Actions
Forum Thread Details
  • Replies 69 replies
  • Subscribers 61 subscribers
  • Views 7043 views
  • Users 0 members are here
Related

Parallella $99 board now open hardware on Github

morgaine

It's probably spreading everywhere like wildfire, but I just read on Olimex's blog that Adapteva's Parallella kickstarter board now has almost all of its development materials on Github in Parallela and Adapteva repos, and is officially being launched as open hardware.

 

The 16-core board is priced at US$99 and its host ARM is a dual-core Cortex-A9 (Xilinx Zynq 7010 or 7020).  It comes with 1GB DDR3, host and client USB, native gigabit Ethernet and HDMI, so at that price this would be a fairly interesting board even without its 16-core Epiphany coprocessor.  (There's a 64-core version planned too.)  For more details see the Parallella Reference Manual.

 

This has all the makings of a pretty fun board.  I hope Element 14 has one eye open in that direction. image

 

Morgaine.

 

 

PS. Note the 4 x Parallella Expansion Connectors (PEC) on the bottom of the board, illustrated on page 19 of the manual and documented on page 26.  They look very flexible for projects, providing access to both Zynq and Epiphany resources.

Parents

  • If the board maker is paying an extra $55 for the chip then an end user price add on of $80 is perfectly reasonable (even rather low)  - just look at how much the big players charge for an extra few G bytes of flash in a phone or pad.

     

    The price of the Zynqs may fall when they have any competition - the MicroSemi (aka Actel) SamrtFusion2 parts look interesting - CortexM3 and FPGA on a chip with lower prices than Zynq (for less performance of course, but  a much simpler part to work with). If the catalogue price for the XC7010 is $53 for 100 it may well be down to $15 for 100k already. FPGA pricing is always difficult to estimate because all the manufacturers like to keep the real prices for volume between them and the actual customer.

     

    MK

Reply

  • If the board maker is paying an extra $55 for the chip then an end user price add on of $80 is perfectly reasonable (even rather low)  - just look at how much the big players charge for an extra few G bytes of flash in a phone or pad.

     

    The price of the Zynqs may fall when they have any competition - the MicroSemi (aka Actel) SamrtFusion2 parts look interesting - CortexM3 and FPGA on a chip with lower prices than Zynq (for less performance of course, but  a much simpler part to work with). If the catalogue price for the XC7010 is $53 for 100 it may well be down to $15 for 100k already. FPGA pricing is always difficult to estimate because all the manufacturers like to keep the real prices for volume between them and the actual customer.

     

    MK

Children
  • in reply to michaelkellett

    Michael Kellett wrote:

     

    If the board maker is paying an extra $55 for the chip then an end user price add on of $80 is perfectly reasonable (even rather low)  - just look at how much the big players charge for an extra few G bytes of flash in a phone or pad.

     

    That's true for a normal consumer end product designed to make profit.  However, Parallella is not that.  It is a board designed with a very different purpose in mind, to raise the worldwide profile of Adapteva and their Epiphany concept of simple processor arrays, and to work out in the community all the software problems of working with it.  The Adapteva CEO has said very clearly that a 5-man company cannot possibly achieve this on its own.

     

    The board is very unlikely to be generating significant profit for Adapteva, or any at all --- a  rough estimate of the cost of the primary BOM components shows that quite clearly now that we have a ballpark figure for the Xilinx devices.  That extremely low or zero profit margin is a direct result of using inappropriately expensive host-side components around the Epiphany.  The Xilinx tail is wagging the Epiphany dog on costs.

     

    The engineers at Adapteva should have recognized that their primary purpose is to promote Epiphany, not the Zynq range.  The main cost component on the Parallella board should therefore have been their own device, and not Xilinx's.  It's extremely likely that the choice of Zynq will be limiting the uptake of Epiphany through excessive costs.  We live in the age of the Pi price niche, so we know how this works now.

     

    In addition, it should be noted that Adapteva's main software goal is (or should be) to gain experience with and generate tools for Epiphany-side computation, and not to solve nor optimize performance for one very specific ARM + FPGA architecture.  I suspect that not only have they made a wrong choice based on costs, but are also going to end up spending time on the wrong problem as a consequence.