Milkymist project news, May 20th
sebastien.bourdeauducq at lekernel.net
Thu May 20 14:59:56 EDT 2010
Read this mail online (with links): http://lekernel.net/blog/?p=1043
The next months will be busy with events around Milkymist. By chronological
order, here is a brief overview of what is going to happen.
* It may not be obvious, but for the last two years I’ve been a Master’s
student in SoC design at the Royal Institute of Technology in Stockholm. My
thesis is about the architecture of the Milkymist SoC, and the seminar will
take place on June, 8th, 15:00 in Stockholm, Sweden (full address available
soon). The defense seminar is public. The draft report is also available, and,
I believe, contains information valuable to many developers.
* Back from Stockholm, I will hold a workshop in Paris, France, about
writing Milkymist/MilkDrop patches. It will take place at 12:00 on June 20th
at the CNAM (292 Rue Saint-Martin 75003 Paris) and is part of a series of
hacking workshops happening there on the same day. More information to come.
* Wolfgang Spraul will present and hopefully demonstrate the project at
Campus Party in Colombia between June 28th and July 6th.
* Finally, I will be present at the Rencontres Mondiales du Logiciel Libre
(Bordeaux, France) on July 6th to talk about the project.
Adam Wang has completed the lay-out of the printed circuit board and has
generated the Gerber plots. The design is available online. The PCB design is
now out to the manufacturer, and boards should come back in a dozen days (see
the schedule page).
Because many people seem to focus exclusively on Milkymist One, I think it is
worthwhile to make a quick reminder so they will not be disappointed. What is
coming out of the factory in a couple of days really are blank and untested
FPGA boards, and nothing else. Porting the existing SoC design to them is
likely to take weeks. It’s not like flashing a new board equipped with a
proven microcontroller with a well-established OS, it’s about porting an
existing SoC design to the bleeding-edge Spartan-6 FPGA technology. We are
likely to encounter synthesizer bugs, timing problems, and high-speed I/O
issues specifically around the DDR SDRAM. There is also substantial background
development to be made (develop a JTAG flasher, modify the NOR flash
controller so it supports the M1 chip, design, implement and test a double-
data-rate I/O system for the SDRAM compatible with the Spartan-6, etc.). All
of this will definitely take time. On top of that, software needs to be
developed or improved: the RTEMS port, the Genode FX GUI toolkit and the
Flickernoise VJ application. The Milkymist One PCB is only the visible part of
the iceberg, really.
A 0.5.1 release is around the corner, which consist mostly of bug fixes.
Nonetheless, those fixes now enable the correct rendering of 62 (and counting)
MilkDrop presets on the Milkymist platform.
Another change has been a rewrite of the SDRAM initialization system, which no
longer requires on-chip RAM, and makes it easier to write alternative
bootloaders like the port of MicroMonitor Michael Walle has been working on.
I have changed the website to highlight the three main areas of the project
(SoC design, conventional hardware design and embedded VJ software
development). Hopefully, it will be more clear to not-so-technical people and
will explain to others that Milkymist is not (just) "an FPGA project".
More information about the discussion