ManuFerHi wrote:I don't think it's the frequency, I've tried to raise from 128.92 to 133mhz step by step and it's always the same result. In addition the cores before to TEST4B work perfectly and go at the same speed and phase, but there was an important change between these two versions in memory management.
Don't touch the frequency, it must be in sync with the main clock (it's exactly 6 times). If not the phase, maybe try to change the input delays in the .sdc (the numbers should be a bit less according to the datasheet). And it might be not the SDRAM chip at all, e.g. if you have some old boards in stock, you can try to resolder one Micron to an Alliance and see if it breaks or not.
Just to emphasize: try change c0 phase shift (from -3000 ps) to something else (-1500 - -3000). Don't touch the frequency!