Dax: A Massively Threaded Visualiza5on and Analysis Toolkit for Extreme Scale

Transcription

1 Dax: A Massively Threaded Visualiza5on and Analysis Toolkit for Extreme Scale GPU Technology Conference March 26, 2014 Kenneth Moreland Sandia Na5onal Laboratories Robert Maynard Kitware, Inc. Sandia National Laboratories is a multi-program laboratory managed and operated by Sandia Corporation, a wholly owned subsidiary of Lockheed Martin Corporation, for the U.S. Department of Energy s National Nuclear Security Administration under contract DE-AC04-94AL SAND NO P

2 A Toolkit for Scien5fic Visualiza5on

3 Sci Vis a Conglomera5on of Many Geometric and Other Tools

4 Mo#va#on

5 Slide of Doom System Parameter Factor Change System Peak 2 PetaFLOPS 1 ExaFLOP 500 Power 6 MW 20 MW 3 System Memory 0.3 PB PB Total Concurrency 225K 1B 10 1B , ,000 Node Performance 125 GF 1 TF 10 TF 8 80 Node Concurrency 12 1,000 10, Network BW 1.5 KB/s 100 GB/s 1000 GB/s System Size (nodes) 18,700 1,000, , I/O Capacity 15 PB PB I/O BW 0.2 TB/s TB/s 10 30

6 Slide of Doom System Parameter Factor Change System Peak 2 PetaFLOPS 1 ExaFLOP 500 Power 6 MW 20 MW 3 System Memory 0.3 PB PB Total Concurrency 225K 1B 10 1B , ,000 Node Performance 125 GF 1 TF 10 TF 8 80 Node Concurrency 12 1,000 10, Network BW 1.5 KB/s 100 GB/s 1000 GB/s System Size (nodes) 18,700 1,000, , I/O Capacity 15 PB PB I/O BW 0.2 TB/s TB/s 10 30

7 Extreme Scale is Threads, Threads, Threads! Jaguar XT5 Titan XK7 Exascale* Cores 224, ,008 and 18,688 gpu 1 billion Concurrency 224,256 way million way billion way Memory 300 Terabytes 700 Terabytes 128 Petabytes To succeed at extreme scale, you need to consider the finest possible level of concurrency Expect each thread to process exactly one element Disallow communica5on among threads *Source: Scien5fic Discovery at the Exascale, Ahern, Shoshani, Ma, et al.

8 Project Goals Reduce the challenges of wri5ng highly concurrent algorithms. Everybody who learns concurrency thinks they understand it, ends up finding mysterious races they thought weren t possible, and discovers that they didn t actually understand it yet aher all. Herb Suker

9 Approach

10 Functor Mapping [Baker, et al. 2010] functor()

11 Applied to Topologies functor()

12 Applied to Topologies functor()

13 Framework

14 Dax Framework Control Environment Execu5on Environment dax::cont dax::exec

15 Dax Framework Control Environment Execu5on Environment Grid Topology Array Handle Invoke dax::cont dax::exec

16 Dax Framework Control Environment Grid Topology Array Handle Invoke Execu5on Environment Cell Opera5ons Field Opera5ons Basic Math Make Cells Worklet dax::cont dax::exec

17 Dax Framework Control Environment Grid Topology Array Handle Invoke Execu5on Environment Cell Opera5ons Field Opera5ons Basic Math Make Cells Worklet dax::cont dax::exec

18 Dax Framework Control Environment Grid Topology Array Handle Invoke Device Adapter Allocate Transfer Schedule Sort Execu5on Environment Cell Opera5ons Field Opera5ons Basic Math Make Cells Worklet dax::cont dax::exec

19 Device Adapter Contents Tag (struct DeviceAdapterFoo { };) Execu5on Array Manager Control Environment Transfer Execu5on Environment Schedule functor Schedule worklet worklet worklet worklet worklet worklet worklet functor Compute Scan Compute Sort Other Support algorithms Stream compact, copy, parallel find, unique

20 Anatomy of a Worklet

21 struct Sine: public dax::exec::workletmapfield { typedef void ControlSignature(Field(In), Field(Out)); typedef _2 ExecutionSignature(_1); DAX_EXEC_EXPORT dax::scalar operator()(dax::scalar v) const { return dax::math::sin(v); } };

22 struct Sine: public dax::exec::workletmapfield { typedef void ControlSignature(Field(In), Field(Out)); typedef _2 ExecutionSignature(_1); DAX_EXEC_EXPORT dax::scalar operator()(dax::scalar v) const { return dax::math::sin(v); } };

23 struct Sine: public dax::exec::workletmapfield { typedef void ControlSignature(Field(In), Field(Out)); typedef _2 ExecutionSignature(_1); DAX_EXEC_EXPORT dax::scalar operator()(dax::scalar v) const { return dax::math::sin(v); } };

24 struct Sine: public dax::exec::workletmapfield { typedef void ControlSignature(Field(In), Field(Out)); typedef _2 ExecutionSignature(_1); DAX_EXEC_EXPORT dax::scalar operator()(dax::scalar v) const { return dax::math::sin(v); } };

25 struct Sine: public dax::exec::workletmapfield { typedef void ControlSignature(Field(In), Field(Out)); typedef _2 ExecutionSignature(_1); DAX_EXEC_EXPORT dax::scalar operator()(dax::scalar v) const { return dax::math::sin(v); } };

26 dax::cont::arrayhandle<dax::scalar> inputhandle = dax::cont::make_arrayhandle(input); dax::cont::arrayhandle<dax::scalar> sineresult; Control Environment Execu5on Environment dax::cont::dispatchermapfield<sine> dispatcher; dispatcher.invoke(inputhandle, sineresult); struct Sine: public dax::exec::workletmapfield { typedef void ControlSignature(Field(In), Field(Out)); typedef _2 ExecutionSignature(_1); DAX_EXEC_EXPORT dax::scalar operator()(dax::scalar v) const { return dax::math::sin(v); } };

27 struct Zip2: public dax::exec::workletmapfield { typedef void ControlSignature(Field(In), Field(In), Field(Out)); typedef _3 ExecutionSignature(_1, _2); DAX_EXEC_EXPORT dax::vector2 operator()(dax::scalar x, dax::scalar y) const { return dax::make_vector2(x, y); } };

28 struct ImagToPolar: public dax::exec::workletmapfield { typedef void ControlSignature(Field(In), Field(In), Field(Out), Field(Out)); typedef void ExecutionSignature(_1, _2, _3, _4); DAX_EXEC_EXPORT void operator()(dax::scalar real, dax::scalar imaginary, dax::scalar &magnitude, dax::scalar &phase) const { magnitude = dax::math::magnitude( dax::make_vector2(real, imaginary)); phase = dax::math::atan2(imaginary, real); }

29 struct Advect: public dax::exec::workletmapfield { typedef void ControlSignature( Field(In), Field(In), Field(In), Field(Out), Field(Out), Field(Out), Field(Out)); typedef void ExecutionSignature( _1, _2, _3, _4, _5, _6, _7); DAX_EXEC_EXPORT void operator()(dax::vector3 startposition, dax::vector3 startvelocity, dax::vector3 acceleration, dax::vector3 &endposition, dax::vector3 &endvelocity, dax::scalar &rotation, dax::scalar &angularvelocity) const {... }

30 struct Contour: public dax::exec::workletinterpolatedcell { typedef void ControlSignature(Topology, Geometry(Out), Field(Point,In)); typedef void ExecutionSignature( Vertices(_1), _2, _3, VisitIndex); template<typename CellTag> DAX_EXEC_EXPORT void operator()( const CellVertices<CellTag>& verts, InterpolatedCellPoints<CellTagTriangle>& outcell, const CellField<dax::Scalar,CellTag> &values, dax::id inputcellvisitindex) const {...

31 Example Code Mandelbulb Fractal

32 Mandelbulb nth power of 3D Vector where

33

34 Mandelbulb Itera5on Fractal created by itera5ng on power opera5on, offseqng by input coordinate. Our implementa5on counts how many itera5ons to escape.

35

36

37

38 Acknowledgements This work was supported by the DOE Office of Science, Advanced Scien5fic Compu5ng Research, under award number , program manager Lucy Nowell. Addi5onal support by the Director, Office of Advanced Scien5fic Compu5ng Research, Office of Science, of the U.S. Department of Energy under Contract No , through the Scien5fic Discovery through Advanced Compu5ng (SciDAC) Ins5tute of Scalable Data Management, Analysis and Visualiza5on. Sandia Na5onal Laboratories is a mul5- program laboratory managed and operated by Sandia Corpora5on, a wholly owned subsidiary of Lockheed Mar5n Corpora5on, for the U.S. Department of Energy s Na5onal Nuclear Security Administra5on under contract DE- AC04-94AL85000.