Patent granted for "system on a chip" framework?

[y_p_w]

The URL would be too long. It's patent 6,601,126, and
is available at <http://patft.uspto.gov/netahtml/srchnum.htm>

There was an EE Times (and other CMP websites) article about
this story.

<http://www.eet.com/semi/news/OEG20030801S0043>

This sounds fishy to me. I've personally worked on SoC designs
using only uni-directional busses with various asynchronous
peripherals - well before the time this patent was filed. I'd
like to see PalmChip try to enforce this patent. The EET article
also mentions that FPGAs have been using this kind of technology
for a while.

Here are some of the "claims" of the patent:

"1. An on-chip interconnection system, comprising:

a single semiconductor integrated circuit (IC);

a plurality of uni-directional buses disposed in the IC;

a peripheral-bus (p-bus) included in the plurality of uni-directional
buses and that uses a simple non-pipelined protocol and supports both
synchronous and asynchronous slave peripherals;

a p-bus controller connected to the p-bus and constituting an only
bus-master, and including a centralized address decoder for generating
a dedicated peripheral select signal, and providing for a connection
to synchronous and asynchronous slave peripherals, and further
providing for an input/output (I/O) backplane that allows a processor
to configure and control any of its slave peripherals; and

an m-bus included in the plurality of uni-directional buses, and for
providing a direct memory access (DMA) connection from any said slave
peripherals to a main memory and permits peripherals to transfer data
directly without processor intervention.

2. The on-chip interconnection system of claim 1, wherein, there are
included no tri-stated-buses, and no bi-directional buses.

3. The on-chip interconnection system of claim 1, wherein, each signal
has only a single buffer driver.

4. The on-chip interconnection system of claim 1, wherein, any
broadcast signals are re-driven by simple buffers with no extra
control logic."

 

[Austin Lesea]

ypw,

I, too, have my doubts on this one.

Even though you may have used this prior to the patent, that would only
allow you to continue using the technique without paying royalties. If no
one published or "disclosed" the technique, then the patent could be
valid. Hopefully you have a public published document prior to their
"discovery"? That would kill it immediately.

Since we have been using unidirectional interconnect since Virtex (about 5
years ago now), with soft processor cores, and peripherals, I also believe
that we (Xilinx) have a prior use claim. Since we also published that we
implemented our "tri-state" buses with unidirectional interconnect in
Virtex (as tristates were too slow), it makes this patent pretty dubious
for any FPGA application.

As well, any combination of cores uses single direction buses in Virtex
and all subsequent families (for speed).

But, if you can not point to a published article describing the technique
in an ASIC/ASSP, then they just might "own it."

Austin

 

[Jerry]

The URL would be too long. It's patent 6,601,126, and
is available at <http://patft.uspto.gov/netahtml/srchnum.htm>

Congratulations for USPTO: good work, boys! I've heard that the next
patent in line is: "round device that minimizes friction while moving
vehicles, a.k.a. wheel", granted for GM, of course...

 

[Jonathan Bromley]

The URL would be too long. It's patent 6,601,126, and
is available at <http://patft.uspto.gov/netahtml/srchnum.htm>

This sounds fishy to me [...]

Here are some of the "claims" of the patent:

[...]

Sounds an awful lot like the ARM-originated AMBA interconnect
spec, which has been in the public domain for years.

Nice reinforcement of my prejudices about patent examiners.
--
Jonathan Bromley, Consultant

DOULOS - Developing Design Know-how
VHDL * Verilog * SystemC * Perl * Tcl/Tk * Verification * Project Services

Doulos Ltd. Church Hatch, 22 Market Place, Ringwood, Hampshire, BH24 1AW, UK
Tel: +44 (0)1425 471223 mail: (e-mail address removed)
Fax: +44 (0)1425 471573 Web: http://www.doulos.com

The contents of this message may contain personal views which
are not the views of Doulos Ltd., unless specifically stated.

 

[y_p_w]

ypw,

I, too, have my doubts on this one.

Even though you may have used this prior to the patent, that would only
allow you to continue using the technique without paying royalties. If no
one published or "disclosed" the technique, then the patent could be
valid. Hopefully you have a public published document prior to their
"discovery"? That would kill it immediately.

It was fairly obvious. I was working at a processor company that
integrated an existing core with several external peripherals to
emulate another processor. We used a lot of unidirectional
communications with address decoders.

Since we have been using unidirectional interconnect since Virtex (about 5
years ago now), with soft processor cores, and peripherals, I also believe
that we (Xilinx) have a prior use claim. Since we also published that we
implemented our "tri-state" buses with unidirectional interconnect in
Virtex (as tristates were too slow), it makes this patent pretty dubious
for any FPGA application.

I'm just curious as to whether a soft core implemented in an FPGA would
be close enough to an SoC to make this patent's claims dubious regardless
its use in any ASICs.

As well, any combination of cores uses single direction buses in Virtex
and all subsequent families (for speed).

But, if you can not point to a published article describing the technique
in an ASIC/ASSP, then they just might "own it."

The AMBA 2.0 spec was published by ARM in May 1999. There are probably
several white papers, academic papers, and published web sites that
outline similar SoC frameworks.

 

[Martin Euredjian]

No surprise to me. Some guy out of MIT got a patent for the mechanics of
the human arm (actually, any articulation-muscle mechanism on any animal on
earth for the last several billion years).

Face it, patents are business tools. They have nothing to do with invention
any more. Very few things any more are inventions, most are
implementations. Most are things that good engineers should be able to
produce given a problem and related constraints.

Oh well.


--
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Martin Euredjian

To send private email:
(e-mail address removed)
where
"0_0_0_0_" = "martineu"

The URL would be too long. It's patent 6,601,126, and
is available at <http://patft.uspto.gov/netahtml/srchnum.htm>

This sounds fishy to me [...]

Here are some of the "claims" of the patent:

[...]

Sounds an awful lot like the ARM-originated AMBA interconnect
spec, which has been in the public domain for years.

Nice reinforcement of my prejudices about patent examiners.
--
Jonathan Bromley, Consultant

DOULOS - Developing Design Know-how
VHDL * Verilog * SystemC * Perl * Tcl/Tk * Verification * Project Services

Doulos Ltd. Church Hatch, 22 Market Place, Ringwood, Hampshire, BH24 1AW, UK
Tel: +44 (0)1425 471223 mail: (e-mail address removed)
Fax: +44 (0)1425 471573 Web: http://www.doulos.com

The contents of this message may contain personal views which
are not the views of Doulos Ltd., unless specifically stated.

 

[y_p_w]

No surprise to me. Some guy out of MIT got a patent for the mechanics of
the human arm (actually, any articulation-muscle mechanism on any animal on
earth for the last several billion years).

That actually makes some sense, if the purpose was to create a
mechanical replica - i.e. prosthetics or a robot arm. This new
patent would seemingly describe a set of techniques that most
engineers would never thought of patenting because they were
(more or less) in the public domain.

Face it, patents are business tools. They have nothing to do with invention
any more. Very few things any more are inventions, most are
implementations. Most are things that good engineers should be able to
produce given a problem and related constraints.

Oh well.

I wouldn't have any problems if someone actually took this idea from
the world of FPGAs - extended it to SoC - and then patent the idea
before anyone else had used it. However - this doesn't seem to be
the case.

I really hope this doesn't turn into the SCO Unix vs Linux fight.

 

[Martin Euredjian]

"Martin Euredjian" <[email protected]> wrote in message

That actually makes some sense, if the purpose was to create a
mechanical replica - i.e. prosthetics or a robot arm. This new
patent would seemingly describe a set of techniques that most
engineers would never thought of patenting because they were
(more or less) in the public domain.

Well, not in this case, at least in my opinion. It describes something they
call "series elastic" actuators. Translation: store energy in a spring (or
spring-like element) as opposed to having the motors directly drive the
joint. In other words, a tendon. Using springs to store energy (or control
force) has been in use for a long, long time, I think.


--
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Martin Euredjian

To send private email:
(e-mail address removed)
where
"0_0_0_0_" = "martineu"

 

[Brad Eckert]

I, too, have my doubts on this one.

Even though you may have used this prior to the patent, that would only
allow you to continue using the technique without paying royalties. If no
one published or "disclosed" the technique, then the patent could be
valid. Hopefully you have a public published document prior to their
"discovery"? That would kill it immediately.

Does a web site count as published? For example, are the cores at
Opencores.org considered publicly published because they reside there?

 

[y_p_w]

Well, not in this case, at least in my opinion. It describes something they
call "series elastic" actuators. Translation: store energy in a spring (or
spring-like element) as opposed to having the motors directly drive the
joint. In other words, a tendon. Using springs to store energy (or control
force) has been in use for a long, long time, I think.

Your average desk lamp comes to mind. Garage door mechanisms are
spring loaded too - the ones for solid garage door would seem to
be similar to this "arm" mechanism.