Bob,

With the changes in Preview 8 the good RX of the SDR-1000 is
now GREAT! My system is: Pentium 4, 3.2 GHz, 1024 MB DDR, XP
Pro SP2, Radeon 9550 256 MB, w/RFE, Waveterminal 192X, WDM-KS,
USB-to-PIO.

The CPU loadings are:
- Display off 3.9%
- Spectrum 5 FPS 4.7%
- Spectrum 15 FPS 6.3%

Well done! Thanks for this Christmas present,
73, Ahti OH2RZ


----- Original Message ----- From: "Robert McGwier" <[EMAIL PROTECTED]>
To: "KD5NWA" <[EMAIL PROTECTED]>
Cc: <FlexRadio@flex-radio.biz>
Sent: Saturday, December 17, 2005 3:12 PM
Subject: Re: [Flexradio] Beta 8, DirectX, new AGC, use of FFTW3


Let me say this once since it has not been made explicit.

Whether or not DirectX helps you or hurts you is COMPLETELY dependent on the quality of your video card and its driver. If you driver and video
card do lots and lots of texturing and rendering, etc. in hardware, it
will give you a huge cut down in CPU.  If you own a POS video card, to
match the POS mobo for the sound card, expect no help from this
setting.  Please continue to use GDI.

(POS == piece of slime, otherwise substitute your favorite)

On my laptop,  the CPU %   moves up and down from 1% to 10% with much
more variance than before.  However,  previous to our use of FFTW3 and
DirectX,  it was a steady 40%.  This latest version is an overall
MASSIVE win.

The agc is now the FULL two track specification that Phil proposed.
The implementation has undergone many iterations and the fast track came
together in the last two weeks while sitting in my hotel room in
Marburg, Germany. Frank and I attended an AMSAT meeting there where it
became absolutely clear that SDR is the future of amateur radio
satellites. Phase 3E will use an SDX (software defined transponder) and AMSAT's current Eagle proposal is a software defined satellite. Now if
we could only get Software defined launches!

Back to the AGC, Phil lives in the land down under. This means that he
and I are inverted!

Phil suggests setting the fast attack threshold to twice that of slow
track to target the same signal.

Phil used AGC voltage (the more negative, the more AGC limiting).   I
use gain. Furthermore, my gain is voltage multiplier. His AGC was dB and inverse gain. The results were like those of the JPL engineers who
spoke metric on one side and feet/yards on the other and sent the Mars
mission into the dirt.   This is entirely my fault and Phil's spec has
been fine since day one.  I really like the new sound this gives.
Thanks so much to Jeff for his constant probing of the agc and his
EXTREMELY useful software analysis and critical remarks.  Also to Dale
for his continued testing and remarks and measurements.  Four days ago
they gave me their latest comments and measurements on overshoot and I
hope we successfully captured the necessary changes in release 8.  If
not, we are still learning. What I can say is this is a sweet sounding
agc now and we should all send our thanks to Phil, VK6APH.

Eric is busting his rear end on getting DirectX to work.  Like all
things, Microsoft works extremely hard to obfuscate as much as
possible. I am becoming more and more convinced that you get the secret keys to the kingdom of their documentation, and probably code examples,
only as a result of attending this huge thousands of dollars developer
courses. Monopolies stink. There is absolutely NOTHING worse than the
documentation for WDM-KS.  DirectX is slightly improved with the
"managed directX" is an abomination.  It silently wrecked the floating
point coprocessor and we were pointing the finger at FFTW3 until Eric
found a page or two full of complaints about the idiotic silent behavior
and then implemented the (behind the hidden handshake and thousands of
dollars of training) hidden flag that stops this behavior.

Bob
N4HY



KD5NWA wrote:

I just tried Beta 8 on my SR-40 and turned on Direct X, maybe I'm
missing something here but my panadapter works just fine, everything
is working fine.

After turning the panadapter on I doubled checked and the setting is
still on Direct X, I don't see a difference  when turning it on or
off as far as CPU usage.


Cecil Bayona
KD5NWA


Reply via email to