All the talk about ARM server SoCs has been focused on who will come up with the breakthrough chip design. Watching trends like OPNFV develop suggests the big breakthrough is more likely to come on the ARMv8 software side. How do you quickly validate ARMv8 software when you don’t have the exact ARMv8 SoC target?
ARM has their answer: the Juno ARM Development Platform. It comes in a couple flavors with different Juno SoCs, two with a Cortex-A57 and Cortex-A53 big.LITTLE cluster, and a newer version with a Cortex-A72 and Cortex-A53 cluster. Also on chip is a Mali-T624 GPU and the CCI-400 cache coherent interconnect, along with a DDR3 memory subsystem, an optional PCIe controller, and a Cortex-M3 core for system control functions.
The Juno ARM Development Platform also comes with a Linaro ARMv8-A reference port plus firmware for the System Control Processor. It’s most of what software developers need to get started – if they are working in Linux, and if they aren’t too concerned with peripheral IP.
From the block diagram above, you can see the “LogicTile”, ARM’s name for expansion daughtercards with an FPGA for custom logic. The AXI interface provides dedicated routing, delivering considerable bandwidth for user-defined IP blocks. This is starting to look like a job for FPGA-based prototyping, except by using the Juno hardware one need not worry about partitioning the processor and GPU cores into FPGAs.
Synopsys agrees. They have just released a HAPS adaptor for the Juno ARM Development Platform. This brings to bear all the synthesis and debug capability of HAPS ProtoCompiler along with a suite of DesignWare IP. In an upcoming webinar, Achim Nohl and Hugo Neto of Synopsys will take a more detailed look at the Juno ARM Development Platform, how it integrates with a HAPS FPGA-based prototyping system, and the capabilities and use cases the hybrid prototype provides for ARMv8 software development.
To register for this event to be presented live on July 21:
One obvious use case is a non-Linaro software stack, for example the Enea stack for OPNFV. I’ll be curious to see if and how the Synopsys team addresses that. In any case, the case for hybrid prototyping combining a Juno SoC with HAPS to quickly and cost-effectively get ARMv8 software running should be of broad interest for teams getting into this space.Share this post via: