ArtsAutosBooksBusinessEducationEntertainmentFamilyFashionFoodGamesGenderHealthHolidaysHomeHubPagesPersonal FinancePetsPoliticsReligionSportsTechnologyTravel

VSCSE Notes on Parallel Computing, Day 2

Updated on July 11, 2012

Task Parallelism Model

The key concept in parallel software design is the conversion of serial work into parallel computations. A dominant method for doing this efficiently is via reduction and reverse reduction trees, which exploit the "divide and conquer" strategy to spread work across a block of threads. This is the idea behind Google's MapReduce (and Hadoop).

Basic Requirements

Parallel computation tasks are only possible if subtasks can be carried out without order preference. The input data must be divided into chunks and each chunk distributed to a thread. The final result is a summary of the actions carried out on each thread. This summary is carried out through "reduction".

Reduction Tree

Each thread in a parallel algorithm processes its input chunk and writes output to a private store on the device. Reduction trees access and combine these private stores into the final output result. Feasible reduction operations must be associative and commutative, as well as possess a well-defined identity. For example, in summation, the identity is zero and the reduction scheme might take the form of a tournament. Other examples include min, max, and product.

Example: Max of an Array

Suppose we have an numeric array of length N and we wish to determine the maximum element contained therein. If we perform sequential comparisons, we will carry out N-1 steps, resulting in O(n) complexity.

Parallel implementation could involve pair-wise comparison. Graphically, this method looks like a tournament bracket (think "March Madness"). Assuming we have N threads, we will carry out N-1 comparisons distributed over log(N) steps. The average parallelism, which is the work per step, is (N-1)/log(N). Note that this is a work-efficient approach since the same total work is carried out as with sequential analysis (this is often not the case).

Thread Divergence

One problem with the tournament-style parallel implementation is thread divergence. As the tournament progresses, threads that "lose" individual comparisons become inactive. Due to the nature of how instructions are communicated to threads under the CUDA framework, this causes a significant performance hit. By optimizing consistent thread activity within warps, which are blocks of threads, divergences can be avoided (or at least minimized).

Keynote from David Kirk

This year, NVIDIA unveiled CUDA 5.0 and the new Kepler GPU architecture with the GK110 chip. The GK110 is the largest chip they've built and is targeted at enthusiasts, professionals, and the HPC community. With over 7 billion transistors, the GK110 is capable of over 1 teraflop. This is quite a milestone. The strategy behind the new design is an increase in floating precision units, but reduction in clock speed. This cuts the heat, which is, as always, a primary issue. They've also increased the number of registers per thread to 255 (Fermi has 63). The major innovation that I found interesting is the concept of dynamic parallelism. This means less I/O with the CPU and more autonomy for the GPU. While previously the CPU controlled the launching and syncing of new kernel routines, now the GPU can launch kernels from within CPU-initiated kernels.

Another note that brought a few smiles is the integration of the LLVM open source compiler in CUDA 5.0. This is perhaps a sign of progress from NVIDIA, who has come under a bit of criticism lately from certain leaders (gods?) of the Linux community.


    0 of 8192 characters used
    Post Comment

    No comments yet.


    This website uses cookies

    As a user in the EEA, your approval is needed on a few things. To provide a better website experience, uses cookies (and other similar technologies) and may collect, process, and share personal data. Please choose which areas of our service you consent to our doing so.

    For more information on managing or withdrawing consents and how we handle data, visit our Privacy Policy at:

    Show Details
    HubPages Device IDThis is used to identify particular browsers or devices when the access the service, and is used for security reasons.
    LoginThis is necessary to sign in to the HubPages Service.
    Google RecaptchaThis is used to prevent bots and spam. (Privacy Policy)
    AkismetThis is used to detect comment spam. (Privacy Policy)
    HubPages Google AnalyticsThis is used to provide data on traffic to our website, all personally identifyable data is anonymized. (Privacy Policy)
    HubPages Traffic PixelThis is used to collect data on traffic to articles and other pages on our site. Unless you are signed in to a HubPages account, all personally identifiable information is anonymized.
    Amazon Web ServicesThis is a cloud services platform that we used to host our service. (Privacy Policy)
    CloudflareThis is a cloud CDN service that we use to efficiently deliver files required for our service to operate such as javascript, cascading style sheets, images, and videos. (Privacy Policy)
    Google Hosted LibrariesJavascript software libraries such as jQuery are loaded at endpoints on the or domains, for performance and efficiency reasons. (Privacy Policy)
    Google Custom SearchThis is feature allows you to search the site. (Privacy Policy)
    Google MapsSome articles have Google Maps embedded in them. (Privacy Policy)
    Google ChartsThis is used to display charts and graphs on articles and the author center. (Privacy Policy)
    Google AdSense Host APIThis service allows you to sign up for or associate a Google AdSense account with HubPages, so that you can earn money from ads on your articles. No data is shared unless you engage with this feature. (Privacy Policy)
    Google YouTubeSome articles have YouTube videos embedded in them. (Privacy Policy)
    VimeoSome articles have Vimeo videos embedded in them. (Privacy Policy)
    PaypalThis is used for a registered author who enrolls in the HubPages Earnings program and requests to be paid via PayPal. No data is shared with Paypal unless you engage with this feature. (Privacy Policy)
    Facebook LoginYou can use this to streamline signing up for, or signing in to your Hubpages account. No data is shared with Facebook unless you engage with this feature. (Privacy Policy)
    MavenThis supports the Maven widget and search functionality. (Privacy Policy)
    Google AdSenseThis is an ad network. (Privacy Policy)
    Google DoubleClickGoogle provides ad serving technology and runs an ad network. (Privacy Policy)
    Index ExchangeThis is an ad network. (Privacy Policy)
    SovrnThis is an ad network. (Privacy Policy)
    Facebook AdsThis is an ad network. (Privacy Policy)
    Amazon Unified Ad MarketplaceThis is an ad network. (Privacy Policy)
    AppNexusThis is an ad network. (Privacy Policy)
    OpenxThis is an ad network. (Privacy Policy)
    Rubicon ProjectThis is an ad network. (Privacy Policy)
    TripleLiftThis is an ad network. (Privacy Policy)
    Say MediaWe partner with Say Media to deliver ad campaigns on our sites. (Privacy Policy)
    Remarketing PixelsWe may use remarketing pixels from advertising networks such as Google AdWords, Bing Ads, and Facebook in order to advertise the HubPages Service to people that have visited our sites.
    Conversion Tracking PixelsWe may use conversion tracking pixels from advertising networks such as Google AdWords, Bing Ads, and Facebook in order to identify when an advertisement has successfully resulted in the desired action, such as signing up for the HubPages Service or publishing an article on the HubPages Service.
    Author Google AnalyticsThis is used to provide traffic data and reports to the authors of articles on the HubPages Service. (Privacy Policy)
    ComscoreComScore is a media measurement and analytics company providing marketing data and analytics to enterprises, media and advertising agencies, and publishers. Non-consent will result in ComScore only processing obfuscated personal data. (Privacy Policy)
    Amazon Tracking PixelSome articles display amazon products as part of the Amazon Affiliate program, this pixel provides traffic statistics for those products (Privacy Policy)