On Fri, Dec 14, 2012 at 11:42 PM, Bruce Perens <[email protected]> wrote:
> I suspect it would be easier to write a fixed-point implementation for a
> generalized
microprocessor than a floating-point one for a DSP chip other than Blackfin.
>
Do you mean to suggest that writing a specialized port for the Blackfin
would be easier than for some other DSPs? Why would that be?
That question is especially relevant to me, as I just started last Friday
trying to get Codec2 to run on Blackfin with the goal of evaluating how
much work it would be to get it to run real-time (hopefully duplex, two or
more channels would be even better). The Blackfin doesn't have hardware
floating point support and I expect that the stock floating point emulation
will be too slow, so I will be looking for ways to speed it up. I have
been collecting ideas from this list and elsewhere, hoping to find a
combination of them that will be fast enough for now. Of course, a
fixed-point port will be a huge improvement long-term, but I have neither
the skill nor the funding to tackle that big of a job right now.
Steve
P.S. Getting codec2 to run on my Linux dev box was as simple as could be.
I was also able to cross-compile it for Blackfin, but the process was a
bit messy, as it first complained of an undefined reference to _rpl_malloc,
then had trouble with generate_codebook because it was a Blackfin rather
than x86 executable. I worked around those issues like this:
export ac_cv_func_malloc_0_nonnull=yes
./configure --host=bfin-uclinux --build=i686-linux
LDFLAGS='-Wl,-elf2flt' CFLAGS=-O2
make # will fail on generate_codebook
cp ../codec2-dev-x86/src/generate_codebook ./src
make # uses incremental build to continue where left off
Ideally ./configure would deal with both of those issues, but I get lost
pretty fast when dealing with the auto-tools, so will keep using the
workarounds for now.
c2enc now runs on Blackfin with no command-line args (printing the usage
info) and crashes with "Data access misaligned address violation" when fed
a file like this "./c2enc 1400 hts1a.raw hts1a_1400.c2". I don't yet know
whether I am overrunning the stack or if there is another problem; I will
be working on that today. Any suggestions about how to troubleshoot it
would be great.
--
Steve Strobel
Link Communications, Inc.
1035 Cerise Rd
Billings, MT 59101-7378
(406) 245-5002 ext 102
(406) 245-4889 (fax)
WWW: http://www.link-comm.com
MailTo:[email protected]
------------------------------------------------------------------------------
LogMeIn Rescue: Anywhere, Anytime Remote support for IT. Free Trial
Remotely access PCs and mobile devices and provide instant support
Improve your efficiency, and focus on delivering more value-add services
Discover what IT Professionals Know. Rescue delivers
http://p.sf.net/sfu/logmein_12329d2d
_______________________________________________
Freetel-codec2 mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/freetel-codec2
