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

Top Linux Distros for Media Creation

I find it interesting how many existing Linux users don't realize there are specialized distributions just for media creation. These distributions come with a bundle of special media-centric applications, a real-time kernel and other tweaks provided by default. This article will provide a tour of these top Linux distros for media creation. I'm confident that even if you've heard of some of these distros, you might not be aware of what makes them unique when compared to a standard desktop Linux distribution. Read more

Portable Android SDR player supports DRM and DAB

Titus SDR’s Android-based “Titus II” Software Defined Radio receiver has a 7-inch touchscreen, a WiFi hotspot, and support for FM, AM, DRM, DAB, and DAB+. Titus SDR is prepping an Android-based wideband digital RF receiver with Software Defined Radio (SDR) capabilities and a hi-fi amplifier. Built around a 7-inch Android tablet, the portable, battery-powered Titus II is billed as the world’s first consumer SDR digital receiver, “bringing true multi-standard radio reception with DRM (AM & VHF bands), DAB(+) and core data applications.” Read more

Programming: OpenJ9, SCons, and Adafruit NeoPixel Stick

  • Some Early Tests Of The Eclipse OpenJ9 Java Virtual Machine
    With IBM's newly open-sourced J9 Java Virtual Machine as the Eclipse OpenJ9, I've run some quick benchmarks to get an idea how its performance is comparing to the de facto Java Virtual Machine, Hotspot.
  • SCons 3.0 Released
    For those that haven't jumped fully on the Meson build system bandwagon, the SCons 3.0 software construction utility is now available.
  • Small Glowing Thing
    Quite a while ago I obtained an Adafruit NeoPixel Stick. It was cheap enough to be an impulse buy but it took me some time to get around to actually doing something with it. I’ve been wanting to play a little more with the ATtiny range of microcontrollers so these things seemed to go together nicely. It turns out that getting an ATtiny programmed is actually rather simple using an Arduino as an ISP programmer. I’ve written up some notes on the procedure at the 57North Hacklab wiki.

Kernel: Next Linux kernel 4.13 RC, the Linux Foundation and Graphics News

  • Linus Torvalds Kicks Off Development of Linux Kernel 4.14, the Next LTS Release
    A day early than expected, Linux creator Linus Torvalds cautiously kicked off the development of the Linux 4.14 kernel series, which looks to be the next LTS (Long Term Support) branch, with the first Release Candidate (RC) milestone. That's right, two weeks after the release of Linux kernel 4.13, which is currently the most stable and advanced kernel series, being adopted by more and more GNU/Linux distributions each day, the first RC development snapshot of Linux kernel 4.14 is ready for public testing, officially closing the merge window. And it looks like some core new functionality will be implemented in this release.
  • Linux Foundation wants to promote sustainable open source development with new initiatives
    During last week’s Open Source Summit North America in Los Angeles, the Linux Foundation announced a series of projects designed to promote sustainability and growth in open source development. We wrote last week about their “Open Source Guides for the Enterprise,” which will see a series of guides by professionals from many different organizations released over the next few months. Following that, the foundation announced the Community Health Analytics for Open Source Software, or CHAOSS, project. With CHAOSS, the Linux Foundation wants to provide a platform for measuring and analyzing open source projects. The foundation also announced that it has granted a CII security badge to 100 projects through a voluntary process for open source projects to prove their security measures stack up professionally.
  • Intel ANV Lands New Vulkan 1.0.61 Extensions, Android Prep Support
  • AMDGPU Increasing Fragment Size For Performance
    Christian König of AMD yesterday sent out an AMDGPU kernel patch for boosting the default fragment size for GCN graphics cards pre-Vega. The patch is quite trivial and is for boosting the default fragment size from 64KB to 2MB, similar to the move made with the latest Vega GPUs. This change is for GFX6/GFX7/GFX8 graphics processors or basically all the GCN cards prior to Vega "GFX9".