FreeCalypso > hg > freecalypso-tools
view loadtools/scripts/dsample.config @ 407:19e5a3e2f9c0
fcup-settime: moved time() retrieval a little closer to the output
A fundamental problem with all simple time transfer tools is that there is
always some delay between the time retrieval on the source system and that
transmitted time being set on the destination, and the resulting time
on the destination system is off by that delay amount. This delay cannot
be fully eliminated when working in a simple environment like ours,
but we should make our best effort to minimize it. In the present case,
moving the atinterf_init() call before the time() retrieval should make
a teensy-tiny improvement.
author | Mychaela Falconia <falcon@freecalypso.org> |
---|---|
date | Sat, 11 Aug 2018 21:52:17 +0000 |
parents | 2b5ed962c2f9 |
children | 49ee210fc4fb |
line wrap: on
line source
# The following parameters go into the <p command sent to the boot ROM # The values to be used have been gleaned from the 20020917 fw image # CLKTCXO input is 13 MHz on the D-Sample, and with Calypso C05 # the max allowed PLL'ed clock is 78 MHz for the DSP and 39 MHz for the ARM. # TI's firmware sets the PLL up to multiply by 6 (giving 78 MHz) with # divide by 2 for the ARM, but the boot ROM doesn't do the latter when # the input clock is 13 MHz. Hence we'll program the PLL to multiply # by 3, putting everything at 39 MHz. pll-config 3/1 rhea-cntl 0x00 # set by 20020917 fw, hence presumed correct # The remaining settings are carried out via loadagent commands init-script cs2-4ws-8mb.init # 8 MiB flash, accessible at 0x03000000 without Compal-like problems, # but the 28F640W30B flash chip has partition quirks, so we need to # tell fc-loadtool about it explicitly instead of using CFI. flash 28f640w30b 0x03000000 # Perform a Iota poweroff when we are done exit-mode iota-off