Used that tool already successful with 2GHz LPDDR4 (4000MTS). Now we want to switch to double sized RAM and different vendor. All provide 32bit output, so only thing to change in RPA is Density and Number of rows.
I could get calibration running, but stress tool fails. It is running only with 1.5GHz. For other frequencies I calculated the proper register values as:
1600: 0x000c8031
1700: 0x001a9061
1800: 0x0012c080
Most boards work with 1.6GHz, some even with 1.7, but none with 1.8 or 2.0 GHz.
As mentioned same pcb is working with smaller RAM at 2GHz. So board itself seems to be ok. Tested two different types of LPDDR4, one is specified for only 1.8 the other even for 2.4GHz - but results are similar. As you see it is not always same pattern that fails. What are my options?
Can I reduce the ARM clock to 1.6GHz as we used it normally? The setting in 3rd line of tool doesn't seem to have an effect?!?
I read about modify driver strength (for pins) and impedance. But how to change? In which direction, in which steps? What range might work? Changing single impedance or all 3 at once? Do you have a procedure for doing that? That are to many parameters to play without knowing what to do.
Once calibrated: can I reload the calibration data (as they might differ from run to run)? I can understand some slight changes when run calibration multiple times, but I even observe in lpddr_timinc.c that different registers (!!!) are used, as for instance one run was{ 0x211143, 0x5a1 } while the other one gives { 0x11143, 0x5a1 }?!? That confuses me a lot.
Success: DDR Calibration completed!!!
'lpddr4_timing.c' is created!
DDR Stress Test Iteration 1
--------------------------------
--Running DDR test on region 1--
--------------------------------
t0.1: data is addr test
....
t0.2: row hop read test
...
t1: memcpy SSN armv8_x32 test
..Address of test2 failure: 0x0000000090EB6A50
Data was: 0x00000000FFFFFBEF
But pattern was: 0x00000000FFFFFBFF
DDR Stress Test Iteration 1
--------------------------------
--Running DDR test on region 1--
--------------------------------
t0.1: data is addr test
....
t0.2: row hop read test
...
t1: memcpy SSN armv8_x32 test
.Address of test1 failure: 0x0000000091E8C0D0
Data was: 0xFFFFFFEF04000000
But pattern was: 0xFFFFFFFF04000000
DDR Stress Test Iteration 1
--------------------------------
--Running DDR test on region 1--
--------------------------------
Additional question: I found in NXP EVK board a config for even 2.4GHz. Is this supported by the tool?Tried that with my boards that work fine at 2GHz, but even calibration failed. Used this:
2200: 0x00113031
2400: 0x0012c031
