[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: [linrad] FFT performance
Hi Roger,
I can't get to the screen where the values are displayed. This is what is
happening with details:
$ ./linrad
ADC sampling is set to 8000, 2 channels (I/Q receiver). Parameters are
saved (W pressed).
At the main screen I press "F" for the hardware test mode. The message
"You are prompted to the parameter selection screens for the following
reason: par_test file missing". Then at the "Set parameters for mode:
HARDWARE TEST MODE RX CHannels: 1" screen, no matter what values I set,
when I press enter, linrad always tells that "Speed far too slow. Try less
demanding FFT!". That is it. The only way to get out is to press ESC, thus
quiting the application.
So, I'm unable to post the performance values since I cannot get them.
73,
-- Edson, 7n4ncl
On Thu, 14 Nov 2002, W3SZ wrote:
> Hi, Edson!
>
> How about posting a listing of ALL of your parameters to the list?
>
> What is your sampling rate? Did you check out Leif's page on hardware
> performance?
>
> Its at
>
> http://ham.te.hik.se/homepage/sm5bsz/linuxdsp/fft1time/fft1time.htm
>
> This is what Leif got with a 650 MHz Pentium III:
>
> SM 5 BSZ - First FFT timing for a large system using Pentium III
> (Coppermine) at 650MHz
> (Dec 30 2000)
> ------------------------------------------------------------------------
> --------
> Sampling rate = 96kHz.
> No of A/D channels = 4.
> No of RF channels = 2.
> Useful bandwidth 2 x 90 kHz.
>
>
>
> SIZE = 32768 16 bit no window, bandwidth 1.46Hz
> Vers Name Time
> 0 Radix 2 DIF C 37.15%
> 1 Radix 2 DIF ASM 34.95%
> 2 Twin Radix 2 ASM 41.75%
> 3 Radix 4 DIT C 33.92%
> 4 Twin radix 4 DIT C 33.10%
> 5 Twin radix 4 DIT SIMD 26.10%
>
> Typical operation
> SIZE = 8192 32 bit sin squared window, bandwidth 23Hz
> Vers Name Time
> 0 Radix 2 DIF C 24.01%
> 1 Radix 2 DIF ASM 22.06%
> 2 Twin Radix 2 ASM 20.17%
> 3 Radix 4 DIT C 22.87%
> 4 Twin radix 4 DIT C 19.70%
> 5 Twin radix 4 DIT SIMD 14.70%
>
> -------------------------------------
>
>
> If we can all see all of your parameters, maybe we can help.
>
> 73,
>
> Roger Rehr
> W3SZ FN20ah
> 2 Merrymount Road
> Reading, PA 19609-1718
> http://www.qsl.net/w3sz
>
>
>
> > -----Original Message-----
> > From: owner-linrad@xxxxxxxxxxxxxxxxxxxxx
> > [mailto:owner-linrad@xxxxxxxxxxxxxxxxxxxxx] On Behalf Of Edson Pereira
> > Sent: Thursday, November 14, 2002 10:00 AM
> > To: linrad@xxxxxxxxxxxxxxxxxxxxx
> > Subject: [linrad] FFT performance
> >
> >
> >
> > Greetings!
> >
> > I'm trying to run linrad on a new system and during the
> > initial calibration linrad keeps telling me that the system
> > is not fast enough and asks me to try to use less demanding
> > FFTs. No matter how I change the values for the FFT
> > paramenters (CW bandwidth for an I/Q direct conversion
> > receiver), the result is always the same.
> >
> > The system is a PIII 1.2 GHz with 512 MB of RAM and a
> > soundcard with a
> > C-Media CMI9738 chipset. The soundcard works fine with other
> > applications.
> > The kernel module is the OSS free cmpci.o.
> >
> > Any suggestions?
> >
> > 73,
> >
> > -- Edson, 7n4ncl
> >
> >
>
>
Hi Roger,
I can't get to the screen where the values are displayed. This is what is
happening with details:
$ ./linrad
ADC sampling is set to 8000, 2 channels (I/Q receiver). Parameters are
saved (W pressed).
At the main screen I press "F" for the hardware test mode. The message
"You are prompted to the parameter selection screens for the following
reason: par_test file missing". Then at the "Set parameters for mode:
HARDWARE TEST MODE RX CHannels: 1" screen, no matter what values I set,
when I press enter, linrad always tells that "Speed far too slow. Try less
demanding FFT!". That is it. The only way to get out is to press ESC, thus
quiting the application.
So, I'm unable to post the performance values since I cannot get them.
73,
-- Edson, 7n4ncl
On Thu, 14 Nov 2002, W3SZ wrote:
> Hi, Edson!
>
> How about posting a listing of ALL of your parameters to the list?
>
> What is your sampling rate? Did you check out Leif's page on hardware
> performance?
>
> Its at
>
> http://ham.te.hik.se/homepage/sm5bsz/linuxdsp/fft1time/fft1time.htm
>
> This is what Leif got with a 650 MHz Pentium III:
>
> SM 5 BSZ - First FFT timing for a large system using Pentium III
> (Coppermine) at 650MHz
> (Dec 30 2000)
> ------------------------------------------------------------------------
> --------
> Sampling rate = 96kHz.
> No of A/D channels = 4.
> No of RF channels = 2.
> Useful bandwidth 2 x 90 kHz.
>
>
>
> SIZE = 32768 16 bit no window, bandwidth 1.46Hz
> Vers Name Time
> 0 Radix 2 DIF C 37.15%
> 1 Radix 2 DIF ASM 34.95%
> 2 Twin Radix 2 ASM 41.75%
> 3 Radix 4 DIT C 33.92%
> 4 Twin radix 4 DIT C 33.10%
> 5 Twin radix 4 DIT SIMD 26.10%
>
> Typical operation
> SIZE = 8192 32 bit sin squared window, bandwidth 23Hz
> Vers Name Time
> 0 Radix 2 DIF C 24.01%
> 1 Radix 2 DIF ASM 22.06%
> 2 Twin Radix 2 ASM 20.17%
> 3 Radix 4 DIT C 22.87%
> 4 Twin radix 4 DIT C 19.70%
> 5 Twin radix 4 DIT SIMD 14.70%
>
> -------------------------------------
>
>
> If we can all see all of your parameters, maybe we can help.
>
> 73,
>
> Roger Rehr
> W3SZ FN20ah
> 2 Merrymount Road
> Reading, PA 19609-1718
> http://www.qsl.net/w3sz
>
>
>
> > -----Original Message-----
> > From: owner-linrad@xxxxxxxxxxxxxxxxxxxxx
> > [mailto:owner-linrad@xxxxxxxxxxxxxxxxxxxxx] On Behalf Of Edson Pereira
> > Sent: Thursday, November 14, 2002 10:00 AM
> > To: linrad@xxxxxxxxxxxxxxxxxxxxx
> > Subject: [linrad] FFT performance
> >
> >
> >
> > Greetings!
> >
> > I'm trying to run linrad on a new system and during the
> > initial calibration linrad keeps telling me that the system
> > is not fast enough and asks me to try to use less demanding
> > FFTs. No matter how I change the values for the FFT
> > paramenters (CW bandwidth for an I/Q direct conversion
> > receiver), the result is always the same.
> >
> > The system is a PIII 1.2 GHz with 512 MB of RAM and a
> > soundcard with a
> > C-Media CMI9738 chipset. The soundcard works fine with other
> > applications.
> > The kernel module is the OSS free cmpci.o.
> >
> > Any suggestions?
> >
> > 73,
> >
> > -- Edson, 7n4ncl
> >
> >
>
>
LINRADDARNIL