PowerVR Series 7XT GPU announcement makes us figure out how this thing works

I'll admit that I've not spent enough time looking into how some of the top mobile graphics cores actually work. That's true in part because we primarily cover desktop parts, and in part due to limited time. The biggest contributor, though, may be the simple fact that the mobile GPU guys don't tend to share many details of their GPU architectures.

That said, the folks at Imagination Technologies have shipped a ton of GPU inside of, you know, every iPhone and iPad ever, and they have been somewhat forthcoming recently about how their GPUs look internally. That openness extends to today's announcement of the PowerVR Series7XT GPUs, which are the new high end of the PowerVR lineup and are based on a newly revised incarnation of the PowerVR Rogue architecture. The folks at Imagination have even provided a reasonably useful functional block diagram.

Yep, looks vaguely like a GPU. Some of the most notable features of the Series7XT family aren't readily apparent from a block diagram, though.

Many of those features involve support for Android 5.0 Lollipop and the Android Extension Pack for OpenGL ES 3.1. Most notable among them is new tessellation hardware that should allow for higher-detail worlds in mobile games and graphics. That addition should bring Rogue's feature set up to par with many desktop GPUs. In fact, Imagination says the Series7XT has "optional" support for DirectX 11.

The Series7XT also adds hardware support for GPU virtualization, so a single Series7XT GPU can be shared among multiple virtual machines running on a hypervisor.

In my view, though, the biggest change in the Series7XT is simply one of scale. To understand that, we need to take a closer look at the GPU's architecture, which brings us to the block diagram above. This diagram has amazing potential to confuse and confound, but I think the bottom line is fairly straightforward. Start with the fact that mobile GPUs like this one still tend to employ fp16 precision by default, rather than the fp32 default used by modern desktop graphics. That's a reasonable choice for real-time mobile graphics, I think.

What this diagram is attempting to show us is, in its most detailed form on the left, a single pipeline from a Rogue cluster. I don't think that pipeline really has six "ALU core" units and a special-function unit (SFU). Instead, I think it generally executes four fp16 operations per cycle, possibly in conjunction with a SFU op. If a program asks for 32-bit precision, then you get two fp32 operations instead. I doubt the hardware can co-issue and process a full slate of fp16 and fp32 instructions simultaneously.

Also, I don't think those "FLOP" boxes represent any sort of hardware. I think they're meant to convey the ability of each unit, at peak, to produce floating-point operations. Thanks to the magic of the fused multiply-add, each "ALU core" can process two flops per cycle, max.

Bottom line, then, each superscalar Series7XT pipeline can process the equivalent of a single pixel (plus maybe a SFU operation) in each clock cycle. Whether that's scheduled as four red components or a single RGBA pixel, I can't tell from here. Regardless, these pipelines are aggregated together in 16-wide groups. Each of these groups is pretty beefy, then, with a total of 64 "ALU cores," as Imagination Technologies calls them. Nvidia has taken to calling these same hardware resources "CUDA cores," and AMD still uses "stream processors" most of the time.

Now consider that the Series7XT can scale up to 16 clusters or 1024 "ALU cores" for fp16—and the equivalent of 512 "ALU cores" for fp32 datatypes. Even the fp32 count, which I believe is the relevant comparison here, is substantially larger than the 192 "CUDA cores" in Nvidia's Tegra K1 SoC. That's simply a statement of scale, not efficiency or throughput, but it helps orient us to the landscape.

Of course, smaller implementations are also possible. The entire Series7XT family is detailed here. Meanwhile, Imagination is also taking this architecture into smaller scale deployments like wearables with the Series7XE family.

This product announcement simply means Series7XT GPUs are available for Imagination Technologies' customers to license, so we probably won't see any solutions based on this IP hitting the market for another six to 12 months. Still, based on what Imagination Technologies is offering its customers, we can probably expect that mobile graphics will continue to scale up at a breakneck pace in the coming years.

Comments closed
    • l33t-g4m3r
    • 5 years ago

    Hmm. Looks like they just copied Kepler by making it feature compatible with open standard api’s, time for a lawsuit. /sarcasm

    • bfar
    • 5 years ago

    The gem here is the optional support for DX11. Nvidia and ATI (and Intel) will take note, because from a hardware perspective there isn’t much of a jump from here into the desktop market. Another competitor in this space would be mana from heaven from a pricing point of view. It may never happen, but no doubt the major desktop players are keeping a close eye on these guys.

      • Deanjo
      • 5 years ago

      Ummm, Nvidia has supported DX in all their tears line since day one. In fact the first Tegras were windows only.

        • l33t-g4m3r
        • 5 years ago

        lol. Tiers, not tears, and tegra didn’t support dx 11 until the k1.

          • Deanjo
          • 5 years ago

          Meh auto correct. Still, nvidia has not made an openGL only Tegra processor ever. They have always supported DX so I have no idea how this is something that nVidia should take note of especially when the K1’s GPU was derived from the desktop version. If anything nvidia has been aiming their development towards solutions that can be utilized on any platform.

          So again, how is any of this even remotely something that nvidia should take note of since they already have done it and it is something that always has been in their roadmaps?

    • blastdoor
    • 5 years ago

    Has anybody implemented that raytracing option we hard about earlier in the year? I think it was called Wizard?

      • ronch
      • 5 years ago

      I’m sure Apple would be very interested in Wizard. /pun

    • the
    • 5 years ago

    How does this thing work? I dunno but if you let the magic smoke escape, it’ll certainly stop working.

    • Ninjitsu
    • 5 years ago

    I think maintaining efficiency when scaled to larger configurations is exactly what mobile architectures still have to prove, before they have a chance at replacing Intel/Nvidia/AMD in the PC area.

      • dodozoid
      • 5 years ago

      But do they intend to replace them in PC area in the first place?

        • End User
        • 5 years ago

        [url=http://www.theverge.com/2014/6/18/5817400/the-best-chromebook<]Yes[/url<].

      • renz496
      • 5 years ago

      isn’t that imagination technologies used to make desktop card before becoming what they did right now?

        • Scrotos
        • 5 years ago

        PowerVR 250, Kyro, Kyro II, Matrox m3D. Best known at the time for powering the Dreamcast and for their “tiling” architecture which, when I asked Carmack about it at QCon 2001 or 2002 (forget which one) he was dubious that it had any value due to increased memory usage.

        [url<]http://en.wikipedia.org/wiki/List_of_PowerVR_products[/url<]

          • egon
          • 5 years ago

          Hyped as a 3Dfx-killer and derogatively referred to as “PowerPR” at the time. PowerVR had its fans but it was a pretty lopsided rivalry – there’d easily have been ten 3Dfx fanboys for every PowerVR contrarian.

          • ozzuneoj
          • 5 years ago

          I actually have a Hercules 3D Prophet 4500 Kyro II card. It was given to me several years ago but I never actually got to try it out. I really wanted one back in the Geforce 2 days but I ended up splurging for a Visiontek Geforce 2 GTS 32Mb instead (to replace my Voodoo 3 2000 PCI). DDR memory and hardware T&L ended up being the more reliable choice… though the Kyro II did benchmark better in many situations despite its shortcomings and far simpler design… which is why I’m sure the technology adapted well to the mobile space.

          Still, its hard to believe that PowerVR went on to design some of the most used mobile graphics processors in the world several years in a row. They probably never would have expected that back in the day. I don’t think anyone would have.

            • Scrotos
            • 5 years ago

            The drivers were the main problem with the tile based cards, from what I recall.

            • ronch
            • 5 years ago

            Same here, although I went from a Voodoo3 3000 to an Elsa Geforce2 GTS. During the days of the Voodoo3, there was the Riva TNT but it seemed like the dark horse contender, so, being my first time to delve into 3D accelerators (my prior S3 ViRGE not included), I didn’t want to take a chance with the TNT and just went with 3dfx. Wasn’t disappointed. Then the following year Nvidia gained more recognition with their Geforce2 lineup. There were other ‘interesting’ options such as those from ATI and S3’s Savage4, but really, why bother with them? PowerVR wasn’t even a consideration.

          • JustAnEngineer
          • 5 years ago

          I had a Matrox m3D. When processors got faster and programs started using MMX, it didn’t take long for the m3D to become a “video decelerator”.

            • Scrotos
            • 5 years ago

            Oh come on, don’t take that away from the trident and s3 stuff at the time. That’s all they got, that decelerater label!

            • ronch
            • 5 years ago

            S3 VIRGE wasn’t bad at 3D… I remember how fast mine was… at rotating a textured cube.

            • adisor19
            • 5 years ago

            The S3 drivers were worse than ATI’s and that says a lot at the time.

            Also, Trident never ever produced anything more than decelerators.

            Adi

        • ET3D
        • 5 years ago

        I admire the way Imagination Technologies has managed to stay in business since that time and stay relevant. They don’t compete with the PC big boys, but they’re selling well, and perhaps they will return to PC space, now that low power is important.

Pin It on Pinterest

Share This