[ECOS] Problems using EP7212's SRAM on Cirrus Logic EDB7209

[ECOS] Problems using EP7212's SRAM on Cirrus Logic EDB7209

[Hoss, Laurent]

We badly need to use the 38K onchip-SRAM for our MP3-Decoder to work in
real time. Currently the decoder allocates 22K dynamically on EDB7209's
DRAM which seems to be much too slow.
( we have tried to port the new, very nice, open source MAD-Decoder
http://www.mars.org/home/rob/proj/mpeg/ )

So I simply tried to replace malloc's return value by 0x60000000
(SRAM-Addr).
Did not work. Therefore I made a test-loop which writes over the whole
SRAM.
Result: Only the first 4K returned the value which was written in. Above
that 4K only 0x00000010 was read.

First I tried to apply the patch to hal_platform_setup.h from the
posting
" http://sources.redhat.com/ml/ecos-discuss/2000-08/msg00156.html "
But that did not work (unable to apply hunk 2...).
Therefore I downloaded v1.8 (with the correct SRAM-size from
25.Aug.2000) of hal_platform_setup.h from CVS, and recompiled ecos (only
v.1.3.1, as the new Config.tool did not work with Windows 95).

Unfortunately the 4K problem remains :(

Another question: Would it be possible to have the whole prg + ecos only
in SRAM ?


Thank you 

Laurent Hoss

Re: [ECOS] Problems using EP7212's SRAM on Cirrus Logic EDB7209

[Lewin A.R.W. Edwards]

Hello Laurent,

>( we have tried to port the new, very nice, open source MAD-Decoder
> http://www.mars.org/home/rob/proj/mpeg/ )

On a tangential note: Is there any particular reason why you are using that 
decoder rather than ARM's? ARM has presumably already hand-optimized in 
assembler any truly critical sections of the code, and observably it runs 
very well (though I haven't tried to integrate it in eCos; I've only tried 
sample applications).

=== Lewin A.R.W. Edwards (Embedded Engineer)
Work: http://www.digi-frame.com/
Personal: http://www.zws.com/ and http://www.larwe.com/

"Und setzet ihr nicht das Leben ein,
Nie wird euch das Leben gewonnen sein."