Language Selection

English French German Italian Portuguese Spanish

Nvidia scientist calls for expanded research into parallelism

Filed under
Hardware

Expanded research is needed into techniques for identifying and preserving parallelism in chip applications, according to chip maker Nvidia's chief scientist.

Comparing the futures of general-purpose CPUs and graphics processors (GPUs), David Kirk told the 17th Hot Chips conference on the Stanford Univeristy campus here on Tuesday (Aug. 16) that a crisis looms in programming technology. He said this could not only blight the future CPU market but also bring an end to improvements in graphics performance despite continued improvements in GPUs.

"If we look at the situation of general-purpose CPUs," Kirk said, "we see architects moving rapidly to multithreading and multicore designs. But we don't see a lot more threads to run. What parallelism there may be in algorithms is often concealed or lost altogether in the programming process."

Kirk grounded his pessimism in the experiences of game developers trying to exploit new multicore CPU chips. "We are already seeing some games limited by CPU throughput. We can render images faster than the CPU can send us the data," Kirk said. "But when game developers try to use dual-core CPUs to help, we have seen virtually no benefit to the second CPU core. And if the developer doesn't clearly understand the interactions of the cores with the caches, we have seen the application actually run slower on a dual-core machine."

Kirk contrasted this situation against the entirely different structure inside the GPU. "Graphics has been called embarrassingly parallel," he said. "In effect, each stage in our pipeline, each vertex in the scene and each pixel in the image is independent. And we have put a lot of effort into not concealing this parallelism with programming."

This allows a GPU developer to simply add more vertex processors and shading engines to handle more vertices and more pixels in parallel, as process technology allows. "We are limited by chip area, not by parallelism," Kirk observed.

Full Story.

More in Tux Machines

OSS in the Back End

  • Open Source NFV Part Four: Open Source MANO
    Defined in ETSI ISG NFV architecture, MANO (Management and Network Orchestration) is a layer — a combination of multiple functional entities — that manages and orchestrates the cloud infrastructure, resources and services. It is comprised of, mainly, three different entities — NFV Orchestrator, VNF Manager and Virtual Infrastructure Manager (VIM). The figure below highlights the MANO part of the ETSI NFV architecture.
  • After the hype: Where containers make sense for IT organizations
    Container software and its related technologies are on fire, winning the hearts and minds of thousands of developers and catching the attention of hundreds of enterprises, as evidenced by the huge number of attendees at this week’s DockerCon 2016 event. The big tech companies are going all in. Google, IBM, Microsoft and many others were out in full force at DockerCon, scrambling to demonstrate how they’re investing in and supporting containers. Recent surveys indicate that container adoption is surging, with legions of users reporting they’re ready to take the next step and move from testing to production. Such is the popularity of containers that SiliconANGLE founder and theCUBE host John Furrier was prompted to proclaim that, thanks to containers, “DevOps is now mainstream.” That will change the game for those who invest in containers while causing “a world of hurt” for those who have yet to adapt, Furrier said.
  • Is Apstra SDN? Same idea, different angle
    The company’s product, called Apstra Operating System (AOS), takes policies based on the enterprise’s intent and automatically translates them into settings on network devices from multiple vendors. When the IT department wants to add a new component to the data center, AOS is designed to figure out what needed changes would flow from that addition and carry them out. The distributed OS is vendor-agnostic. It will work with devices from Cisco Systems, Hewlett Packard Enterprise, Juniper Networks, Cumulus Networks, the Open Compute Project and others.
  • MapR Launches New Partner Program for Open Source Data Analytics
    Converged data vendor MapR has launched a new global partner program for resellers and distributors to leverage the company's integrated data storage, processing and analytics platform.
  • A Seamless Monitoring System for Apache Mesos Clusters
  • All Marathons Need a Runner. Introducing Pheidippides
    Activision Publishing, a computer games publisher, uses a Mesos-based platform to manage vast quantities of data collected from players to automate much of the gameplay behavior. To address a critical configuration management problem, James Humphrey and John Dennison built a rather elegant solution that puts all configurations in a single place, and named it Pheidippides.
  • New Tools and Techniques for Managing and Monitoring Mesos
    The platform includes a large number of tools including Logstash, Elasticsearch, InfluxDB, and Kibana.
  • BlueData Can Run Hadoop on AWS, Leave Data on Premises
    We've been watching the Big Data space pick up momentum this year, and Big Data as a Service is one of the most interesting new branches of this trend to follow. In a new development in this space, BlueData, provider of a leading Big-Data-as-a-Service software platform, has announced that the enterprise edition of its BlueData EPIC software will run on Amazon Web Services (AWS) and other public clouds. Essentially, users can now run their cloud and computing applications and services in an Amazon Web Services (AWS) instance while keeping data on-premises, which is required for some companies in the European Union.

today's howtos

Industrial SBC builds on Raspberry Pi Compute Module

On Kickstarter, a “MyPi” industrial SBC using the RPi Compute Module offers a mini-PCIe slot, serial port, wide-range power, and modular expansion. You might wonder why in 2016 someone would introduce a sandwich-style single board computer built around the aging, ARM11 based COM version of the original Raspberry Pi, the Raspberry Pi Compute Module. First off, there are still plenty of industrial applications that don’t need much CPU horsepower, and second, the Compute Module is still the only COM based on Raspberry Pi hardware, although the cheaper, somewhat COM-like Raspberry Pi Zero, which has the same 700MHz processor, comes close. Read more

DAISY: A Linux-compatible text format for the visually impaired

If you're blind or visually impaired like I am, you usually require various levels of hardware or software to do things that people who can see take for granted. One among these is specialized formats for reading print books: Braille (if you know how to read it) or specialized text formats such as DAISY. Read more