[llvm-commits] CVS: llvm-test/MultiSource/Benchmarks/mediabench/adpcm/LICENSE.TXT Makefile README

Dinakar Dhurjati dhurjati at cs.uiuc.edu
Tue Jan 11 10:05:56 PST 2005



Changes in directory llvm-test/MultiSource/Benchmarks/mediabench/adpcm:

LICENSE.TXT added (r1.1)
Makefile added (r1.1)
README added (r1.1)
---
Log message:


Added media bench to llvm-test


---
Diffs of the changes:  (+107 -0)

Index: llvm-test/MultiSource/Benchmarks/mediabench/adpcm/LICENSE.TXT
diff -c /dev/null llvm-test/MultiSource/Benchmarks/mediabench/adpcm/LICENSE.TXT:1.1
*** /dev/null	Tue Jan 11 12:05:55 2005
--- llvm-test/MultiSource/Benchmarks/mediabench/adpcm/LICENSE.TXT	Tue Jan 11 12:04:14 2005
***************
*** 0 ****
--- 1,9 ----
+ adpcm - Part of the mediabench Benchmark Suite
+ -------------------------------------------------------------------------------
+ All files are licensed under the LLVM license with the following additions:
+ 
+ Please see individiual files for additional copyright information.
+ Redistribution must follow the additional restrictions required by the
+ individual files.
+ 
+ 


Index: llvm-test/MultiSource/Benchmarks/mediabench/adpcm/Makefile
diff -c /dev/null llvm-test/MultiSource/Benchmarks/mediabench/adpcm/Makefile:1.1
*** /dev/null	Tue Jan 11 12:05:56 2005
--- llvm-test/MultiSource/Benchmarks/mediabench/adpcm/Makefile	Tue Jan 11 12:04:14 2005
***************
*** 0 ****
--- 1,6 ----
+ # MultiSource/mediabench/adpcm Makefile: Build all subdirectories automatically
+ 
+ LEVEL = ../../../..
+ PARALLEL_DIRS  := rawcaudio rawdaudio
+ 
+ include $(LEVEL)/Makefile.programs


Index: llvm-test/MultiSource/Benchmarks/mediabench/adpcm/README
diff -c /dev/null llvm-test/MultiSource/Benchmarks/mediabench/adpcm/README:1.1
*** /dev/null	Tue Jan 11 12:05:56 2005
--- llvm-test/MultiSource/Benchmarks/mediabench/adpcm/README	Tue Jan 11 12:04:14 2005
***************
*** 0 ****
--- 1,92 ----
+ This directory contains version 1.2 of a simple 16 bit PCM <-> 4 bit
+ ADPCM coder and decoder. See the source and the manpage for comments
+ on the algorithm.
+ 
+ If you have previously used version 1.0: there was a bug in it. This
+ version fixes the bug. (1.1 fixed that, but had another slight bug
+ that caused loss of quality on SGI machines).
+ 
+ The routines have been tested on an SGI Indigo running Irix 4 and
+ 5 and on a Sun Sparc SLC running SunOS 4.1.1. There are no real
+ machine dependencies, though, so there should be no problems on other
+ architectures. 
+ 
+ There is a timing program to test how fast the stuff runs on your
+ machine. Here are some sample numbers:
+ 
+ R4000 Indigo:	compress: 1.1Msample/sec, decompress: 1.7Msample/sec
+ R3000 Indigo:	compress: 410Ksample/sec, decompress: 850Ksample/sec
+ Sun SLC:	compress: 250Ksample/sec, decompress: 420Ksample/sec
+ Mac-IIsi	compress   21Ksample/sec, decompress:  35Ksample/sec
+ 486/DX2-33 SCO: compress  550Ksample/sec, decompress: 865Ksample/sec
+ 486/33 linux:	compress  278Ksample/sec, decompress: 464Ksample/sec
+ 386/33 gcc:	compress  117Ksample/sec, decompress: 168Ksample/sec
+ (the Borland and MS compilers are a little slower)
+ 
+ Also included are some simple test programs that convert audio files
+ to/from 4bit ADPCM samples. Rawcaudio/rawdaudio convert to/from raw
+ 16bit PCM files; sgicaudio/sgidaudio convert to/from SGI AIFF files
+ and suncaudio/sundaudio convert to/from SUN/NeXT format audio files.
+ The Sun program was donated by Hughes Doug, hughes at sde.mdso.vf.ge.com
+ (it is actually only one program and looks at the program name to
+ decide whether to compress or decompress). Oh yeah, the sun and sgi
+ compressors will probably only compile on those machines. Also, for
+ the sgi compressor you need the Digital Media Developers Option.
+ 
+ Note: I have had to make a few changes to the sun audio converter, but
+ unfortunately I have no way of testing the result (since it needs
+ Solaris 1.somethingorother which I don't have). If the program does
+ not work and someone fixes it I would like to get a copy.
+ 
+ Finally, there is a uuencoded adpcm sound file of me saying "hello
+ world" (with a cold) that you can use to test that everything works
+ correctly.
+ 
+ If you use this package I would like to hear from you.
+ I am especially interested in people who can test interoperability
+ with proven Intel/DVI ADPCM coders. I have tried to get the algorithm
+ correct, but you never know....
+ 
+ Answers to often-asked questions:
+ - No, this is *not* a G.721 coder/decoder. The algorithm used by G.721
+   is very complicated, requiring oodles of floating-point ops per
+   sample (resulting in very poor performance). I have not done any
+   tests myself but various people have assured my that 721 quality is
+   actually lower than DVI quality.
+ 
+ - No, it probably isn't a RIFF ADPCM decoder either. Trying to decode
+   RIFF ADPCM with these routines seems to result in something
+   recognizable but very distorted.
+ 
+ - No, it is not a CDROM-XA coder either, as far as I know. I haven't
+   come across a good description of XA yet.
+ 
+ - No, testc and testd are not full-blown audio-file converters, only
+   simple test programs. If you write a file-converter (like caudio.c,
+   which compresses Sun audio files) and you're willing to donate it
+   I'd be very happy to include it in the distribution.
+ 
+ - No, I know of no standard audio file format that supports idvi
+   adpcm. If you find one, please let me know.
+ 
+ - Also, I know next-to-nothing about who IMA are. Their address, though,
+   from their May 1992 proceedings (where I got the algorithm from):
+ 	IMA Compatability Project Headquarters
+ 	9 Randall Court
+ 	Annapolis MD 21401
+ 	USA
+ 
+ 	Phone: (410) 626-1380
+ 	Fax:   (410) 263-0590
+ 
+ They might be able to send back issues of their proceedings, I don't
+ know.
+ 
+ 	Jack Jansen
+ 	Centre for Mathematics and Computer Science
+ 	Kruislaan 413
+ 	Amsterdam
+ 	the Netherlands
+ 
+ 	+31 20 592 4098
+ 	Jack.Jansen at cwi.nl






More information about the llvm-commits mailing list