Announcement

I will write a open source arm vhdl model from 
this  month on. I hope to have it ready in 
approximately 2 months using LeonSoc as a framework 
and implementing a arm1 integerunit. Anyone that wants 
to participate is wellcome.
Konrad Eisele

> I will write a open source arm vhdl model from
> this  month on. I hope to have it ready in
> approximately 2 months using LeonSoc as a framework
> and implementing a arm1 integerunit. Anyone that wants
> to participate is wellcome.
> Konrad Eisele

Many tried, all got kicked in the back by ARM. If you can create something
useful, get ready for a contact from one of their lawyers,

Andras Tantos

"Andras Tantos" <andras_tantos@tantos.yahoo.com> wrote in message news:<3fa942a5@news.microsoft.com>...
> > I will write a open source arm vhdl model from
> > this  month on. I hope to have it ready in
> > approximately 2 months using LeonSoc as a framework
> > and implementing a arm1 integerunit. Anyone that wants
> > to participate is wellcome.
> > Konrad Eisele
> 
> Many tried, all got kicked in the back by ARM. If you can create something
> useful, get ready for a contact from one of their lawyers,
> 
> Andras Tantos

why write it from scratch? the nnARM source codes can be found with google
with no problems. the synthesis result is unfortunatly quite big fits
barely into XC2S600E

would be nice to have smaller core though ;)

antti
PS the ARMlayers tried to take nnARM off from public internet, but its
still downloadable, so the layers are not doing very good job.

antti@case2000.com (Antti Lukats) wrote in message news:<80a3aea5.0311060129.38e3476a@posting.google.com>...
> "Andras Tantos" <andras_tantos@tantos.yahoo.com> wrote in message news:<3fa942a5@news.microsoft.com>...
> > > I will write a open source arm vhdl model from
> > > this  month on. I hope to have it ready in
> > > approximately 2 months using LeonSoc as a framework
> > > and implementing a arm1 integerunit. Anyone that wants
> > > to participate is wellcome.
> > > Konrad Eisele
> > 
> > Many tried, all got kicked in the back by ARM. If you can create something
> > useful, get ready for a contact from one of their lawyers,
> > 
> > Andras Tantos
> 
> why write it from scratch? the nnARM source codes can be found with google
> with no problems. the synthesis result is unfortunatly quite big fits
> barely into XC2S600E
> 
> would be nice to have smaller core though ;)
> 
> antti
> PS the ARMlayers tried to take nnARM off from public internet, but its
> still downloadable, so the layers are not doing very good job.

Doesn't really matter, good enough in this case, lets any potential
commercial user get the message loud & clear. If its a 600 target,
thats one very expensive Arm compared to real thing. For an opensource
cpu to be useable, it must be competitive in size, speed, power with
commercial cpus.

johnjaksonATusaDOTcom

"Andras Tantos" <andras_tantos@tantos.yahoo.com> wrote in message news:<3fa942a5@news.microsoft.com>...
> > I will write a open source arm vhdl model from
> > this  month on. I hope to have it ready in
> > approximately 2 months using LeonSoc as a framework
> > and implementing a arm1 integerunit. Anyone that wants
> > to participate is wellcome.
> > Konrad Eisele
> 
> Many tried, all got kicked in the back by ARM. If you can create something
> useful, get ready for a contact from one of their lawyers,

I wrote my own clone earlier this year (just haven't got around to
releasing it yet). I spent quite a bit of time studying the patent
issues, and believe that
you should be able to implement the basic V4 instruction set (This
doens't include Thumb). The only real hurdle is then likey to be
patent 5,386,563 which deals with the exception processing mechanism.
However, I believe this patent to be invalid, as there is a
significant amount of prior art (Search for "ARM's dubious patents" in
comp.arch).

Cheers,
JonB

antti@case2000.com (Antti Lukats) wrote in message news:<80a3aea5.0311060129.38e3476a@posting.google.com>...
> "Andras Tantos" <andras_tantos@tantos.yahoo.com> wrote in message news:<3fa942a5@news.microsoft.com>...
> > > I will write a open source arm vhdl model from
> > > this  month on. I hope to have it ready in
> > > approximately 2 months using LeonSoc as a framework
> > > and implementing a arm1 integerunit. Anyone that wants
> > > to participate is wellcome.
> > > Konrad Eisele
> > 
> > Many tried, all got kicked in the back by ARM. If you can create something
> > useful, get ready for a contact from one of their lawyers,
> > 
> > Andras Tantos
> 
> why write it from scratch? the nnARM source codes can be found with google
> with no problems. the synthesis result is unfortunatly quite big fits
> barely into XC2S600E
> 
> would be nice to have smaller core though ;)
> 
> antti
> PS the ARMlayers tried to take nnARM off from public internet, but its
> still downloadable, so the layers are not doing very good job.

I'm not sure if it has been worked on since the last time I saw it,
but I don't believe the nnARM could really be used as is. Didn't it
just come with a behavioural memory interface? I also seem to remember
the pipeline being pretty shallow, meaning it isn't exactly going to
go very fast.

Cheers,
Jon

> > > > I will write a open source arm vhdl model from
> > > > this  month on. I hope to have it ready in
> > > > approximately 2 months using LeonSoc as a framework
> > > > and implementing a arm1 integerunit. Anyone that wants
> > > > to participate is wellcome.
> > > > Konrad Eisele
> > > 
> > > Many tried, all got kicked in the back by ARM. If you can create something
> > > useful, get ready for a contact from one of their lawyers,
> > > 
> > > Andras Tantos
> > 
> > why write it from scratch? the nnARM source codes can be found with google
> > with no problems. the synthesis result is unfortunatly quite big fits
> > barely into XC2S600E
> > 
> > would be nice to have smaller core though ;)
> > 
> > antti
> > PS the ARMlayers tried to take nnARM off from public internet, but its
> > still downloadable, so the layers are not doing very good job.
> 
> Doesn't really matter, good enough in this case, lets any potential
> commercial user get the message loud & clear. If its a 600 target,
> thats one very expensive Arm compared to real thing. For an opensource
> cpu to be useable, it must be competitive in size, speed, power with
> commercial cpus.

And if you want to be better than the competition, don't base your
design on ARM. It has (IMHO) a huge flaw in the architecture
(shifting), that means if you want decent performance, you end up
having to add tons of extra logic.

Jon

johnjakson@yahoo.com (john jakson) wrote in message news:<adb3971c.0311061709.5ca8cff7@posting.google.com>...
> antti@case2000.com (Antti Lukats) wrote in message news:<80a3aea5.0311060129.38e3476a@posting.google.com>...
> > "Andras Tantos" <andras_tantos@tantos.yahoo.com> wrote in message news:<3fa942a5@news.microsoft.com>...
> > > > I will write a open source arm vhdl model from
> > > > this  month on. I hope to have it ready in
> > > > approximately 2 months using LeonSoc as a framework
> > > > and implementing a arm1 integerunit. Anyone that wants
> > > > to participate is wellcome.
> > > > Konrad Eisele
> > > 
> > > Many tried, all got kicked in the back by ARM. If you can create something
> > > useful, get ready for a contact from one of their lawyers,
> > > 
> > > Andras Tantos
> > 
> > why write it from scratch? the nnARM source codes can be found with google
> > with no problems. the synthesis result is unfortunatly quite big fits
> > barely into XC2S600E
> > 
> > would be nice to have smaller core though ;)
> > 
> > antti
> > PS the ARMlayers tried to take nnARM off from public internet, but its
> > still downloadable, so the layers are not doing very good job.
> 
> Doesn't really matter, good enough in this case, lets any potential
> commercial user get the message loud & clear. If its a 600 target,
> thats one very expensive Arm compared to real thing. For an opensource
> cpu to be useable, it must be competitive in size, speed, power with
> commercial cpus.
> 
> johnjaksonATusaDOTcom

Does anyone know when the arm license is going to expire?

> > Doesn't really matter, good enough in this case, lets any potential
> > commercial user get the message loud & clear. If its a 600 target,
> > thats one very expensive Arm compared to real thing. For an opensource
> > cpu to be useable, it must be competitive in size, speed, power with
> > commercial cpus.
> > 
> > johnjaksonATusaDOTcom
> 
> Does anyone know when the arm license is going to expire?

I think the exception processing patent was filed around 90-92, so its
got quite a bit in it left..

JonB

jon@beniston.com (Jon Beniston) writes:
> I wrote my own clone earlier this year (just haven't got around to
> releasing it yet). I spent quite a bit of time studying the patent
> issues, and believe that
> you should be able to implement the basic V4 instruction set (This
> doens't include Thumb). The only real hurdle is then likey to be
> patent 5,386,563 which deals with the exception processing mechanism.
> However, I believe this patent to be invalid, as there is a
> significant amount of prior art (Search for "ARM's dubious patents" in
> comp.arch).

If you only need binary compatibility with user-mode ARM software, you
can implement a different exception handling mechanism and avoid the
patented technique entirely.