian National Uni versity)\n---------------------\nChallenges of C++ Heterogeneous Programmi ng Using SYCL Implementation Experience: the Four Horsemen of the Apocalyp se\n\nThe C++ Direction Group has set a future direction for C++ and inclu des a guidance towards Heterogeneous C++. The introduction of the executor s TS means for the first time in C++ there will be a standard platform for writing applications which can execute across a wide range of architectur es includi...\n\n\nVictor Lomuller (Codeplay Software Ltd)\n\nTag: Program Transformation, Programming Systems\n\nRegistration Category: Workshop Re g Pass END:VEVENT END:VCALENDAR