Hello,
I'm new here, I worked before only in the proprietary world. I started to think about a project in my new company (called Gnuside) that could increase the open source hardware world (not meaning it will be a huge advance, but it will (I hope) contribute) :
We are about to design a hardware containing those general characteristics : - a ~1GHz dual-core CPU (here is he main node of my question) - 1 or 2 GB of DDR2/3 - 16 GB of flash for the system - SATA controller (which will be populated by 2 hard-drives in the public delivered release) - 2 or 3 ethernet 1 Gbps ports - COM port - a programmation/debug port.
Why I'm talking about that ?
We plan to put a Freescale processor of the family of P102x, from what I heard this processor has it's documentation available for everyone, and seems a good choice for an open source hardware platform. From what I heard ARM processors which are in the center of many open source hardware platforms are not that easy to use for open source community. Are PPC better ? I was at first thinking about the future T2080 which put back the Altivec solutions and a new architecture and packet engines that seems promising, but everything about them are under NDA. Even if this could be circumvented by writing some open source test programs around the basic functionalities of the CPU, would it be a big problem for the community ?
What do you think about all this ? I'm asking that to better know how to manage an open source hardware project, without messing things up.
Regards
Roland Laurès
PS: The project using this platform will be Firstbrick and has a hardware and software part, and I'm managing the hardware part.
On Tue, 2012-10-02 at 01:20 +0200, Roland Laurès wrote:
Hello,
I'm new here, I worked before only in the proprietary world. I started to think about a project in my new company (called Gnuside) that could increase the open source hardware world (not meaning it will be a huge advance, but it will (I hope) contribute) :
We are about to design a hardware containing those general characteristics :
- a ~1GHz dual-core CPU (here is he main node of my question)
- 1 or 2 GB of DDR2/3
- 16 GB of flash for the system
- SATA controller (which will be populated by 2 hard-drives in the public delivered release)
- 2 or 3 ethernet 1 Gbps ports
- COM port
- a programmation/debug port.
Why I'm talking about that ?
We plan to put a Freescale processor of the family of P102x, from what I heard this processor has it's documentation available for everyone, and seems a good choice for an open source hardware platform. From what I heard ARM processors which are in the center of many open source hardware platforms are not that easy to use for open source community. Are PPC better ? I was at first thinking about the future T2080 which put back the Altivec solutions and a new architecture and packet engines that seems promising, but everything about them are under NDA. Even if this could be circumvented by writing some open source test programs around the basic functionalities of the CPU, would it be a big problem for the community ?
Hi Roland,
I think the furore over the specs for the SoC on Raspberry Pi has damaged the repuation of ARM for open source. However the secrecy is really about the GPU, not the ARM processor itself. With so many open systems on ARM, it is as good as any proprietary processor. Whether you choose ARM or another manufacturer (and many of them will be using ARM anyway), if it has a proprietary processor, there will be a degree of secrecy.
I think Pandaboard or Hackberry are close to your actual specs.
As an alternative you could consider one of the open processor's around and run on an FPGA. OpenRISC, Lattice Mico, LEON and OpenSPARC are out there. But you'll struggle to hit 1GHz, and I don't know if you'll get dual core in any of them.
What do you think about all this ? I'm asking that to better know how to manage an open source hardware project, without messing things up.
Well messing up is how you learn:) Most of the best open source came from something that got messed up!
HTH and best wishes,
Jeremy