home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky comp.sys.dec:6745 comp.dsp:2948
- Path: sparky!uunet!spool.mu.edu!cass.ma02.bull.com!mips2!bull.bull.fr!clbull.frcl.bull.fr!bull.bull.fr!julienas!ircam!fingerhu
- From: fingerhu@ircam.fr (Michel Fingerhut)
- Newsgroups: comp.sys.dec,comp.dsp
- Subject: Re: Alpha fft performance
- Message-ID: <1993Jan5.134034.22043@ircam.fr>
- Date: 5 Jan 93 13:40:34 GMT
- References: <1992Dec31.164221.27734@aplcen.apl.jhu.edu>
- Organization: IRCAM, Paris (France)
- Lines: 22
- In-Reply-To: <1993Jan4.154245.13258@crl.dec.com>
- To: msp@nadia, phd@nadia, jpj@nadia, ircam-alpha@nadia
- Cc: lghys@nadia
-
- Voici l'extrait d'un article donnant ces informations:
-
- For a 1024 point, complex, single precision FFT (i.e. just fits in the on-chip
- 8K D cache), we have a measured time of 96,000 cycles. For the various Alpha
- systems, this translates to:
-
- 133 MHz clock 722 microseconds
- 150 Mhz 640 microseconds
- 200 MHz 480 microseconds
-
- The algorithm is a radix-4 algorithm, and is basically just a C translation
- of the FORTRAN code in "DFT/FFT and Convolution Algorithms" by Burrus and
- Parks (with a few tweaks, of course). It was compiled with GCC 2.3 and
- the execution time was measured with the Alpha's process cycle counter.
-
- The Digital Extended Math Library (DXML) for Alpha reportedly can do even
- better, but I don't have any current performance numbers. For more info on
- DXML, contact your local Digital sales rep. [Nous devons le recevoir dans
- la pre-release prochaine].
-
- For some more information on signal processing on Alpha, see our technical
- report "Are DSP Chips Obsolete?" [disponible chez nous].
-