[Discuss-gnuradio] sparc <--> x86 data exchange

[cswiger]

Gang - any hints on what happens to float data when transfered
from a sparc machine to an x86?  I understand byte-swap but this
looks like word-swap.

A dial_tone_2_file.py script is run on a sparc
machine creating dial_tone_out_sparc.
The same script is run on an x86 creating dial_tone_out_x86.


od of dial_tone_out_sparc on sparc:

0000000 3194 d46d 3d0a 3b9b 3d8a 015a 3dce 70b4
0000020 3e09 192a 3e2a 8692 3e4b 6464 3e6b 9727



hexdump of dial_tone_out_x86 on x86:

0000000 d46d 3194 3b9b 3d0a 015a 3d8a 70b4 3dce
0000010 192a 3e09 8692 3e2a 6464 3e4b 9727 3e6b

where you can clearly see every PAIR of bytes are swapped.
3194 d46d -->  d46d 3194



Copying the sparc file to the x86 box,
hexdump of dial_tone_out_sparc on x86:

0000000 9431 6dd4 0a3d 9b3b 8a3d 5a01 ce3d b470
0000010 093e 2a19 2a3e 9286 4b3e 6464 6b3e 2797

where you can clearly see a BYTE-swap has occurred on the file.

running $ dd conv=swab if=dial_tone_out_sparc of=dial_tone_out_sparc_swab

hexdump of dial_tone_out_sparc_swab on x86:

0000000 3194 d46d 3d0a 3b9b 3d8a 015a 3dce 70b4
0000010 3e09 192a 3e2a 8692 3e4b 6464 3e6b 9727

Now, it looks the same as the octal dump on the sparc, but
16bit words are still reversed compared to the x86 hexdump of
dial_tone_out. Is this something to do with the ultrasparc II
being a 64-bit architecture?

Creating a dial-tone on the sparc machine and sending it over a
network to audio on the x86 machine comes out noise.

TIA
--Chuck