DSPRelated.com
Forums
Online · Live + on demand
February 9-11, 2027
Price goes up Jan 1: register now to lock in today's rate
Register Now →

Suggation needed for 6713

Started by "H.264encoderondm642" ●January 4, 2007
Nilesh-
> I have code which is open source code. So there is no code from my side.

Ok, so b_frame.c references:

mrefD
imgY_refD
imgUV_tmpD
.
.
.

and these symbols are not found by CCS? Then I suggest to open each of your other
.obj files, one by one using a hex editor, and search for "_mrefD". If you don't
find it, then -- no matter what you may think -- you're not supplying those symbols
to the link process.

-Jeff
> Jeff Brower wrote:
>
> Nilesh-
>
> > I have tried all possible ways to tackle this undefined symbol problem.
> I have noticed few things that are strange
> > for me. For e.g. my .bss section in mapping file have length of 0. This
> is for decoder code ==> Somehow, you're making this way too hard. Linking is hardly the thing
> that should cause a many-day problem.
>
> Who makes file "b_frame.c" -- you? Or you get this from a third-party?
>
> -Jeff
>
> > .bss 0 00000000 00000000 UNINITIALIZED
> >
> > while for my succesfull encoder it shows this as ==>
> >
> > .bss 0 00005da0 00000158 UNINITIALIZED
> > 00005da0 00000158 b_frame.obj (.bss)
> >
> > Also similar to following Lines are there for each undefined symbol
> e.g.
> >
> > DW$60 .dwtag DW_TAG_variable, DW_AT_name("imgUV"), DW_AT_symbol_
> name("_imgUV")
> > .dwattr DW$60, DW_AT_type(*DW$T$131)
> > .dwattr DW$60, DW_AT_declaration(0x01)
> > .dwattr DW$60, DW_AT_external(0x01)
> > ****************************************************************
> >
> > .global _imgUV
> > .bss _imgUV,4,4
> > DW$42 .dwtag DW_TAG_variable, DW_AT_name("imgUV"), DW_AT_symbol_
> name("_imgUV")
> > .dwattr DW$42, DW_AT_location[DW_OP_addr _imgUV]
> > .dwattr DW$42, DW_AT_type(*DW$T$145)
> > .dwattr DW$42, DW_AT_external(0x01)
> >
> >
> > Now second code in mapping file has .global directive with it in
> comparison with first. And second is from mapping
> > file of encoder which is working ok. While first code was from mapping
> file of decoder giving linking error. So
> > though symbol is defined it is not globally defined. Which I think is
> definitely wrong.
> >
> > Also for your information I have som definition in my main file asbr />

> > #define GLOBAL_VARS
> > And global.h file that is included everywhere as br />

> > #ifdef GLOBAL_VARS
> > #define EXTERN_GLOBAL
> > #else
> > #define EXTERN_GLOBAL extern
> > #endif
> > And I have used symbol before all global variable in global.h file as
> XTERN_GLOBAL
> >
> > I tested all code on vc++ and it is working fine. But am not able to
> figure out why undefined symbol error is there.
> > Also my .cmd file is OK as below=>
> >
> >
> > -c
> > -heap 0x1F4FFFF
> > -stack 0x2000000
> > -l rts6700.lib
> >
> > MEMORY
> > {
> > L2SRAM: origin = 0x00000000 length = 00008000h
> > CE0: origin = 0x80000000 length = 0xB0000
> > CEO: origin = 0x800B0000 length = 0x3F4FFFF
> > CE1: origin = 0x90000000 length = 0x3FFFFFF
> > CE2: origin = 0xA0000000 length = 0x3FFFFFF
> > CE3: origin = 0xB0000000 length = 0x3FFFFFF
> > }
> >
> > SECTIONS
> > {
> > /* .sysmem > CEO
> > .stack > CEO*/
> > .sysmem: {*(.sysmem) } > CEO | CE1 | CE2 | CE3
> > .stack: {*(.stack) } > CEO | CE1 | CE2 | CE3
> >
> > .cinit > L2SRAM
> > .bss: { /*b_frame.obj (.bss)*/
> > *(.bss) } > L2SRAM
> > .const > L2SRAM
> > .cio > L2SRAM
> > .switch > L2SRAM
> > .data > L2SRAM
> > .tables > L2SRAM
> >
> > .far > CE0
> > .external > CE0
> > .text > CE0
> > }
> >
> > I need kindly your suggestions
> > Regards,
> > Nilesh
> >
> > Jeff Brower wrote:
> > Nilesh- > ## Do u means to say that I can use embedded JTAG emulator
> >> with just usb cable to pc and from CCS? Yes. > What configuration
> setting I have to make for communication
> >> through this then? The version of CCS that comes with DSK boards
> handles this automatically. > ## If I need to
> >> use one CCS project then :
> >> My small portion of code i.e. integer dct is going to be on
> >> DSP 6713 so I need to download that part to 6713. So I was
> >> thinking to make 2 CCS project separately.
> >> But if I can have this from one CCS project also possible then
> >> can u please explain me, how can I differentiate between part
> >> to be downloaded on target and remaining part that will be
> >> running on host through CCS. Do I need to have different
> >> sectional mapping to be used? I can understand why you would ask this,
> but you're thinking of the "tree" instead
> >> of the "forest". Here are the main possibilities for host-side
> software: -CCS only. Edit/compile/build/link
> >> code, debug functions,
> > waveform and data display functions, file I/O -Simulink/MATLAB + CCS.
> Block diagrams with one or more
> > blocks that run on the DSP. Simulink uses CCS for DSP
> > code download and communication -Visual Studio + HPI DLL. C/C++ code
> that talks to DSK
> > board (See previous post for notes) Hopefully this explains why you
> don't need more than one CCS project. The
> > main thing you are missing is that CCS is both a GUI for code
> development (compile code) and also contains debug,
> > I/O, and other target (board) communication functions. I suppose this
> is why TI calls it an IDE -- integrated
> > development environment. > ## Also I am getting following kind of
> errors while linking on CCS 6713:
> >> [Linking...] "C:\CCStudio_v3.1\C6000\cgtools\bin\cl6x" -@"Debug.lkf"
> Linking errors are basic and you should be
> >> able to figure out on your own by studying TI's example projects. As
> just a brief suggestion, I might guess that
> >> you don't have b_frame listed in your .cmd (linker command) file.
> -Jeff
> >
> > undefinedfirst referenced
> > symbolin file
> > -------------------------
> > _mrefD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\
> Debug\b_frame.obj
> > _imgY_refD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\
> Debug\b_frame.obj
> > _imgUV_tmpD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\
> Debug\b_frame.obj
> > _P_intervalD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\
> Debug\b_frame.obj
> > _ftimeD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\
> Debug\b_frame.obj
> > _prevP_trD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\
> Debug\b_frame.obj
> > _p_traceD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\
> Debug\b_frame.obj
> > _imgUVD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\
> Debug\b_frame.obj
> > _imgY_tmpD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\
> Debug\b_frame.obj
> > _tot_timeD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\
> Debug\b_frame.obj
> > _frame_noD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\
> Debug\b_frame.obj
> > _imgY_pfD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\
> Debug\b_frame.obj
> > _fbD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\
> Debug\b_frame.obj
> > _allocaD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\
> Debug\b_frame.obj
> > _p_refD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\
> Debug\b_frame.obj
> > _imgUV_prevD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\
> Debug\b_frame.obj
> > _nal_startcode_followsD:\a_nilesh\code comp sudio Projects\10 Jan
> deco\TryDec\Debug\b_frame.obj
> > _p_outD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\
> Debug\b_frame.obj
> > _Bframe_ctrD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\
> Debug\b_frame.obj
> > _imgY_prevD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\
> Debug\b_frame.obj
> > _nextP_trD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\
> Debug\b_frame.obj
> > _mcefD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\
> Debug\b_frame.obj
> > _imgUV_pfD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\
> Debug\b_frame.obj
> > _errortextD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\
> Debug\b_frame.obj
> > _refFrArrD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\
> Debug\b_frame.obj
> > _imgUV_refD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\
> Debug\b_frame.obj
> > _imgYD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\
> Debug\b_frame.obj
> > >>error: symbol referencing errors - './Debug/TryDec.out' not built
> > >> Compilation failure
> > Build Complete,
> > 2 Errors, 153 Warnings, 0 Remarks.
> > What can be reason? I have included .h file containing definition of
> global symbols as extern everywhere in .c file
> > where errors are being reported.
> >
> > Best Regards,
> > Nilesh
> >
> >
> > Jeff Brower wrote: Nilesh- > By JTAG u means
> to say emulator , right? If yes I have ext.
> > emulator option
> >> on dsk 6713. And if emulator becomes unavailable then is there any
> other way
> >> to communicate with host? I mean usb or other. Yes I'm using term
> 'JTAG' interchangeably with emulator. But with
> >> DSK 6713, keep in
> > mind that you don't need a separate emulator unit (such as Spectrum
> Digital XDS510),
> > because CCS supports emulation over the USB port connection. This is a
> feature that
> > TI builds into CCS for DSK boards, but third-party boards typically
> don't have (which
> > is why most 3p boards require the external emulator to use CCS debug
> functions). > In my case do I need to make 2
> > ccs projects? One for host and other for dsk.
> >> Or one program will suffice? Also what about memory mapping of project
>
> >> running on ccs should it be hypothetical then? Just one CCS project.
> Your DSP code can perform host file I/O, but
> >> it's slow and you
> > have to be careful with it. > Can I use vc++ to communicate with dsk?
> And will it be easy as compared to
> >> ccs? I hope not. What is your say? Well that's a good question. For
> DSK 6711, TI offers a DLL that can be used to
> >> allow
> > Visual Studio C/C++ programs to communicate via the HPI. It works but
> has some
> > reliability issues. But for DSK 6713, the HPI is left unconnected, and
> all
> > communication goes through the USB + JTAG connection. EducationalDSP
> makes a small
> > daughtercard that you can add to the DSK 6713 and connect HPI to Visual
> Studio
> > programs (http://www.educationaldsp.com/stockproduct_dsk6xxxhpi.htm).
> -Jeff > Jeff Brower wrote: Nilesh-
> >>
> >> > As I said last mail that we do not have 64xx series dsp's available
> in our
> >> > college I have no other option but to use 6713 dsp for implementing
> integer
> >> > dct. Here I have some queries as below:
> >> >
> >> >
> >> > Now if I want real time implementation of H.264 decoder on TI with
> just
> >> > integer dct on 6713 implemented, it means that I have to use file
> I\O's.
> >> > Is this facility available on 6713?
> >>
> >> Yes, available via CCS + JTAG, independent of DSP type.
> >>
> >> > I want to write residual dct coeff. data in dsp memory and get its
> dct and
> >> > again exchange idct o\p residual data with host pc through dsp
> memory,
> >> > which will then continue with rest of decoding. Will this be a
> simple
> >> > routine?
> >>
> >> The code for host file I/O is relatively simple but transfer rates are
> not fast.
> >>
> >> > Can I use interrupts for this real time exchange of dct? Which will
> be
> >> > better and less time to design.
> >> >
> >> > What type of synchronization will I have to use between host and
> dsp?
> >> > Assuming that I will use some flag in memory that will be polled by
> dsp
> >> > and host to start processing dct coeff. And idct residuals'
> >> > respectively. That is this is a sequential process for each frame or
>
> >> > macroblock even.
> >>
> >> For questions about host data transfer using CCS + JTAG, you should
> first study TI
> >> docs about "RTDX". You should not be setting up your own low-level
> interrupts and
> >> semaphores. The exception would be if you are using HPI, which you
> won't be doing on
> >> DSK 6713.
> >>
> >> > I initially do not mind how slow I have to keep my frame rate or
> >> > processed by host pc I want to display decoded bitstream in original
> yuv
> >> > format on my host while processing is being done, will this work?
> >>
> >> Should be possible, but image update rate will be slow.
> >>
> >> > What kind of difference is between 6711 and 6713? Is they are
> compatible
> >> > for code and h\w, memory mapping point of view?
> >>
> >> For your application, main differences are speed (6713 is twice as
> fast) and amount
> >> of onchip SRAM (6713 has 4x amount).
> >>
> >> -Jeff
> >>
> >> > Jeff Brower wrote: Nilesh-
> >> >
> >> > > Thanx for quick reply. In fact after reading all resopnses in
> lastr couple of days only i decided to ask query.
> >> Still
> >> > > i have some Q's in my mind. These are as below:
> >> > > 1. Is 64xx series latest as compared to 67xx?
> >> >
> >> > Yes. 300 MHz 6713 came out about 2003. 1 GHz 6416T device came out
> in 2005. Currently TI's fastest available
> >> device
> >> > is 6455, also 1 GHz but with 2x speed 64x+ core, released in 2006.
> >> >
> >> > > 2. How much approx. cost of 6413 dsp board will? i don't know
> weather it is or not available in my college so
> >> > > i have to check that.
> >> >
> >> > I think 6416 DSK is same cost as others, or close to it -- about USD
> 400.
> >> >
> >> > > 3. For Integer DCT implementation fixed point is best processor to
> use i guess is that right? and i think 64xx
> >> > > seris is fixed point dsp series...?
> >> >
> >> > Yes. 16x16 multipliers, 32-bit accumulators.
> >> >
> >> > > 4. I agree that 67xx based implementaion will not be an efficient
> method. But will it worth some value at least
> >> > > entry level work exp. as i am doing my final year project and not
> a professional work , so if no 6416 is
> >> available i
> >> > > mean to say...
> >> >
> >> > As I mentioned, for academic implementation one sub-block of H.264
> algorithm, 6713 should be Ok assuming the main
> >> > point is to obtain DSP learning and experience. I would add that
> after graduation during your job search, it would
> >> > look better if you had done image processing work using 641x.
> >> >
> >> > -Jeff
> >> >
> >> > > Jeff Brower wrote: Nilesh-
> >> > >
> >> > >> I am currently working on H.264 feature Integer DCT. I am
> modifying code
> >> > >> and want to implement this feature on 6713. I want to know that
> weather it
> >> > >> will be possible to do so?
> >> > >> Actually initially I was using setup for dm642 ccs. But since it
> is not
> >> > >> available in my college I decided to go for 6713. I am new to
> this and want
> >> > >> to know will it be possible. I just want to put integer dct
> implementation on 6713.
> >> > >> Any valuable suggation plz
> >> > >
> >> > > Suggest to look over recent C6x group discussion on subject "Using
> C6713 for image
> >> > > processing". Here is the thread:
> >> > >
> >> > > http://tech.groups.yahoo.com/group/c6x/message/8112
> >> > >
> >> > > For a specific sub-block of H.264, using 6713 might be Ok, but for
> fixed-point based
> >> > > algorithms it's going to a lot slower than DM642. You might try to
> locate a DSK
> >> > > 6416, which is low-cost and your Univ/college should have
> available.
> >> > >
> >> > > -Jeff
>
Sir,
I have code which is open source code. So there is no code from my side.
Regards,
Nilesh

Jeff Brower wrote: Nilesh-

> I have tried all possible ways to tackle this undefined symbol problem. I have noticed few things that are strange
> for me. For e.g. my .bss section in mapping file have length of 0. This is for decoder code ==>

Somehow, you're making this way too hard. Linking is hardly the thing that should cause a many-day problem.

Who makes file "b_frame.c" -- you? Or you get this from a third-party?

-Jeff

> .bss 0 00000000 00000000 UNINITIALIZED
>
> while for my succesfull encoder it shows this as ==>
>
> .bss 0 00005da0 00000158 UNINITIALIZED
> 00005da0 00000158 b_frame.obj (.bss)
>
> Also similar to following Lines are there for each undefined symbol e.g.
>
> DW$60 .dwtag DW_TAG_variable, DW_AT_name("imgUV"), DW_AT_symbol_name("_imgUV")
> .dwattr DW$60, DW_AT_type(*DW$T$131)
> .dwattr DW$60, DW_AT_declaration(0x01)
> .dwattr DW$60, DW_AT_external(0x01)
> ****************************************************************
>
> .global _imgUV
> .bss _imgUV,4,4
> DW$42 .dwtag DW_TAG_variable, DW_AT_name("imgUV"), DW_AT_symbol_name("_imgUV")
> .dwattr DW$42, DW_AT_location[DW_OP_addr _imgUV]
> .dwattr DW$42, DW_AT_type(*DW$T$145)
> .dwattr DW$42, DW_AT_external(0x01)
>
>
> Now second code in mapping file has .global directive with it in comparison with first. And second is from mapping
> file of encoder which is working ok. While first code was from mapping file of decoder giving linking error. So
> though symbol is defined it is not globally defined. Which I think is definitely wrong.
>
> Also for your information I have som definition in my main file asbr /> > #define GLOBAL_VARS
> And global.h file that is included everywhere as br /> > #ifdef GLOBAL_VARS
> #define EXTERN_GLOBAL
> #else
> #define EXTERN_GLOBAL extern
> #endif
> And I have used symbol before all global variable in global.h file as XTERN_GLOBAL
>
> I tested all code on vc++ and it is working fine. But am not able to figure out why undefined symbol error is there.
> Also my .cmd file is OK as below=>
>
>
> -c
> -heap 0x1F4FFFF
> -stack 0x2000000
> -l rts6700.lib
>
> MEMORY
> {
> L2SRAM: origin = 0x00000000 length = 00008000h
> CE0: origin = 0x80000000 length = 0xB0000
> CEO: origin = 0x800B0000 length = 0x3F4FFFF
> CE1: origin = 0x90000000 length = 0x3FFFFFF
> CE2: origin = 0xA0000000 length = 0x3FFFFFF
> CE3: origin = 0xB0000000 length = 0x3FFFFFF
> }
>
> SECTIONS
> {
> /* .sysmem > CEO
> .stack > CEO*/
> .sysmem: {*(.sysmem) } > CEO | CE1 | CE2 | CE3
> .stack: {*(.stack) } > CEO | CE1 | CE2 | CE3
>
> .cinit > L2SRAM
> .bss: { /*b_frame.obj (.bss)*/
> *(.bss) } > L2SRAM
> .const > L2SRAM
> .cio > L2SRAM
> .switch > L2SRAM
> .data > L2SRAM
> .tables > L2SRAM
>
> .far > CE0
> .external > CE0
> .text > CE0
> }
>
> I need kindly your suggestions
> Regards,
> Nilesh
>
> Jeff Brower wrote:
> Nilesh- > ## Do u means to say that I can use embedded JTAG emulator
>> with just usb cable to pc and from CCS? Yes. > What configuration setting I have to make for communication
>> through this then? The version of CCS that comes with DSK boards handles this automatically. > ## If I need to
>> use one CCS project then :
>> My small portion of code i.e. integer dct is going to be on
>> DSP 6713 so I need to download that part to 6713. So I was
>> thinking to make 2 CCS project separately.
>> But if I can have this from one CCS project also possible then
>> can u please explain me, how can I differentiate between part
>> to be downloaded on target and remaining part that will be
>> running on host through CCS. Do I need to have different
>> sectional mapping to be used? I can understand why you would ask this, but you're thinking of the "tree" instead
>> of the "forest". Here are the main possibilities for host-side software: -CCS only. Edit/compile/build/link
>> code, debug functions,
> waveform and data display functions, file I/O -Simulink/MATLAB + CCS. Block diagrams with one or more
> blocks that run on the DSP. Simulink uses CCS for DSP
> code download and communication -Visual Studio + HPI DLL. C/C++ code that talks to DSK
> board (See previous post for notes) Hopefully this explains why you don't need more than one CCS project. The
> main thing you are missing is that CCS is both a GUI for code development (compile code) and also contains debug,
> I/O, and other target (board) communication functions. I suppose this is why TI calls it an IDE -- integrated
> development environment. > ## Also I am getting following kind of errors while linking on CCS 6713:
>> [Linking...] "C:\CCStudio_v3.1\C6000\cgtools\bin\cl6x" -@"Debug.lkf" Linking errors are basic and you should be
>> able to figure out on your own by studying TI's example projects. As just a brief suggestion, I might guess that
>> you don't have b_frame listed in your .cmd (linker command) file. -Jeff
>
> undefinedfirst referenced
> symbolin file
> -------------------------
> _mrefD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\Debug\b_frame.obj
> _imgY_refD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\Debug\b_frame.obj
> _imgUV_tmpD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\Debug\b_frame.obj
> _P_intervalD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\Debug\b_frame.obj
> _ftimeD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\Debug\b_frame.obj
> _prevP_trD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\Debug\b_frame.obj
> _p_traceD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\Debug\b_frame.obj
> _imgUVD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\Debug\b_frame.obj
> _imgY_tmpD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\Debug\b_frame.obj
> _tot_timeD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\Debug\b_frame.obj
> _frame_noD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\Debug\b_frame.obj
> _imgY_pfD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\Debug\b_frame.obj
> _fbD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\Debug\b_frame.obj
> _allocaD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\Debug\b_frame.obj
> _p_refD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\Debug\b_frame.obj
> _imgUV_prevD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\Debug\b_frame.obj
> _nal_startcode_followsD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\Debug\b_frame.obj
> _p_outD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\Debug\b_frame.obj
> _Bframe_ctrD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\Debug\b_frame.obj
> _imgY_prevD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\Debug\b_frame.obj
> _nextP_trD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\Debug\b_frame.obj
> _mcefD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\Debug\b_frame.obj
> _imgUV_pfD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\Debug\b_frame.obj
> _errortextD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\Debug\b_frame.obj
> _refFrArrD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\Debug\b_frame.obj
> _imgUV_refD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\Debug\b_frame.obj
> _imgYD:\a_nilesh\code comp sudio Projects\10 Jan deco\TryDec\Debug\b_frame.obj
> >>error: symbol referencing errors - './Debug/TryDec.out' not built
> >> Compilation failure
> Build Complete,
> 2 Errors, 153 Warnings, 0 Remarks.
> What can be reason? I have included .h file containing definition of global symbols as extern everywhere in .c file
> where errors are being reported.
>
> Best Regards,
> Nilesh
>
>
> Jeff Brower wrote: Nilesh- > By JTAG u means to say emulator , right? If yes I have ext.
> emulator option
>> on dsk 6713. And if emulator becomes unavailable then is there any other way
>> to communicate with host? I mean usb or other. Yes I'm using term 'JTAG' interchangeably with emulator. But with
>> DSK 6713, keep in
> mind that you don't need a separate emulator unit (such as Spectrum Digital XDS510),
> because CCS supports emulation over the USB port connection. This is a feature that
> TI builds into CCS for DSK boards, but third-party boards typically don't have (which
> is why most 3p boards require the external emulator to use CCS debug functions). > In my case do I need to make 2
> ccs projects? One for host and other for dsk.
>> Or one program will suffice? Also what about memory mapping of project
>> running on ccs should it be hypothetical then? Just one CCS project. Your DSP code can perform host file I/O, but
>> it's slow and you
> have to be careful with it. > Can I use vc++ to communicate with dsk? And will it be easy as compared to
>> ccs? I hope not. What is your say? Well that's a good question. For DSK 6711, TI offers a DLL that can be used to
>> allow
> Visual Studio C/C++ programs to communicate via the HPI. It works but has some
> reliability issues. But for DSK 6713, the HPI is left unconnected, and all
> communication goes through the USB + JTAG connection. EducationalDSP makes a small
> daughtercard that you can add to the DSK 6713 and connect HPI to Visual Studio
> programs (http://www.educationaldsp.com/stockproduct_dsk6xxxhpi.htm). -Jeff > Jeff Brower wrote: Nilesh-
>>
>> > As I said last mail that we do not have 64xx series dsp's available in our
>> > college I have no other option but to use 6713 dsp for implementing integer
>> > dct. Here I have some queries as below:
>> >
>> >
>> > Now if I want real time implementation of H.264 decoder on TI with just
>> > integer dct on 6713 implemented, it means that I have to use file I\O's.
>> > Is this facility available on 6713?
>>
>> Yes, available via CCS + JTAG, independent of DSP type.
>>
>> > I want to write residual dct coeff. data in dsp memory and get its dct and
>> > again exchange idct o\p residual data with host pc through dsp memory,
>> > which will then continue with rest of decoding. Will this be a simple
>> > routine?
>>
>> The code for host file I/O is relatively simple but transfer rates are not fast.
>>
>> > Can I use interrupts for this real time exchange of dct? Which will be
>> > better and less time to design.
>> >
>> > What type of synchronization will I have to use between host and dsp?
>> > Assuming that I will use some flag in memory that will be polled by dsp
>> > and host to start processing dct coeff. And idct residuals'
>> > respectively. That is this is a sequential process for each frame or
>> > macroblock even.
>>
>> For questions about host data transfer using CCS + JTAG, you should first study TI
>> docs about "RTDX". You should not be setting up your own low-level interrupts and
>> semaphores. The exception would be if you are using HPI, which you won't be doing on
>> DSK 6713.
>>
>> > I initially do not mind how slow I have to keep my frame rate or
>> > processed by host pc I want to display decoded bitstream in original yuv
>> > format on my host while processing is being done, will this work?
>>
>> Should be possible, but image update rate will be slow.
>>
>> > What kind of difference is between 6711 and 6713? Is they are compatible
>> > for code and h\w, memory mapping point of view?
>>
>> For your application, main differences are speed (6713 is twice as fast) and amount
>> of onchip SRAM (6713 has 4x amount).
>>
>> -Jeff
>>
>> > Jeff Brower wrote: Nilesh-
>> >
>> > > Thanx for quick reply. In fact after reading all resopnses in lastr couple of days only i decided to ask query.
>> Still
>> > > i have some Q's in my mind. These are as below:
>> > > 1. Is 64xx series latest as compared to 67xx?
>> >
>> > Yes. 300 MHz 6713 came out about 2003. 1 GHz 6416T device came out in 2005. Currently TI's fastest available
>> device
>> > is 6455, also 1 GHz but with 2x speed 64x+ core, released in 2006.
>> >
>> > > 2. How much approx. cost of 6413 dsp board will? i don't know weather it is or not available in my college so
>> > > i have to check that.
>> >
>> > I think 6416 DSK is same cost as others, or close to it -- about USD 400.
>> >
>> > > 3. For Integer DCT implementation fixed point is best processor to use i guess is that right? and i think 64xx
>> > > seris is fixed point dsp series...?
>> >
>> > Yes. 16x16 multipliers, 32-bit accumulators.
>> >
>> > > 4. I agree that 67xx based implementaion will not be an efficient method. But will it worth some value at least
>> > > entry level work exp. as i am doing my final year project and not a professional work , so if no 6416 is
>> available i
>> > > mean to say...
>> >
>> > As I mentioned, for academic implementation one sub-block of H.264 algorithm, 6713 should be Ok assuming the main
>> > point is to obtain DSP learning and experience. I would add that after graduation during your job search, it would
>> > look better if you had done image processing work using 641x.
>> >
>> > -Jeff
>> >
>> > > Jeff Brower wrote: Nilesh-
>> > >
>> > >> I am currently working on H.264 feature Integer DCT. I am modifying code
>> > >> and want to implement this feature on 6713. I want to know that weather it
>> > >> will be possible to do so?
>> > >> Actually initially I was using setup for dm642 ccs. But since it is not
>> > >> available in my college I decided to go for 6713. I am new to this and want
>> > >> to know will it be possible. I just want to put integer dct implementation on 6713.
>> > >> Any valuable suggation plz
>> > >
>> > > Suggest to look over recent C6x group discussion on subject "Using C6713 for image
>> > > processing". Here is the thread:
>> > >
>> > > http://tech.groups.yahoo.com/group/c6x/message/8112
>> > >
>> > > For a specific sub-block of H.264, using 6713 might be Ok, but for fixed-point based
>> > > algorithms it's going to a lot slower than DM642. You might try to locate a DSK
>> > > 6416, which is low-cost and your Univ/college should have available.
>> > >
>> > > -Jeff