Am 17.10.2017 um 16:46 schrieb H. Nikolaus Schaller hns@goldelico.com:
Am 17.10.2017 um 15:56 schrieb H. Nikolaus Schaller hns@goldelico.com:
Am 17.10.2017 um 15:52 schrieb Michael Mrozek EvilDragon@openpandora.org:
Am Dienstag, den 17.10.2017, 15:07 +0200 schrieb H. Nikolaus Schaller:
Hi,
Do you still have multiple PCBs with Samsung 4GB RAM lying around?
I think I have one but it doesn't boot as good as both IMG boards.
If you add the wire there, does that fix the problems?
Those are not reworked, so they should be fine.
Good additional test!
Summary so far:
- C48: Samsung 4GB makes problems (doesn't boot well)
- C41: IMG 4GB works perfectly
here is another unplanned test result from this good board:
root@letux:~# time memtester 3GB memtester version 4.3.0 (32-bit) Copyright (C) 2001-2012 Charles Cazabon. Licensed under the GNU General Public License version 2 (only).
pagesize is 4096 pagesizemask is 0xfffff000 want 3072MB (3221225472 bytes) got 2924MB (3067002880 bytes), trying mlock ...locked. Loop 1: Stuck Address : ok Random Value : ok Compare XOR : ok Compare SUB : ok Compare MUL : ok Compare DIV : ok Compare OR : ok Compare AND : ok Sequential Increment: ok Solid Bits : ok Block Sequential : ok Checkerboard : ok Bit Spread : ok Bit Flip : ok Walking Ones : ok Walking Zeroes : ok 8-bit Writes : -
then it stopped.
But because battery went empty (3.032V)! So this 4GB CPU board runs even stable at low battery.
- C35: works better with higher VDD - cables have almost no effect
three more test results with this board:
* memtester only showed 2 bit-flips (again the same bit!) at 1.45V
Bit Spread : ok Bit Flip : testing 93FAILURE: 0xffffffff != 0xfffff7ff at offset 0x337dfa44. FAILURE: 0xffffffff != 0xfffff7ff at offset 0x53679a44. Walking Ones : ok
* a second memtester loop did find 5 similar FAILURES
* and it did run one memtester loop without failures at 1.5V Now this battery is also drained to 3.08V).
So we have a clearly DDR3L voltage dependent board. Which is the same effect as some Samsung 4GB boards show. This which makes me wonder a little if OMAP5 EMIF supply also plays a role. Anyways my plan is already to look there intensively when doing the PCB layout improvements.
We would need more IM8G boards to get better statistics to show if it is just this unit or a general effect.
Maybe your proposal of adding wires to the Samsung RAM board will give new insights. But now I have to recharge the batteries (Pyra and mine)...
BR, Nikolaus