XScope Issues on OSX?

[stephencwelch]

Hello, I am currently working on an audio application using a slice board. I spent more time than I would care to admit yesterday working on getting the XScope working. I was finally able to get the scope working on Windows 7 (after trying OSX and ubuntu builds), but the results are still patchy - the scope does not always work, and seems to prevent my code from running depending on the implementation.

Has anyone else had issues getting the scope working or had a perhaps the opposite - a completely frictionless xScope experience? Any tips, tricks, or info here would be much appreciated, I am happy to provide my setup details as needed.

Thanks!

Stephen

[Redeye]

Stephen, I feel your pain.

Yes, I run mostly on OSX and I've had more than a few struggles with xScope. I don't know whether it's worse on OSX than any other platform, but yes it can be pretty patchy. I recently spent a long time searching for an error on my custom PCB until I eventually found out that the reason xScope wasn't working was a bug in the tools. I'd hope that isn't your problem if you're using XMOS boards, but I wouldn't rule it out.

Some advice from my experiences with xScope :

1. Real-time mode is hopeless, it just doesn't work in my experience.
2. If your Mac goes to sleep mode, you'll need to disconnect/reconnect the XTAG2 before the tools will see it again.
3. Matt at XMOS has been really helpful with the problems I've had with the tools (but others aren't always so helpful).
4. If you're tracing audio data in xScope, the files can get very large very quickly and it can take a long time between pressing the stop button in the tools and the output file getting written. The IDE goes unresponsive in the meantime, which is quite annoying.

I've learned to work around the various problems, but it has been a bit painful at times. I'm hoping there will be stability/speed improvements in xScope in the v13 tools.

[stephencwelch]

Thanks for the perspective, glad to hear I'm not the only one having troubles. The scope would be a very useful tool for me if it worked in real time. I noticed that in the xScope demo video, the engineer is using ubuntu, that's why I gave it a shot on my ubuntu machine, but the install turned out to be super buggy unfortunately, and I ran into some driver issues that I couldn't resolve quickly. I've been able to debug pretty successfully today using strategic print commands, and based on your experience, I may stick with this approach in the meantime.

This does make me nervous for the future though - right now we are just prototyping on a slice board, when we move to our own boards later this year I'm sure debugging will be much tougher, and perhaps impossible without the scope. Hopefully I can figure something out by then.

Thanks!

[pstnotpd]

...I noticed that in the xScope demo video, the engineer is using ubuntu, that's why I gave it a shot on my ubuntu machine, but the install turned out to be super buggy unfortunately, and I ran into some driver issues that I couldn't resolve quickly. ...

Which issues did you get on ubuntu? I'm running 13.04 64bit and after resolving the eclipse bug as described in the FAQ it is running quite smoothly.

Granted that I haven't tried the xScope yet.....

[XMatt]

Stephen, I feel your pain.

Yes, I run mostly on OSX and I've had more than a few struggles with xScope. I don't know whether it's worse on OSX than any other platform, but yes it can be pretty patchy. I recently spent a long time searching for an error on my custom PCB until I eventually found out that the reason xScope wasn't working was a bug in the tools. I'd hope that isn't your problem if you're using XMOS boards, but I wouldn't rule it out.

Some advice from my experiences with xScope :

1. Real-time mode is hopeless, it just doesn't work in my experience.
2. If your Mac goes to sleep mode, you'll need to disconnect/reconnect the XTAG2 before the tools will see it again.
3. Matt at XMOS has been really helpful with the problems I've had with the tools (but others aren't always so helpful).
4. If you're tracing audio data in xScope, the files can get very large very quickly and it can take a long time between pressing the stop button in the tools and the output file getting written. The IDE goes unresponsive in the meantime, which is quite annoying.

I've learned to work around the various problems, but it has been a bit painful at times. I'm hoping there will be stability/speed improvements in xScope in the v13 tools.

Hi Redeye,

As a general point the 13.0 beta 1 release has had a lot of work put into it to improve the stability of xSCOPE so this would be worth trying. I believe we have addressed all the issues you mention in version 13.

To answer the points your have raised,

1. There was a bug in the underlying eclipse platform we use for xTIMEcomposer which was stopping the real time display working correctly on OSX, we believe this is now fixed but if it is not we will look into it again.
2. The XTAG-2 firmware does not support USB suspend at this time, this is targeted to be fixed in production release of 13, at the moment you will as you say have to disconnect and reconnect the XTAG-2 after your machine goes to sleep.
3. Thanks ;-)
4. In version 13 xSCOPE now streams to file rather than buffers, this will result in much better performance and resolve this problem with it having to write out a large file at the end, you will still however get a large file if you are tracing a lot of data.

[XMatt]

Hello, I am currently working on an audio application using a slice board. I spent more time than I would care to admit yesterday working on getting the XScope working. I was finally able to get the scope working on Windows 7 (after trying OSX and ubuntu builds), but the results are still patchy - the scope does not always work, and seems to prevent my code from running depending on the implementation.

Has anyone else had issues getting the scope working or had a perhaps the opposite - a completely frictionless xScope experience? Any tips, tricks, or info here would be much appreciated, I am happy to provide my setup details as needed.

Thanks!

Stephen

Hi Stephen,

The examples that are provided within the tools for xSCOPE should allow you to test that everything is working correctly. If they dont or you have issues then we will look into it as they are all tested to work correctly on slicekit.

Matt

[stephencwelch]

Thanks pstnotpd, yeah I followed that FAQ, but it actually caused some display issues, I generally don't like installing packages without a package manager as the FAQ suggests - where did you install pango an glib?

Thanks.

[stephencwelch]

Hi Stephen,

The examples that are provided within the tools for xSCOPE should allow you to test that everything is working correctly. If they dont or you have issues then we will look into it as they are all tested to work correctly on slicekit.

Matt

Thanks Matt, yeah I tried this on OSX with no luck, worked fine on Windows. I'll try the beta version as you suggest.

Thanks.

[pstnotpd]

Thanks pstnotpd, yeah I followed that FAQ, but it actually caused some display issues, I generally don't like installing packages without a package manager as the FAQ suggests - where did you install pango an glib?

I built them under my XMOS install directory (/home/psm/XMOS) and adjusted the actual xtimecomposer startup file by adjusting the PKG_CONFIG_PATH and LD_LIBRARY_PATH as follows:

Code: Select all

export PKG_CONFIG_PATH=$PKG_CONFIG_PATH:/home/psm/XMOS/xTIME_ubuntu_deps/lib/pkgconfig
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/psm/XMOS/xTIME_ubuntu_deps/lib

abspath=$(cd ${0%/*} && echo $PWD/${0##*/})
path_only=`dirname "$abspath"`
cd $path_only
source SetEnv
cd - > /dev/null
xtimecomposer

That way it only applies to the xtimcomposer startup and leaves the rest of the OS alone. Works fine.

[Redeye]

Hi Redeye,

As a general point the 13.0 beta 1 release has had a lot of work put into it to improve the stability of xSCOPE so this would be worth trying. I believe we have addressed all the issues you mention in version 13.

To answer the points your have raised,

1. There was a bug in the underlying eclipse platform we use for xTIMEcomposer which was stopping the real time display working correctly on OSX, we believe this is now fixed but if it is not we will look into it again.
2. The XTAG-2 firmware does not support USB suspend at this time, this is targeted to be fixed in production release of 13, at the moment you will as you say have to disconnect and reconnect the XTAG-2 after your machine goes to sleep.
3. Thanks ;-)
4. In version 13 xSCOPE now streams to file rather than buffers, this will result in much better performance and resolve this problem with it having to write out a large file at the end, you will still however get a large file if you are tracing a lot of data.

Thanks Matt - as if to prove my point about your excellent support!!

I'll give the v13 tools a try at some point then. I have to admit to being slightly nervous about making the switch as I know there are a lot of changes in it. To be honest I'm far more interested in stability and speed/performance improvements than any of the new features.