DDR3 Corruption on Wandboard

Discussion of your EDM baseboard, your add-on boards or other peripherals for your wandboard.

Re: DDR3 Corruption on Wandboard

Postby Tapani » Thu Jul 17, 2014 1:56 am

Actually, my first guess on what might be wrong was "heat". See my previous post.

We did CPU burn in tests when we designed the heatsink, but you seem to be stressing all subsystems at the same time (we tried too), but maybe you are stressing them more efficiently :-)
Just to test, try adding a fan and run the failing units at 1GHz, and see if the situation improves.

Running at 1.2 will fail for sure. Freescale sells the ones that passes 1.2GHz as 1.2GHz marked CPUs, the rest as 1GHz ones.
Tapani
 
Posts: 712
Joined: Tue Aug 27, 2013 8:32 am

Re: DDR3 Corruption on Wandboard

Postby tdefeo23 » Thu Jul 17, 2014 4:42 pm

Ok, so I just realized that I am breaking the rules by running the chip at 1.2Ghz since it is only rated at 1.0Ghz :D

I am still having issues at 996Mhz, and I think it is interesting that the failure mode at 996Mhz is the same as the failure mode when running at 1.2Ghz, it just happens less frequently at 996Mhz, and does not fail at 792Mhz.

I am working with Freescale to try to identify the root cause, and will follow up if I find out anything more.

If anyone is having similar issues, or knows of a potential solution, let me know, I'm open to trying anything at this point!

Thanks,
Tony
tdefeo23
 
Posts: 13
Joined: Mon Jan 20, 2014 10:34 pm

Re: DDR3 Corruption on Wandboard

Postby tdefeo23 » Thu Jul 17, 2014 4:49 pm

Hi Tapani,

Sorry, I missed your last reply.
The temperature measured from both the internal sensor, and from a thermometer on the case is around ~50C, which is well within spec.
We did try the fan anyway, and it didn't make a difference (other then cooling it down to around 30C).

Thanks again for your help, I'll keep looking.....

Tony
tdefeo23
 
Posts: 13
Joined: Mon Jan 20, 2014 10:34 pm

Re: DDR3 Corruption on Wandboard

Postby jcsistemas2001 » Wed Sep 24, 2014 12:32 pm

tdefeo23,
I curious about this post and results. Any news about this?
In your testing you included power measurement (of the board)?
Regards
Julio
jcsistemas2001
 
Posts: 25
Joined: Thu Oct 17, 2013 5:53 am

Re: DDR3 Corruption on Wandboard

Postby wandboard » Fri Sep 26, 2014 10:21 pm

In the past months some minor patches have been submitted on the wandboard gitserver, as well as the links on the wandboard download page for the sourcecode and the demo trial images have been kept updated.

We didn't hear of this issue for a long time.
wandboard
Site Admin
 
Posts: 24
Joined: Thu Aug 29, 2013 4:35 pm

Re: DDR3 Corruption on Wandboard

Postby Tapani » Tue Sep 30, 2014 3:03 am

This issue was about memory timings in our Android 4.2 release.

The old timings worked perfectly on the actual frequencies used (400 and 533 MHz), but the OP did stress testing on other frequencies than that. In one or two frquencies they got rare one bit errors after days of stressing.

Somewhere around the release of Android 4.3 we tweaked the memory timings. The resulting timings now work with even with the one or two frequencies were the rare errors occured.
The new timings are now standard in our software, and are present in for instance the Android 4.4 release.
Tapani
 
Posts: 712
Joined: Tue Aug 27, 2013 8:32 am

Re: DDR3 Corruption on Wandboard

Postby tdefeo23 » Fri Oct 03, 2014 3:36 am

Just to follow up, unfortunately, due to our product schedule, we couldn't afford to spend any more time on this issue. In the end, we switched the design to use an Aaeon IMX6 Q7 module, which has worked well. So, I never got a chance to try any new memory timings, or Android 4.4. Hopefully the timing changes fixed it for you. Thanks again,
Tony
tdefeo23
 
Posts: 13
Joined: Mon Jan 20, 2014 10:34 pm

Re: DDR3 Corruption on Wandboard

Postby mindentropy » Sun May 21, 2017 7:21 am

I too started facing the exact same issue. Most of the time the UI crashes and then zygote, netd etc crashes. Initially my Android and the wandboard Dual did not have any crashes at all. After at least 5 months I see this problem.

I too started DDR stress testing using u-boot option in the DDR stress tool and I could get errors at 400MHz DDR clock frequency and 800MHz CPU in the the 60th iterations. If I test it at 1GHz I get a failure almost immediately.

After doing the stress test my board does not even boot-up the Linux kernel. If it boots forward from the kernel I get undefined instructions etc. Does running the stress test cause it to erase some calibration data stored somewhere?
mindentropy
 
Posts: 21
Joined: Sun Aug 31, 2014 9:32 am

Re: DDR3 Corruption on Wandboard

Postby Snew952 » Thu Jan 11, 2018 6:19 am

You could quickly download and install and also install this apk on your smartphone. machineliker after that do not bear in mind to include this wonderful.
Snew952
 
Posts: 2
Joined: Thu Jan 11, 2018 6:19 am


Previous

Return to Hardware and peripherals

Who is online

Users browsing this forum: No registered users and 14 guests