Re: [openrisc] Re:Getting the size down

[Michael Unnebäck <michael@voxi.se>]

Jolly good. The same thing should be done in gcc so that it would be possible to
compile gcc for openrisc with or without hardware support for multiply and
multiply-and-accumulate. Instead of just inserting mul gcc must insert a handful
of opcodes doing the same thing.

Don't forget that the option with mult implemented bitserial would be
attractive. The hardware cost for that type of mult implementation would be very
low. No extra external opcode fetching would be needed. I have a design that
could be used (after conversion from VHDL to verilog). 

regards
/Michael
 
Quoting Damjan Lampret <lampret@opencores.org>:

> Damon,
> 
> if you update your sources, you will now find two defines that remove MAC
> and multiplier. They are called OR1200_MAC_IMPLEMENTED and
> OR1200_MULT_IMPLEMENTED. These two defines do probably what you did, except
> they are now standard. By default they are both enabled. Note if you want
> to
> have MAC, you also need to enable multiplier.
> 
> regards,
> Damjan
> 
> ----- Original Message -----
> From: "Damon Brantley" <brantley@mcloudteleco.com>
> To: <openrisc@opencores.org>
> Sent: Wednesday, April 23, 2003 7:41 PM
> Subject: Re: [openrisc] Re:Getting the size down
> 
> 
> > I took out the multiplier and it cut the utilization on my gate array
> from
> > 96% to 72%.
> > This includes the vga interface I added. With just synthesizing
> or1200_top,
> > utilization went from 89% to 70%.
> > These are just the numbers that xst spits out when the synthesis is done.
> > Still it is enough for me
> > to push on.
> >
> > It did take more than a simple removal of the multiplier. My test c
> > program, which just bounces a character
> > around on the display, acted very flaky with just removing the multiplier
> > instantiation.
> > I had to chase down the connections to the related modules and def things
> > out as needed. Possibly there
> > were some dangling outputs that were generating noise.
> >
> > Regards,
> > Damon
> >
> >
> >
> >
> > At 04:33 PM 4/23/2003 -0700, you wrote:
> > >Damon,
> > >
> > >You can comment (disable) the following assuming you don't need them:
> > >OR1200_CFGR_IMPLEMENTED (this one most likely you don't need)
> > >OR1200_TT_IMPLEMENTED (do you need tick timer?)
> > >OR1200_CASE_DEFAULT (some case defaults will be removed, might confuse
> some
> > >synthesis tools)
> > >
> > >If you update your sources from the cvs, then you will also be able to
> > >disable implementations of not needed exceptions. For deeply embedded
> > >applications running firmware certain exceptions like alignment, illegal
> > >instruction and bus error are not needed at all.
> > >
> > >Also like Michael said you can remove multiplier if you don't needed.
> Right
> > >now there is no define to disable multiplier. I can add one if there is
> > >demand, and even beter is idea of simple small shifting multplier.
> > >
> > >regards,
> > >Damjan
> > >
> > >----- Original Message -----
> > >From: "Damon Brantley" <brantley@mcloudteleco.com>
> > >To: <openrisc@opencores.org>
> > >Sent: Wednesday, April 23, 2003 4:42 AM
> > >Subject: [openrisc] Re:Getting the size down
> > >
> > >
> > >>
> > >> Here it is.
> > >>
> > >> Thanks,
> > >> Damon
> > >>
> > >> >Can you send your or1200_defines.v so I can see if you really
> "removed"
> > >> >everything possible.
> > >> >
> > >> >regards,
> > >> >Damjan
> > >>
> > >> >----- Original Message -----
> > >> >From: "Damon Brantley" <brantley@m... >
> > >> >To: <openrisc@o... >
> > >> >Sent: Tuesday, April 22, 2003 7:52 PM
> > >> >Subject: [openrisc] Getting the size down
> > >>
> > >>
> > >> >> I am currently working with a board made by Digilent that has a
> Spartan
> > >> IIe
> > >> >> with 300k
> > >> >> gates. By removing all the cache and debug logic, the openrisc cpu
> uses
> > >> >> about 84% of the available space.
> > >> >> My last test project,which interfaced with the vga display, pushed
> the
> > >> >> utilization up to 96%.
> > >> >> Are there any other places where I can do some pruning and get the
> size
> > >> >> down more. To get ucLinux up and
> > >> >> running I need at least enough space to add the memory controller
> to
> > >> access
> > >> >> some external memory
> > >> >> and the UART.
> > >> >>
> > >> >>
> > >
> > >
> >
> >---------------------------------------------------------------------------
> -
> > >----
> > >
> > >
> > >>
> > >>
> > >
> > >--
> > >To unsubscribe from openrisc mailing list please visit
> > http://www.opencores.org/mailinglists.shtml
> > >
> >
> > --
> > To unsubscribe from openrisc mailing list please visit
> http://www.opencores.org/mailinglists.shtml
> >
> 
> --
> To unsubscribe from openrisc mailing list please visit
> http://www.opencores.org/mailinglists.shtml
> 



--
To unsubscribe from openrisc mailing list please visit http://www.opencores.org/mailinglists.shtml