[U-Boot-Users] SDRAM "anti-problem"
I have the "problem" that i can't corrupt my SDRAM. To verify my SDRAM refresh settings are reasonable, I did my best in reprogramming my MPC865's UPM to not refresh the SDRAM often enough.
I set the timer loop field (TLFA) to minimum, the PTA to maximum and the PTP to maximum to have the refresh come too seldom to keep the data.
Whatever, the SDRAM is very unimpressed and keeps the data I wrote into it without a single bit corrupted for at least minutes.
Another problem is that the mpcbdm I use to test the SDRAM takes up to 30 secs to read/write about 60K of data into/out of RAM (burst mode disabled). Is that a mpcbdm issue? I can't believe the SDRAM can be so slow.
Peter Asemann
On Wednesday 25 May 2005 10:37, Peter Asemann wrote:
Another problem is that the mpcbdm I use to test the SDRAM takes up to 30 secs to read/write about 60K of data into/out of RAM (burst mode disabled). Is that a mpcbdm issue? I can't believe the SDRAM can be so slow.
In my experience the BDM connection is much, much slower than the RAM.
If I load a kernel into the RAM with the BDM I have to get up and walk away, get a sandwitch, have a soda, etc... But If I transfer it in over TFTP it's nearly instantaneous.
NZG.
In message 200505251309.50618.ngustavson@emacinc.com you wrote:
In my experience the BDM connection is much, much slower than the RAM.
This is true.
If I load a kernel into the RAM with the BDM I have to get up and walk away, get a sandwitch, have a soda, etc...
Ummm... on most 8xx systems the BDI2000 needs about 7 seconds to load and burn 1 MB of image to flash. You must be really *fast* to fetch a sandwich in this time --- or you are using the wrong tools.
But If I transfer it in over TFTP it's nearly instantaneous.
Of course, this is faster, but then, on some flash chips the flah drivers in U-Boot are far from optimal.
Best regards,
Wolfgang Denk
participants (3)
-
NZG -
Peter Asemann -
Wolfgang Denk