I read in some A80 datasheet (or in some place in so many post around) about an additional Cortex-M3 inside the A80. I don't have any idea about it.
On Wednesday, October 1, 2014 3:45:06 AM UTC-4, Maxime Ripard wrote: > > On Mon, Sep 29, 2014 at 01:46:50PM -0700, javqui wrote: > > Hi, > > I'm working on a couple of projects requiring the classic Micro > controller > > features (low power, deterministic real time processing) and the classic > > UX, flexibility and functionality of Linux /android. > > > > Most SoCs today provide many high level external hardware interfaces > (like > > Camera, USB, HDMI, etc) but some projects require additional drivers and > > interfaces to handle different external hardware. Usually we solve the > > interconnectivity with extra MCUs, FPGAs or other specialized chip > > interfaces available. > > > > Sometimes, we design product boards with two solutions: a Cortex A SoC > like > > Allwinner/rockchip/Omap series and a small MCU Cortex M like the STM32 > > series, but with a powerful A80, it could change forever. > > > > I will receive my first Optimus board soon, and I want to customize the > > kernel to create a classic Linux running on the powerful 4x A15+ GPU and > > Nucleus (or Free RTOS) on one or two of the A7 of the Allwinner A80 Soc. > (I > > made similar kernel works with MTK SoCs in the past, but never try to > run > > two operating systems in the same chip at the same time) > > > > Both projects require continuous operation and deterministic real time > > response on the low power processor(s) (RTOS on A7). > > User interaction (Linux on the A15 + GPU side ) is only eventual, so > termal > > issues by running almost all processors at the same time occasionally, > > should not be a problem. > > > > If anyone anticipate a significant barrier to build a kernel of this > type, > > please share it here, I will really appreciate. I will share the results > > and evaluation test here > > What might be easier for you, and probably less intrusive from the > kernel point of view, would be to use the co-processor that some > Allwinner SoCs have. I know the A31 has one, and I'm pretty sure the > A80 too. > > That would leave Linux in charge of the "real" CPUs, while offloading > your RT tasks to a smaller processor, without having to deal with all > the segmentation in the bootloader. > > And if you're used to using Cortex-M, you shouldn't need all that > horsepower anyway. > > Maxime > > -- > Maxime Ripard, Free Electrons > Embedded Linux, Kernel and Android engineering > http://free-electrons.com > -- You received this message because you are subscribed to the Google Groups "linux-sunxi" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
