Re: [f-cpu] C/VHDL code and fcpusim

[Yann Guidon <whygee@f-cpu.org>]

hi,

jaap stolk wrote:
> 
> hi,
> 
> YGASM/JWSASM:
> there is not a lot of difference, i only made one
> change:
> 
> i used ygasm from stable_yg_12_31_2001.tgz
> and added the line:
>        current_instruction = 0;
> after:
>       /* reset for the next round : */
> in:        /ygasm/ygasm_bin.c

well, i could reuse this version but i had worked on
it after december, until i found that i wanted a sophisticated
memory allocation library (i don't trust glibc...)

My goal is to allow forward label declaration, which would
make FROMFS possible, as well as many other things.

The problem is that i don't want to promote memory leaks
(for example, when a very large application is compiled,
the assembler could kill the computer if the allocation is
not correctly done).

Maybe i should concentrate on this now ?
This would give you enough visibility on your code
so you can advance it without waiting for my updates.

> > > about an extra 0-cycle unit for the register moves:
> > > we also could add an extra ouput to an existing
> > > unit, and use that as a 0-cycle bypass.
> > ?
> i added a EU_MOV read / write port to the xbar, that
> dous the read-bus -> write-bus bypass.
> (see: /include/***_ports.h)
ok

> but useing the read port of an other unit could work
> also.
here, i don't understand.

> all bypasses before and after a register move work
> ok now. :-)
did you solve the problem of the 2nd bypass ?

that is : after the first obvious bypass, there is
another bypass condition due to the write latency of
the register set. The register set must be bypassed
so that the old value is not read.
Usually, there is a "latch" before the register set
write ports : these 2 values (2 write ports) can be
reused/reinjected on the Xbar.

> > > i'll upload my snapshot as soon as i added your
> > > changes.
> > i'll make something quick so you can work.
> we are getting into a bit of a snapsot-war.... :-)
it's not a war, and quite the contrary :-)
i'm happy that things move and give tangible results.
Like many people, i'm fed up of the useless endless
"discussions" on this list and your code is a bliss.

> i'll try and display_color[] display_normal[] to the
> other units.
isn't it handy ? :-)

> > btw, did you install simili ?
> yes, and read some of the documentation, but i haven't
> run it yet. i spend all morning getting my modem to
> work, and then found that my kernel has no PPP
> support :-(
been there, done that, and still doesn't work because
cardmgr doesn't recognize my PCMCIA modem ...

> what's your opinion about two latency outputs from
> the decoder (one for each write port) ?
i still don't understand what it means.
is it related to the IMUL thing ? If yes, then
i think it's necessary.

> jaap.
WHYGEE
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
*************************************************************
To unsubscribe, send an e-mail to majordomo@seul.org with
unsubscribe f-cpu       in the body. http://f-cpu.seul.org/