We’re expanding our PowerVR Series6 IP family with two new members: PowerVR G6230 and G6430 – two graphics processor cores with two and four clusters, respectively.

Providing our partners and our ecosystem with our latest and greatest IP accompanied by the right tools they require to succeed in the business is vital to us as a company. These new additions to our Series6 line-up will offer an opportunity for our licensees to have a wider choice.

A brief history

But let’s rewind a bit. A while ago we introduced the first PowerVR Series6 GPUs cores, the PowerVR G6200 and G6400, which are designed to offer the best performance while also keeping the smallest area possible and staying true to our ultra-low power approach. Here are Imagination Technologies’ Tony King-Smith, VP of Marketing and Kristof Beets, Business Development Manager, detailing some of the benefits of the new PowerVR Series6 architecture:

 

Imagination Technologies details PowerVR Series6 at CES 2012
 

With PowerVR G6230 and G6430, we’ve “gone all out”, aiming to set a new benchmark for high performance. They have been optimised for maximum efficiency but still manage to keep power consumption to a minimum even with an incremental increase in area. Consequently, they complement the G6200 and G6400 cores by delivering an unmatched power and performance per area rating without sacrificing anything from the Series6 standard feature set.

PowerVR GPUs – the new performance standard for the mobile space

So what does “going all out” mean after all? With a 20x increase in performance compared to current GPU cores targeting comparable markets, the Series6 architecture is able to offer amazing computing horsepower (think in the range of 100s of GFLOPS to even the magnitude of TFLOPS), making it the perfect choice for handling complex user-interfaces and ultra-realistic gaming as well as applications like image processing or augmented and virtual reality. And with that kind of computing prowess, we’re talking about reaching levels of performance that approach the HPC market. But you know how the best things come in small packages. This means we can still give you the best performance per mm2 and per mW in the industry.

PowerVR G6230 block diagram

PowerVR G6230 block diagram

Also, because of our innovative multi-cluster architectural approach, we’ve been able to build on the success of PowerVR Series 5/5XT and offer a completely scalable roadmap which makes no compromises in terms of supporting all modern APIs (OpenCL, RenderScript Compute and Direct Compute), reducing memory bandwidth and providing a versatile line of cores that meets and exceeds the demands of our partners. Furthermore, with the widely used PVRTC texture compression and a new generation of our Tile Based Deferred Rendering architecture, you can rest assured we are continuing our mission to set the standard in the GPU market.

PowerVR G6430 block diagram

PowerVR G6430 block diagram

The hundreds of millions of GPU cores shipped coupled with a market share of 78% in mobile devices GPU IP are strong signs we have been doing things right. So congratulations to our team of engineers who have been hard at work to deliver these amazing graphics cores which will bring a new wave of devices in your hands, homes and cars, making life a bit more colourful (always at the highest resolution, of course!)

For more announcements about our PowerVR Series6 please keep coming back to our blog, follow our Twitter account closely and look out for webinars and live events that showcase our amazing technology.

 

Comments

  • SeanLumly

    How is lossless compression of the frame-buffer achieved? If you are compressing the data and pushing that to memory, wouldn’t special display logic be required to show that compressed buffer on the display?

    • Hi Sean,
      We have not yet revealed details about the framebuffer algorithm implementation so unfortunately I cannot comment on that part of your question. At the system level you would indeed require compression to be supported for the complete display chain for optimal return on investment, however it should be noted that a large number of the render targets in applications today are not “on-screen” but are rendered to “off-screen” buffers and later blended to the final display buffer. This means that lossless compression can reduce the bandwidth usage for all those intermediate reads and writes even if the final write by the GPU and the final read by the display controller is not compressed. Some common usage cases that would benefit: UI compositing within Android, Shadowmaps and reflection maps in games.
      I hope this helps.

    • alexvoica

      Hi Sean,
      We have not yet revealed details about the framebuffer
      algorithm implementation so unfortunately I cannot comment on that part
      of your question. At the system level you would indeed require
      compression to be supported for the complete display chain for optimal
      return on investment, however it should be noted that a large number of
      the render targets in applications today are not “on-screen” but are
      rendered to “off-screen” buffers and later blended to the final display
      buffer. This means that lossless compression can reduce the bandwidth
      usage for all those intermediate reads and writes even if the final
      write by the GPU and the final read by the display controller is not
      compressed. Some common usage cases that would benefit: UI compositing
      within Android, Shadowmaps and reflection maps in games.
      I hope this helps.

      • top.commentor

        really?? performance in magnitudes of tflops?? is that even possible in this decade??

        • alexvoica

          Hi,

          Because the PowerVR ‘Rogue’ architecture is now cluster based, there are certain application areas where our silicon partners could achieve the TFLOPS range if they choose to go for the maximum number of clusters and optimize their SoCs for performance.

          Best regards,
          Alex.

  • SeanLumly

    Ah, of course! How silly of me to forget intermediate buffers, which are incredibly common.

    This must save *tons* of bandwidth, and should help secure PowerVRs performance lead for yet another arch generation. Bravo.

    Thank you so much, David.

  • B

    powervr should come back to the desktop and bring somthing new to the Nvidia/AMD (ATI) cards. Owned both 3dfx and power vr back in the day. loved them. Loved the add in 3d only cards.

  • Davide

    Too bad that there aren’t third party producers of cards equipped with the PowerVR.
    I’m a follower of powerVR chip since the M3D time.

  • zlatan4177

    Sounds great.
    Can you comment on what’s in the USC? And how about HSA, what features are the new architecture support?

    • Hi Zlatan,

      Unfortunately I cannot answer your first question just yet, but we do have an upcoming blog post about the HSA Foundation. It will explain what it is and how Imagination will contribute to it, so watch this space.

      Best regards,
      Alex.

    • alexvoica

      Hi Zlatan,

      Unfortunately I cannot answer your first question just yet, but we do
      have an upcoming blog post about the HSA Foundation. It will explain
      what it is and how Imagination will contribute to it, so watch this
      space.

      Best regards,
      Alex.