I have a new machine, based on a Pentium D 940 at 3.2Ghz with 1GB of 533Mhz DDR2. I installed the x86_64 FC6 distribution on it, and then Gnu Radio and its dependencies. I ran into small problems with my gr-radio-astronomy applications, that were easy to fix. It turns out that Numeric.zeros() is now fussy about type correctness in the "number of elements" argument. Also, in two places, I inadvertently used a "C" idiom: d *= 0.1 Which caused no problem previously, but caused Python to barf in the new FC6 environment. I replaced it with: d = d * 0.1 The new machine is working out quite well. I can process a full 8Mhz of bandwidth, doing both spectral and Continuum modes. I was able to double the FFT length, as well as double the FFT rate, which should make my spectral analysis more sensitive.
on 2007-03-15 18:28
on 2007-03-15 19:51
On Thu, Mar 15, 2007 at 01:26:31PM -0400, Marcus Leech wrote: > d *= 0.1 > more sensitive. Thanks for the report! Eric