Hi CCS users,
Am I the only one here that has various problems with running CCS 3.1
under XP SP2? I believe that I've not yet seen any trouble reports
regarding CCS 3.1 in this forum.
I do like the new connect/disconnect feature for the emulator and
finally the 'Find in Files ...' function is correctly working now. But
I
have come across various new problems that I have never encountered in
CCS 2.21:
- C55xx Rev2.x CPU Cycle Accurate Simulator does crash very often.
- Profiling via the clock counter does not work. Reported cycles are way
below the real numbers.
- Still have not found out why opening DSP/BIOS projects trigger error
messages when I'm logged into XP without administrator privileges.
Although code can be run with the emulator, all DSP/BIOS analysis
features do not work. As is the case with most serious problems,
epic@epic... was clueless and not able to help with this particular
problem either.
Have you guys come across similar problems? Do you guys just keep
working with CCS 2.21 or did I just simply miss something here?
Best Regards,
Roland
CCS 3.1 Problems?
Started by ●October 19, 2005
Reply by ●October 20, 20052005-10-20
Roland-
> Am I the only one here that has various problems with running CCS 3.1
> under XP SP2? I believe that I've not yet seen any trouble reports
> regarding CCS 3.1 in this forum.
> I do like the new connect/disconnect feature for the emulator and
> finally the 'Find in Files ...' function is correctly working now. But I
> have come across various new problems that I have never encountered in
> CCS 2.21:
>
> - C55xx Rev2.x CPU Cycle Accurate Simulator does crash very often.
>
> - Profiling via the clock counter does not work. Reported cycles are way
> below the real numbers.
>
> - Still have not found out why opening DSP/BIOS projects trigger error
> messages when I'm logged into XP without administrator privileges.
> Although code can be run with the emulator, all DSP/BIOS analysis
> features do not work. As is the case with most serious problems,
> epic@epic... was clueless and not able to help with this particular
> problem either.
>
> Have you guys come across similar problems? Do you guys just keep
> working with CCS 2.21 or did I just simply miss something here?
What are you expecting, Visual Studio? Given there are orders of magnitude fewer DSP
developers than GCC / Visual Studio developers, I think CCS is exceptional --
certainly compared against other DSP vendor tools.
But yes, lots of bugs and weirdnesses, you are right about this, and it has always
been the case for many years. What we do is find ways to get by. Sometimes we use
v2.x if that's needed, sometimes run it on another PC that for whatever reason works
better, sometimes just avoid the feature or function that causes the problem.
For example, your profiling issues above? Just run the code on a DSK 5510 board and
account for difference in clock frequency with your target C55xx device. Once you
have real code running, you can profile it many ways.
-Jeff
> Am I the only one here that has various problems with running CCS 3.1
> under XP SP2? I believe that I've not yet seen any trouble reports
> regarding CCS 3.1 in this forum.
> I do like the new connect/disconnect feature for the emulator and
> finally the 'Find in Files ...' function is correctly working now. But I
> have come across various new problems that I have never encountered in
> CCS 2.21:
>
> - C55xx Rev2.x CPU Cycle Accurate Simulator does crash very often.
>
> - Profiling via the clock counter does not work. Reported cycles are way
> below the real numbers.
>
> - Still have not found out why opening DSP/BIOS projects trigger error
> messages when I'm logged into XP without administrator privileges.
> Although code can be run with the emulator, all DSP/BIOS analysis
> features do not work. As is the case with most serious problems,
> epic@epic... was clueless and not able to help with this particular
> problem either.
>
> Have you guys come across similar problems? Do you guys just keep
> working with CCS 2.21 or did I just simply miss something here?
What are you expecting, Visual Studio? Given there are orders of magnitude fewer DSP
developers than GCC / Visual Studio developers, I think CCS is exceptional --
certainly compared against other DSP vendor tools.
But yes, lots of bugs and weirdnesses, you are right about this, and it has always
been the case for many years. What we do is find ways to get by. Sometimes we use
v2.x if that's needed, sometimes run it on another PC that for whatever reason works
better, sometimes just avoid the feature or function that causes the problem.
For example, your profiling issues above? Just run the code on a DSK 5510 board and
account for difference in clock frequency with your target C55xx device. Once you
have real code running, you can profile it many ways.
-Jeff
Reply by ●October 20, 20052005-10-20
Jeff-
No, I do not expect the stability of Visual Studio. I have been working
with CCS since its beginning, when it was still distributed by GO DSP,
so I do know that one has to live with workarounds for coping with its
faults. I also agree with you that CCS is an exceptional tool within the
entire embedded world. But you miss my point here Jeff, which is that
CCS 2.21 was quite stable as compared to the newest release of 3.1. You
probably will tell me know to forget about 3.1 and go back to 2.21
(which I probably will have to do). I just feel a little resentment in
accepting that a new release of an IDE is allowed to perform worse than
its previous version. If this should be a pattern, then TI better stop
its continued enhancement work of CCS right away. The fact that a
simulator crashes so frequently is a serious fault and there is no
workaround except not using the simulator at all. I do expect that an
often used feature which has worked fine in a previous release will
still work fine in a next release, otherwise upgrading makes no sense.
Ironically, going back to 2.21 would also mean for me that I will have
to run Visual Studio next to CCS, because I'll have to resort to using
Visual Studio whenever I want to perform a 'Find in Files ..' search.
Regards,
Roland
-----Original Message-----
From: Jeff Brower [mailto:jbrower@jbro...]
Sent: Thursday, October 20, 2005 17:10
To: Welte Roland
Cc: code-comp@code...
Subject: Re: [code-comp] CCS 3.1 Problems? Roland-
> Am I the only one here that has various problems with running CCS 3.1
> under XP SP2? I believe that I've not yet seen any trouble reports
> regarding CCS 3.1 in this forum. I do like the new connect/disconnect
> feature for the emulator and finally the 'Find in Files ...' function
> is correctly working now. But I have come across various new problems
> that I have never encountered in CCS 2.21:
>
> - C55xx Rev2.x CPU Cycle Accurate Simulator does crash very often.
>
> - Profiling via the clock counter does not work. Reported cycles are
> way below the real numbers.
>
> - Still have not found out why opening DSP/BIOS projects trigger error
> messages when I'm logged into XP without administrator privileges.
> Although code can be run with the emulator, all DSP/BIOS analysis
> features do not work. As is the case with most serious problems,
> epic@epic... was clueless and not able to help with this particular
> problem either.
>
> Have you guys come across similar problems? Do you guys just keep
> working with CCS 2.21 or did I just simply miss something here?
What are you expecting, Visual Studio? Given there are orders of
magnitude fewer DSP developers than GCC / Visual Studio developers, I
think CCS is exceptional -- certainly compared against other DSP vendor
tools.
But yes, lots of bugs and weirdnesses, you are right about this, and it
has always been the case for many years. What we do is find ways to get
by. Sometimes we use v2.x if that's needed, sometimes run it on another
PC that for whatever reason works better, sometimes just avoid the
feature or function that causes the problem.
For example, your profiling issues above? Just run the code on a DSK
5510 board and account for difference in clock frequency with your
target C55xx device. Once you have real code running, you can profile
it many ways.
-Jeff
No, I do not expect the stability of Visual Studio. I have been working
with CCS since its beginning, when it was still distributed by GO DSP,
so I do know that one has to live with workarounds for coping with its
faults. I also agree with you that CCS is an exceptional tool within the
entire embedded world. But you miss my point here Jeff, which is that
CCS 2.21 was quite stable as compared to the newest release of 3.1. You
probably will tell me know to forget about 3.1 and go back to 2.21
(which I probably will have to do). I just feel a little resentment in
accepting that a new release of an IDE is allowed to perform worse than
its previous version. If this should be a pattern, then TI better stop
its continued enhancement work of CCS right away. The fact that a
simulator crashes so frequently is a serious fault and there is no
workaround except not using the simulator at all. I do expect that an
often used feature which has worked fine in a previous release will
still work fine in a next release, otherwise upgrading makes no sense.
Ironically, going back to 2.21 would also mean for me that I will have
to run Visual Studio next to CCS, because I'll have to resort to using
Visual Studio whenever I want to perform a 'Find in Files ..' search.
Regards,
Roland
-----Original Message-----
From: Jeff Brower [mailto:jbrower@jbro...]
Sent: Thursday, October 20, 2005 17:10
To: Welte Roland
Cc: code-comp@code...
Subject: Re: [code-comp] CCS 3.1 Problems? Roland-
> Am I the only one here that has various problems with running CCS 3.1
> under XP SP2? I believe that I've not yet seen any trouble reports
> regarding CCS 3.1 in this forum. I do like the new connect/disconnect
> feature for the emulator and finally the 'Find in Files ...' function
> is correctly working now. But I have come across various new problems
> that I have never encountered in CCS 2.21:
>
> - C55xx Rev2.x CPU Cycle Accurate Simulator does crash very often.
>
> - Profiling via the clock counter does not work. Reported cycles are
> way below the real numbers.
>
> - Still have not found out why opening DSP/BIOS projects trigger error
> messages when I'm logged into XP without administrator privileges.
> Although code can be run with the emulator, all DSP/BIOS analysis
> features do not work. As is the case with most serious problems,
> epic@epic... was clueless and not able to help with this particular
> problem either.
>
> Have you guys come across similar problems? Do you guys just keep
> working with CCS 2.21 or did I just simply miss something here?
What are you expecting, Visual Studio? Given there are orders of
magnitude fewer DSP developers than GCC / Visual Studio developers, I
think CCS is exceptional -- certainly compared against other DSP vendor
tools.
But yes, lots of bugs and weirdnesses, you are right about this, and it
has always been the case for many years. What we do is find ways to get
by. Sometimes we use v2.x if that's needed, sometimes run it on another
PC that for whatever reason works better, sometimes just avoid the
feature or function that causes the problem.
For example, your profiling issues above? Just run the code on a DSK
5510 board and account for difference in clock frequency with your
target C55xx device. Once you have real code running, you can profile
it many ways.
-Jeff
Reply by ●October 22, 20052005-10-22
Roland-
> No, I do not expect the stability of Visual Studio. I have been working
> with CCS since its beginning, when it was still distributed by GO DSP,
> so I do know that one has to live with workarounds for coping with its
> faults. I also agree with you that CCS is an exceptional tool within the
> entire embedded world. But you miss my point here Jeff, which is that
> CCS 2.21 was quite stable as compared to the newest release of 3.1. You
> probably will tell me know to forget about 3.1 and go back to 2.21
> (which I probably will have to do). I just feel a little resentment in
> accepting that a new release of an IDE is allowed to perform worse than
> its previous version. If this should be a pattern, then TI better stop
> its continued enhancement work of CCS right away. The fact that a
> simulator crashes so frequently is a serious fault and there is no
> workaround except not using the simulator at all. I do expect that an
> often used feature which has worked fine in a previous release will
> still work fine in a next release, otherwise upgrading makes no sense.
Yes I see your point more clearly now. Something that was working should still be
working, even if the feature is not the first thing most users will encounter. That
implies less than rigorous testing.
> Ironically, going back to 2.21 would also mean for me that I will have
> to run Visual Studio next to CCS, because I'll have to resort to using
> Visual Studio whenever I want to perform a 'Find in Files ..' search.
Yes I know, Find-in-Files is part of the litmus-test for CCS vs. Visual Studio --
"are we there yet?" :-)
One suggestion I might make is to get hooked up on the next round of CCS beta
testing. If you're interested, mail me off-line and I can ask TI people I know to
consider you on the next one.
-Jeff
> -----Original Message-----
> From: Jeff Brower [mailto:jbrower@jbro...]
> Sent: Thursday, October 20, 2005 17:10
> To: Welte Roland
> Cc: code-comp@code...
> Subject: Re: [code-comp] CCS 3.1 Problems?
>
> Roland-
>
> > Am I the only one here that has various problems with running CCS 3.1
> > under XP SP2? I believe that I've not yet seen any trouble reports
> > regarding CCS 3.1 in this forum. I do like the new connect/disconnect
> > feature for the emulator and finally the 'Find in Files ...' function
> > is correctly working now. But I have come across various new problems
> > that I have never encountered in CCS 2.21:
> >
> > - C55xx Rev2.x CPU Cycle Accurate Simulator does crash very often.
> >
> > - Profiling via the clock counter does not work. Reported cycles are
> > way below the real numbers.
> >
> > - Still have not found out why opening DSP/BIOS projects trigger error
>
> > messages when I'm logged into XP without administrator privileges.
> > Although code can be run with the emulator, all DSP/BIOS analysis
> > features do not work. As is the case with most serious problems,
> > epic@epic... was clueless and not able to help with this particular
> > problem either.
> >
> > Have you guys come across similar problems? Do you guys just keep
> > working with CCS 2.21 or did I just simply miss something here?
>
> What are you expecting, Visual Studio? Given there are orders of
> magnitude fewer DSP developers than GCC / Visual Studio developers, I
> think CCS is exceptional -- certainly compared against other DSP vendor
> tools.
>
> But yes, lots of bugs and weirdnesses, you are right about this, and it
> has always been the case for many years. What we do is find ways to get
> by. Sometimes we use v2.x if that's needed, sometimes run it on another
> PC that for whatever reason works better, sometimes just avoid the
> feature or function that causes the problem.
>
> For example, your profiling issues above? Just run the code on a DSK
> 5510 board and account for difference in clock frequency with your
> target C55xx device. Once you have real code running, you can profile
> it many ways.
>
> -Jeff >
>
> No, I do not expect the stability of Visual Studio. I have been working
> with CCS since its beginning, when it was still distributed by GO DSP,
> so I do know that one has to live with workarounds for coping with its
> faults. I also agree with you that CCS is an exceptional tool within the
> entire embedded world. But you miss my point here Jeff, which is that
> CCS 2.21 was quite stable as compared to the newest release of 3.1. You
> probably will tell me know to forget about 3.1 and go back to 2.21
> (which I probably will have to do). I just feel a little resentment in
> accepting that a new release of an IDE is allowed to perform worse than
> its previous version. If this should be a pattern, then TI better stop
> its continued enhancement work of CCS right away. The fact that a
> simulator crashes so frequently is a serious fault and there is no
> workaround except not using the simulator at all. I do expect that an
> often used feature which has worked fine in a previous release will
> still work fine in a next release, otherwise upgrading makes no sense.
Yes I see your point more clearly now. Something that was working should still be
working, even if the feature is not the first thing most users will encounter. That
implies less than rigorous testing.
> Ironically, going back to 2.21 would also mean for me that I will have
> to run Visual Studio next to CCS, because I'll have to resort to using
> Visual Studio whenever I want to perform a 'Find in Files ..' search.
Yes I know, Find-in-Files is part of the litmus-test for CCS vs. Visual Studio --
"are we there yet?" :-)
One suggestion I might make is to get hooked up on the next round of CCS beta
testing. If you're interested, mail me off-line and I can ask TI people I know to
consider you on the next one.
-Jeff
> -----Original Message-----
> From: Jeff Brower [mailto:jbrower@jbro...]
> Sent: Thursday, October 20, 2005 17:10
> To: Welte Roland
> Cc: code-comp@code...
> Subject: Re: [code-comp] CCS 3.1 Problems?
>
> Roland-
>
> > Am I the only one here that has various problems with running CCS 3.1
> > under XP SP2? I believe that I've not yet seen any trouble reports
> > regarding CCS 3.1 in this forum. I do like the new connect/disconnect
> > feature for the emulator and finally the 'Find in Files ...' function
> > is correctly working now. But I have come across various new problems
> > that I have never encountered in CCS 2.21:
> >
> > - C55xx Rev2.x CPU Cycle Accurate Simulator does crash very often.
> >
> > - Profiling via the clock counter does not work. Reported cycles are
> > way below the real numbers.
> >
> > - Still have not found out why opening DSP/BIOS projects trigger error
>
> > messages when I'm logged into XP without administrator privileges.
> > Although code can be run with the emulator, all DSP/BIOS analysis
> > features do not work. As is the case with most serious problems,
> > epic@epic... was clueless and not able to help with this particular
> > problem either.
> >
> > Have you guys come across similar problems? Do you guys just keep
> > working with CCS 2.21 or did I just simply miss something here?
>
> What are you expecting, Visual Studio? Given there are orders of
> magnitude fewer DSP developers than GCC / Visual Studio developers, I
> think CCS is exceptional -- certainly compared against other DSP vendor
> tools.
>
> But yes, lots of bugs and weirdnesses, you are right about this, and it
> has always been the case for many years. What we do is find ways to get
> by. Sometimes we use v2.x if that's needed, sometimes run it on another
> PC that for whatever reason works better, sometimes just avoid the
> feature or function that causes the problem.
>
> For example, your profiling issues above? Just run the code on a DSK
> 5510 board and account for difference in clock frequency with your
> target C55xx device. Once you have real code running, you can profile
> it many ways.
>
> -Jeff >
>
