March 2015 - Boost - lists.boost.org

GSoC 2015 Boost Compute Proposal
by Aditya Avinash Atluri 27 Mar '15

27 Mar '15

Hi, This is my proposal. Please suggest edits. Thank you!! :) *Personal Details:* Name: Aditya Atluri University: The George Washington University Major: Computer Engineering Degree: MS Email: adityaavinash1(a)gmail.com Homepage: adityaatluri.github.io Availability: 1. 7 days a week 2. 27 April to 28 August 3. Nothing. *Background Information:* *Educational Background:* I did my undergrad from Indian Institute of Technology, Dhanbad, India. Currently, I am pursuing my Masters at The George Washington University. The courses I have taken are, High Performance Computing, Advanced Computer Micro Architecture, Computer Graphics, Advanced Topics in Computer Graphics, VLSI Design, Low Power VLSI Design, Compiler, Machine Learning. I have did lots of projects in these courses which can be seen on my website and in my bio[1]. *Programming Background and Interests:* I am researcher, I program GPUs with IRs, writing work arounds and using a specific hardware for other applications (example, implementing ray tracing on rasterizer on GPUs, using tessellators for generating procedural textures). I wrote high productive parallel programming abstractions for GPUs[2]. I presented my work at GTC 2014. For this work, NVIDIA supported me with a couple of Tesla K40s. As CUDA, OpenGL, OpenCL, Metal, Thrust are C/C++ APIs I use C++ extensively. I also create databases on GPUs which is now in deep interest from NVIDIA, Google and MapD. I worked with Mesa, Open Source graphics drivers for Linux to build ARB_shader_atomic_counters for R600 backend. I also worked at IBM software labs, where I worked majorly on Java. *Why I opted Boost?* I use boost on daily basis as a part of PyCUDA. I write work arounds which involve some of boost libraries. The next thing I am interested in Boost is its Compute library. I have used other GPU STL style libraries but, none are as good as Compute (simple and elegant). I couldn't contribute to Compute via GSOC last year due to my CPT. I don't want to lose that chance again. This time, I want to take Boost.Compute to a whole new level. I like to program anything related to GPUs, compilers, drivers, APIs. *Interest in Boost.Compute:* As I have mentioned earlier, I am all-in for anything related to GPUs. Either graphics or compute libraries. I believe in Boost.Compute that it can be a widely used API, hence I want to make it more powerful and more useful. *Previous Work:* I work on graphics drivers on regular basis. I write loads of OpenGL compute and shader code (part of my TA job and Mesa testing on local systems). I am good at using APIs, making optimizations, less memory footage. For example, increasing the number of draw calls per second without stalling CPU, writing shaders with more ALU operations than Load/Store operations. Aligning data properly to memory width size, minimizing padding done by compilers by struct unrolling (to get exact multiple of word size). These are a few of the optimizations I do every day. I am currently working on running Metal Shading Language on Intel processors (using AVX).[3] *Plans beyond SoC*: I want to extend Compute to ARM and Qualcomm devices that support both OpenCL and ARM Neon. This brings all "compute-intense" APIs under one roof. C++ 98/03 - 03 C++ 98/14 - 03 C++ STL - 4 Boost C++ Libraries - (Knowledge: 2, Usage: 2) Git: 3 I program using Visual Studio and Xcode. Using Terminal on Linux. I am good with Doxygen. *Project Proposal:* Introduction: Boost Compute library is so far the best portable STL-like C++ library for accelerated computing leveraging the underlying parallel hardware. Boost Compute works well on OpenCL compatible desktop/notebook devices. In this project, we implement Boost Compute on mobile devices. Current mobile devices doesn’t provide good STL style abstractions that are accelerated by SIMD and OpenCL. With the introduction of Metal graphics API for iOS devices, development of compute application for non-iOS and iOS devices requires knowledge multiple skill sets. With this project, we bring all compute abstractions of different mobile devices under one roof that is, Boost.Compute. Project Goals: Make C++ extensions from Objective-C. Integrate device containers to Boost.Compute front end. Make Metal API data types visible to user through Boost.Compute. Implementing BOOST_COMPUTE_FUNCTION() to run custom kernels. Most important, implementing algorithms for Boost.Compute on iOS devices using Metal Shader Language. *Implementation:* Building Boost.Compute APIs for iOS devices. - Metal graphics APIs are used as backend to create Boost.Compute APIs in conjunction with Metal Shading Language Make C++ extensions from Objective-C. - Metal is Objective-C graphics API whereas Boost.Compute is C++ compute API. Hence a bridge between these two APIs should be made to use Compute on iOS devices. Hence, we build C++ extensions from Objective-C APIs. As C++ is also used in building iOS apps, writing Boost.Compute for iOS devices makes it easier to leverage compute performance from an iOS app. Build resource management Metal API backend for Boost.Compute - Here we develop infrastructure which make "algorithms" run on the hardware. Like, creating automated CommandQueues, CommandBuffers, Resources (buffers or textures) - As Metal APIs does not have ‘context’ related API, mapping can be done to Metal command buffer API. Some OpenCL APIs which are natively available (using in Boost.Compute) are not present in Metal API, appropriate mapping from Boost.Compute to Metal API commands will be found. Integrate device containers from Metal API to Boost.Compute front end. - Boost.Compute provide vector and array containers. These should be declared in Metal and integrated to Boost. Device containers include, device vectors, initializing vectors, moving vectors, similarly for arrays, etc., and their STL type functionality (.size(), .length()) - Implementing iterators for the containers should also be done. Make Metal API data types visible to user through Boost.Compute. - Mapping Metal data types to current Boost.Compute data types, while making them consistent with current Boost.Compute APIs. Certain new data types to Metal are being added which improve the performance and decrease memory footprint. Making these visible to the programmer and building building algorithms with them is done in this section. Implement algorithms for Boost.Compute in Metal Shading Language. - This makes the biggest challenge as Metal compute shaders support two modes of execution (fast and precise modes). Algorithms written in each mode should be iterated over to judge the best performance without losing precision for speed. Testing the existing Boost.Compute code on iOS devices and debugging for errors. - With Metal emulators not being available for Xcode, hardware testing will be done[4]. *Timeline:* The timeline is straight forward, for details check Implementation section. 27 April – 25 May: Community bonding, testing the Xcode and iOS infrastructure to find any possible build problems. Build a test run environment (iOS app) to check any other hardware/software issues in deploying Metal API code to iOS device. 25 May – 10 June: Wrap C++ Metal extensions with C++, submit patches. 10 June – 25 June: Implement data types and containers for Boost from C++ Metal APIs. Make appropriate changes to the code after getting feedback from community. Submit patches. 25 June – 3 July: Review community feedback and change patches accordingly. 3 July – 3 August: Write BUILD_COMPUTE_FUNCTION() and Boost.Compute algorithms on Metal Shading Language and integrate it to Boost.Compute Metal APIs. Submit patches. 3 August – 17 August: Write tests and debug by running them on iOS test app 17 August – 28 August: Cleaning up code and review with community. Available Hardware: iPhone 4s, 6Plus, Mac Mini. Mentor: Kyle Lutz (kyle.r.lutz(a)gmail.com) *Competency:* The project I started which creates the actual Metal API, is iaMetal (Metal for Intel processors)[3]. It does not uses Metal SL, it converts Metal SL to AVX extensions. [1]. https://github.com/adityaatluri/adityaatluri.github.co/blob/master/CV.md [2]. https://github.com/urutu/Urutu [3]. https://github.com/iaMetal/iaMetal [4]. https://devforums.apple.com/message/971605#971605 -- Regards, Aditya Avinash Atluri, Graduate Student, Electrical and Computer Engineering, The George Washington University, Washington, DC.

1 0

[asio][android][clang-x86] cannot locate symbol "___tls_get_addr"
by Dmitry Moskalchuk 27 Mar '15

27 Mar '15

As you can see at http://www.boost.org/development/tests/master/developer/asio.html, there are many fails on Android/x86, when tests are built by clang. There was ticket filed to CrystaX NDK tracker about that: https://tracker.crystax.net/issues/912. After some research, I've found reason of fails and created pull request for Boost.Asio: https://github.com/boostorg/asio/pull/17. Shortly speaking, there is intersection of several bugs - one is inability of clang to emulate thread-local variables without support from underlying runtime (gcc can do that), and second is misconfiguration in Boost.Asio headers. I've tested locally with change from https://github.com/boostorg/asio/pull/17 and all '___tls_get_addr' errors disappeared. Now I'm waiting for that pull request to be merged into upstream, but I don't see any activity in Boost.Asio repository. Is there something wrong with my pull request? It's very simple so I wonder why it not yet merged? I'd like to see it merged to the upstream (including 'master') before 1.58 release - just because this bug cause not only Boost.Asio fails, but several others libraries too. -- Dmitry Moskalchuk

2 3

[GSoC] Proposal for uBLAS
by Ganesh Prasad 27 Mar '15

27 Mar '15

Hello everybody, Here is the final draft of my proposal for GSoC'15 and I have made it public. https://www.google-melange.com/gsoc/proposal/review/student/google/gsoc2015… The same text can be found on Gist at https://gist.github.com/GnsP/72dce0ac96394f54797c Reviews are requested and welcome. Best wishes Ganesh Prasad

1 0

No automatic super project commits at the moment.
by Daniel James 27 Mar '15

27 Mar '15

Hi, Github is currently going through a DDOS attack, which is why the super project isn't getting updated for new commits. I imagine testers will also have problems running. Daniel

1 0

GSOC 2015 Proposal
by Aditya Avinash Atluri 27 Mar '15

27 Mar '15

Hi, This is my proposal for GSoC 2015. Please suggest me some edits. Thank you!! :) https://www.google-melange.com/gsoc/proposal/review/student/google/gsoc2015… -- Regards, Aditya Avinash Atluri, Graduate Student, Electrical and Computer Engineering, The George Washington University, Washington, DC.

5 4

Proposal "Enhanced vector and deque containers" project (GSoC 2015)
by Nico Hartmann 27 Mar '15

27 Mar '15

Hello everyone, as a student with quite some experience developing C++ I finally decided to participate in an open source project instead of just writing all the code for myself. To do so I chose to join the boost community in the course of this year's Google Sommer of Code. Therefore I wrote a proposal for the "enhanced vector and deque containers" project that really drew my interest. I like to invite you to take a look at my proposal as well as the requested implementation of a small vector class (including unit tests) to prove my capability to write C++ code. Thank you very much Nico project announcement: https://svn.boost.org/trac/boost/wiki/SoC2015#a5.Enhancedvectoranddequecont… project proposal (GSoC 2015): https://www.google-melange.com/gsoc/proposal/review/student/google/gsoc2015… small vector class implementation on github: https://github.com/nh-/gsoc_vector

1 1

GSOC 2015 Proposal
by vivek dhiman 26 Mar '15

26 Mar '15

Hello all This is my proposal for GSoC 2015. http://www.google-melange.com/gsoc/proposal/public/google/gsoc2015/vdhiman/… It would be awesome if you guys could suggest some changes/edits. Thanks Vivek Dhiman

2 1

Boost.uBLAS a library for matrix computations Project: Data Frame
by Sahba E 26 Mar '15

26 Mar '15

Hello all, My name is Sahba Etezad. I'm studying masters of Computing Science in Simon Fraser University, Canada.I'm really interested in working on ublas library and matrix computations. Particularly, I'm really passionate about the data frame project. My thesis concerns graph embedding so I've been working with Boost graph library for over a year now. I've been also coding in Matlab for over 4 years, and I think my familiarity with matrix operations and all that can be an asset for this project. I've finished implementing the competency code. https://github.com/setezad/ublasAnd I know it's really last minute, but I'd really appreciate it if you would give me some advice on what I should include to provide a precise project proposal. It has been mentioned that a data frame similar to R should be implemented with extension to implementing the summary and aggregate functions. I studied the two links provided in the project description ( http://pandas.pydata.org/pandas-docs/dev/generated/pandas.DataFrame.html, and http://cran.r-project.org/web/packages/data.table/index.html ). Now my question is knowing that a data frame can have countless attributes and methods, should I specify what attributes and methods I'd like to implement? Or will it be decided later and need not to be proposed by me? Thank you very much. Regards,Sahba Etezad

1 0

[SORT] Parallel Algorithms
by Francisco José Tapia 26 Mar '15

26 Mar '15

Sorry by the delay, As promised, the sorting parallel algorithms On my computer (Quad Core) the GCC std::sort and stable_sort are slight faster than my version of intro_sort and merge_sort (stable sort is a synonym of merge_sort). In other computer with an Intel I3, my algorithms are slight faster than in the GCC version. Trying to optimize the parallel implementation. I modified the parallel version for to use the std::sort and std::stable_sort. The parallel_sort version response good, but the parallel_stable_sort increase the time around a 60% compared with the version with countertree::parallel_merge_sort . After many test, I discover the cause of the problem. The demonstration is in the test_stress_sort_01.cpp. In this program we have a big vector of 100 000 000 elements uint64_t. The program make parts of the same size for each HW core. On my computer 4 parts and 4 cores. The program execute the part sequentially checking the time, and after do the same but in parallel, each thread sort a part, and check the time. This is done with the std::stable sort and countertree::merge_sort. The data input is the same for all. The next lines are the result show max. OpenMP threads: 4 (4 processors) Sort of 100000000 elements in a vector --------------------- Random elements, few repeated ( rand() )----------------------- Stable Sort Part 0 secs :2.77955 Part 1 secs :2.75015 Part 2 secs :2.76068 Part 3 secs :2.76284 Parallel Sort secs :8.49796 Countertree stable sort Part 0 secs :2.72802 Part 1 secs :2.71614 Part 2 secs :2.72074 Part 3 secs :2.722 Parallel Sort secs :4.80692 Surprisingly the parallel sort with countertree::stable_sort is much more faster than the sort with std::stable_sort in the GCC 4.8 and 4.9 compiler. This is the secret of the speed of the countertree::parallel_merge_sort. This is very important, because when the size of the elements is small, the GCC and the countertree algorithms have similar performance. But when the size of the elements grow, the std::stable sort is slower than the countertree::merge_sort. This difference is increased in the parallel version. The next lines are part of benchmark_sort_05.cpp *************************************************************** Sort of N = 6250000 elements of size 128 ***************************************************************** Random elements, few repeated ( rand() )----------------------- STL std::sort :2.9188 secs countertree::intro_sort :2.87864 secs parallel::sort :2.46113 secs tbb::parallel_sort :1.86458 secs countertree::parallel_sort:1.89263 secs std::stable_sort :8.77963 secs countertree::merge_sort :5.70694 secs parallel::stable_sort :8.03946 secs countertree::parallel_merge_sort :4.90815 secs I have in mind a new version of the stable sort in order to improve the speed, reducing the number of internal copy of elements. I hope to have time in Christmas, but I can't guarantee. If the results are the expected I will send you You have too, a version of indirect_sort, which sort pointers to the elements, and at end sort the elements according the pointers. You can replace any sort algorithm by their indirect version, obtaining the same results, with the only difference of the time needed. You have a benchmark_indirect_sort.cpp where see the comparison of the time needed of several sort methods and their indirect version with elements of different size. This code is a preliminary version, done for my countertree library. If you consider useful, I will change the name space and document the code, the algorithms and the ideas obtained from the benchmarks. Please feel free for ask me any question or for to modify or change anything. The goal is to provide good sort methods to the users. In the zip file you have the code and a file with a description of each code. Yours

5 20

[Locale] Boost.Locale cannot Run in Visual Studio 2010
by Sarvagya Pant 26 Mar '15

26 Mar '15

I have been trying to learn Boost.Locale Module. Consider this http://www.boost.org/doc/libs/1_55_0/libs/locale/doc/html/conversions_8cpp-… example. The code gets successfully compiled but when I try to run it, it gives the error Unhandled exception at 0x74d34b32 in LearningCPP.exe: Microsoft C++ exception: std::bad_cast at memory location 0x00a6fa48.. I have attached the stack-trace of the program also. Here it is: KernelBase.dll!74d34b32() [Frames below may be incorrect and/or missing, no symbols loaded for KernelBase.dll] KernelBase.dll!74d34b32() > msvcr100d.dll!_nh_malloc_dbg(unsigned int nSize, int nhFlag, int nBlockUse, const char * szFileName, int nLine) Line 302 + 0x1d bytes C++ msvcr100d.dll!malloc(unsigned int nSize) Line 56 + 0x15 bytes C++ LearningCPP.exe!std::use_facet<boost::locale::converter<char> >(const std::locale & _Loc) Line 526 C++ LearningCPP.exe!boost::locale::to_upper<char>(const char * str, const std::locale & loc) Line 225 + 0x9 bytes C++ LearningCPP.exe!main() Line 170 + 0xc4 bytes C++ LearningCPP.exe!__tmainCRTStartup() Line 555 + 0x19 bytes C LearningCPP.exe!mainCRTStartup() Line 371 C kernel32.dll!75268543() ntdll.dll!776bac69() ntdll.dll!776bac3c() Did anyone faced this issue when using Boost.Locale in Visual Studio 2010? If so, how do I resolve this issue? *PS:* I have used option Use Unicode Character Set under the Character Set option of Visual Studio. I believe the libraries are correctly linked too. -- *sarvagya*

1 0