Sound Blaster AWE 32/64 HOWTO
by Marcus Brinkmann v1.2, 11 January 1998
This document describes how to install and configure a Sound Blaster 32 (SB AWE 32, SB AWE 64) card from Creative Labs in a Linux System using the AWE Sound Driver Extension written by Takashi Iwai. It also covers some special tools and players for the SB AWE series. Reference system is a Debian GNU/Linux System, but every other Linux Distribution should also work.
- 1.1 Acknowledgments
- 1.2 Revision History
- 1.3 New versions of this document
- 1.4 Feedback
- 1.5 Distribution Policy
- 2.1 Introduction
- 2.2 Some general notes about the SB AWE cards
- 2.3 Some general notes about the Plug and Play cards
- 2.4 Some general notes about loadable kernel modules
- 2.5 Some general notes about the kernel sound driver
- 4.2 Output - The Raw Audio Device
- 4.3 Output - The OPL-2/OPL-3 Synthesis
- 4.4 Output - The Wave Table Synthesis
- 4.5 Mixing
- 4.6 Input - Sampling with the Raw Audio Device
- 4.7 The MIDI Port
This is the Sound Blaster AWE HOWTO. It gives you detailed information about getting the most out of your Sound Blaster 32 or better, including Wave Table synthesis. This document covers all SB cards up from the SB 32.
The Sound Blaster series is well-known in the DOS and Windows community, and a lot of Linux users want to use this sound card under Linux as well. Unfortunately, Creative Labs only provides a sound driver for Windows and DOS, so it is not trivial to install and use a SB card under Linux. This document tries to describe how one can use the features of the SB AWE series under a Linux environment.
This documents contains information I got from the AWE Driver FAQ and the ISA PnP FAQ. See section Sources for author and location of this documents. Thanks to the authors, the SB AWE support is possible.
A lot of essential work was done by Hannu Savolainen, who developed the sound driver that comes with the Linux kernel. Thank you!
Thanks to the SGML Tools package, this HOWTO is available in several formats, all generated from a common source file.
- Version 1.0
- Version 1.1
corrected spelling (thanks Curt!), added version requirement for
isapnp, now available via sunsite and lots of mirrors
- Version 1.2
French version now available!, minor corrections, lots of spelling fixed (thanks to ispell)
The latest version can be found on my Home Page, namely at Sound Blaster AWE HOWTO.
New versions of this document will be uploaded to various anonymous ftp sites that archive such information including ftp://sunsite.unc.edu/pub/Linux/docs/HOWTO/mini.
Hypertext versions of this and other Linux HOWTOs are available on
many World-Wide-Web sites, including http://sunsite.unc.edu/LDP/. Most
Linux CD-ROM distributions include the HOWTOs, often under the
If you make a translation of this document into another language, let me know and I'll include a reference to it here.
i would especially welcome information about the startup procedure (boot scripts etc.) of other famous Linux distributions, e. g. Red Hat or SuSE Linux.
Copyright 1997 Marcus Brinkmann.
This HOWTO is free documentation; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version.
This document is distributed in the hope that it will be useful, but without any warranty; without even the implied warranty of merchantability or fitness for a particular purpose. See the GNU General Public License for more details.
You can obtain a copy of the GNU General Public License by writing to the Free Software Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
This document tries to help you install and use a Sound Blaster AWE 32 or Sound Blaster AWE 64 from Creative Labs in your Linux system. The reference system is a Debian GNU/Linux system on an Intel i586 platform, but it should work with any other Linux distribution as well as on every platform that is supported by the Linux sound driver (differences are mentioned where they appear, if any).
Be sure to read the Linux Sound HOWTO (see section Additional Information) carefully. I consider my document a supplement to the Sound HOWTO, and often you can find more information about things I have left out there.
The SB AWE 32 sound card provides a raw audio device, standard OPL-2/OPL-3 synthesis, a MPU-401 MIDI port and 32 voices EMU 8000 Wave Table synthesis (for an explanation on these and other terms see the Linux Sound HOWTO). One goal of this document is to help you get all these features to work properly.
The SB AWE 64 has the capabilities of the SB AWE 32 and an additional Wave Guide synthesis Creative Labs is especially proud of. The problem for Linux users is, that the additional 32 voices are software generated and output via the raw wave devices. Because Creative Labs sees no market in Linux drivers, a Wave Guide synthesis sound driver is only available for Windows 3.1 and Windows 95.
This means that, from a Linux user's point of view, the SB AWE 32 and SB AWE 64 are almost identical. From now on I will only refer to the SB AWE in general and will only mention differences where they appear (if any).
Most modern cards for the Intel platform are ISA PnP cards, which is an abbreviation for ``Plug and Play''. This means, that the card has to be configured by the operation system, and this has to be done through an initialization routine at boot time. In general, there are at least three possible ways to do this:
- You have a PnP Bios, which means that your Bios is aware of PnP cards and can configure them. If you think you can use all the features of your SB AWE PnP just because you have a PnP Bios, you are out of luck. Even if the Bios claims to support PnP cards it only initializes a subset of the ports and addresses used by your sound card. You will probably be able to play raw wave data, but you won't be able to play MIDI music, for example. For this reason, a PnP Bios is not an option.
- You have an operating system that supports PnP cards. The current stable Linux kernel (2.0.x) is not such an operating system, so we have to wait for future kernel releases, that will support PnP devices.
- You have a special program, started at boot time, that initializes your PnP cards. This is the way we do it.
The most commonly used software to initialize PnP cards under Linux are the ``isapnptools'' (see section Sources and section Getting Started). They provide a predictable way to configure all ISA PnP cards in your machine, not only your sound card.
Some device drivers can be built as modules instead of compiling them into the kernel. You can find more information about modules in the Kernel HOWTO and the Module HOWTO (see section Additional Information).
If you have a PnP card, you must install sound support as a loadable kernel module. This means, that you can't build the sound driver into the kernel, but you will have to build it as a module that can be loaded into the kernel at runtime. This is because the kernel will be installed before your ISA PnP card can be configured, and your sound driver has to be loaded after your ISA PnP card is configured.
The sound kernel module can be loaded manually via
modprobe -a sound or in the
appropriate boot script of your Linux system (in Debian, it is
sufficient to append a single line containing
/etc/modules). Another approach is to launch
kerneld, a daemon that installs and removes kernel
modules as needed.
kerneld may not be the best solution for the
AWE sound driver module, because it takes time to load the module
in the kernel, especially if you want to use Wave Table synthesis
and load big Sound Font banks, which you have to do each time after
inserting the module. Because
kerneld removes unused
modules after one minute by default, it is perhaps better to insert
the sound module manually or at boot time. Note that inserting the
sound module manually or at boot time does prevent
kerneld from removing it when it is idle. By the way:
You can manually insert the sound module and use
kerneld at the same time. The two methods don't
kerneld does not care about the sound
This is especially useful if your mixer settings get disturbed
after removing and reinstalling the module. A solution for this
problem will be given in section Mixing
(there is described how you can start your mixer automatically when
the sound module gets inserted). However, it takes time for
kerneld to load the module, to load the sound font
bank and to start your mixer, and for this and other reasons it is
better to install the sound module at boot time and not let
kerneld remove it.
You can install sound support in the kernel as a built-in or as a loadable module. If you have a PnP card, you have to install sound support as a module, because the PnP card needs to be initialized via the isapnptools before the module gets installed.
As you can imagine, you will have to recompile the kernel. I will give you a few hints about it below. For now, let's talk about the sound support in the kernel source. The kernel ships with the Free (Lite) Version of the OSS (USS) sound driver. The current version of this driver (3.5.4) does not support the SB AWE in full, but the SB 16 part of it. So you can have a raw audio device and OPL-2/OPL-3 synthesis if you use it, but you will not be able to play midi music with Wave Table synthesis.
If you want to use the Wave Table device, you can either buy the commercial sound driver from 4Front Technologies (someone please confirm me that it can do it, please), or patch your kernel with the AWE 32 Sound Driver Extension by Takashi Iwai. The former is beyond the scope of this document, I assume you want to use the latter.
The AWE32 Sound Driver Extension (see section Sources) is published under the GNU copyright license and ships with a number of tools to make use of the EMU 8000 wave synthesis of the SB AWE cards.
awedrv software is already included in newer
kernel source trees (starting from some 2.1.x kernels, but perhaps
you want to upgrade the sources in your kernel tree, they may be
These are the requirements for SB AWE support under Linux:
- a functional Linux system (e.g. the Debian GNU/Linux distribution),
- a SB AWE 32 or compatible card (e.g. SB32, SB32 PnP, SB AWE64, ...),
- the sources of the Linux kernel, including the sound driver
OSS/Free (normally included, check in
- the AWE 32 sound driver extensions.
If you have a PnP card, you will also need:
- the isapnptools software package.
Look in section Sources for information where you can obtain these programs.
If you have a Debian GNU/Linux system, this means that you will
need the packages
awe-drv and perhaps the package
isapnptools. You will want some of the other
awe-* packages although they are not essential. I
strongly recommend the
kernel-package for easy kernel
compiling and installing.
Let's assume you have properly installed your card in a socket on your motherboard and perhaps already tested it under a DOS or Windows environment.
The next step is to initialize the card at boot time if and only if
it is a PnP card. Follow the documentation in the source package to
compile and install
isapnptools (or just use the
Debian binary package), and make a
# pnpdump > /etc/isapnp.conf
as root. This will generate a hopefully proper configuration file
for your PnP cards, but with all devices commented out. Please edit
this file carefully, and compare the values for DMA channels, IO
bases and interrupts with the configuration of the cards under an
DOS or Window environment if possible (with Win95, look at the
resources used by the card under
resources in the
device manager). If you have problems, read the
documentation that ships with the
isapnptools often fails to detect
all three I/O ports of the SB AWE Wave Table device. Please check
WaveTable entry in your
isapnp.conf against the example at the end of this
HOWTO. Adjust the I/O base addresses if necessary.
CAUTION: According to the
isapnp fails to program the number of the
logical device. If you encounter error messages like the following:
Error occurred executing request 'LD 2' on or around line...
then try adding
(VERIFYLD N) at the top of
isapnp.conf. You have to use at least version 1.10 for
this to work. If you can't use version 1.10 or newer, you can also
POKE the logical device numbers directly. Please refer to the
isapnp-faq for more information on this approach. If
it doesn't work for you, please contact me (and the
isapnp people will be interested, too, I would think).
CAUTION: Make sure that the last line is
(WAITFORKEY), that will sometimes be omitted by older
isapnp.conf for only one PnP card (the sound
card) can be found at the end of the document (see section Sample <tt>isapnp.conf</tt>).
If you have a Debian system, no further adjustments are necessary.
isapnp will be started at boot time in
/etc/init.d/boot with this snippet of script code,
that you may want to include in your boot scripts:
# Configure the isa plug and play boards before loading # modules. Need to do this before loading modules to get # a chance of configuring and starting PnP boards before # the drivers mess all this up. # if [ -x /etc/init.d/isapnp ] then /etc/init.d/isapnp start fi
#! /bin/sh # /etc/init.d/isapnp: configure Plug and Play boards test -x /sbin/isapnp || exit 0 /sbin/isapnp /etc/isapnp.conf exit 0
If you have another Linux distribution, you are on your own. I don't know what to do (anybody willing to submit more specific information?). Just make sure that isapnp gets started before the modules will be loaded (see below).
Before recompiling the kernel, you have to apply the AWE Driver
Extension to the sound driver. Even if your kernel source tree
already includes the
awedrv extension (check
/drivers/sound/lowlevel/ for that), you may want to
awedrv software. Follow the installation
instructions in the
awedrv source directory. In brief,
you have to run an installation script that applies the patches to
the kernel sources.
Be careful if you have upgraded the kernel source tree after
running the install script. The script just checks if a certain
file exists - if it exists, it does not apply the necessary patch.
You are well advised to remove the file
drivers/sound/lowlevel/awe_wav.c before running the
script after upgrading the kernel source.
Next you have to configure the kernel for sound support. I hope you
know a bit about kernel compiling; see the Sound HOWTO and the
Kernel HOWTO for details. Go in the source directory of your kernel
/usr/src/linux for example), and do
# make config
make menuconfig or
make xconfig. Then
you have to configure your kernel in the standard way. Use this
opportunity to make a small and powerful kernel, especially
designed for your system. Be sure to answer
module support with
Y, if you want to install
the sound driver as a loadable module (a must if you have
a PnP card), but I'm sure you want to do it anyway.
At one point, you will be asked if you want sound card support. You
are free to answer with
Y or with
you have not a PnP card. You must answer with
M, for module, if you have a PnP card. You have to
compile sound card support as a module if you have a PnP card
because the PnP card has to be initialized before the module gets
The following questions you should answer with
Sound Blaster (SB, SBPro, SB16, clones) support (CONFIG_SB) [Y/n/?] Generic OPL2/OPL3 FM synthesizer support (CONFIG_ADLIB) [Y/n/?] /dev/dsp and /dev/audio support (CONFIG_AUDIO) [Y/n/?] MIDI interface support (CONFIG_MIDI) [Y/n/?] FM synthesizer (YM3812/OPL-3) support (CONFIG_YM3812) [Y/n/?] lowlevel sound driver support [Y/n/?] AWE32 support (CONFIG_AWE32_SYNTH) [Y/n/?]
Only the latter is actually for the Wave Table synthesis. The others are SB 16 options from the OSS/Free sound driver.
In addition, you have to configure the sound card I/O port. Look at
isapnp.conf file for hints, if you have one. For
me, the following default values are sufficient. Note that the
default value from the kernel configuration script may be wrong
(especially the values
SB_MPU_BASE seem to be incorrect for most cards).
I/O base for SB Check from manual of the card (SBC_BASE)  Sound Blaster IRQ Check from manual of the card (SBC_IRQ)  Sound Blaster DMA 0, 1 or 3 (SBC_DMA)  Sound Blaster 16 bit DMA (_REQUIRED_for SB16, Jazz16, SMW) 5, 6 or 7 (use 1 for 8 bit cards) (SB_DMA2)  MPU401 I/O base of SB16, Jazz16 and ES1688 Check from manual of the card (SB_MPU_BASE)  SB MPU401 IRQ (Jazz16, SM Wave and ES1688) Use -1 with SB16 (SB_MPU_IRQ) [-1]
Now recompile the kernel. Debian users should use the
kernel-package. This package makes the kernel compile
as easy as installing a debian package. Look at the documentation
/usr/doc/kernel-package/. Here is a hint:
# make-kpkg clean # make-kpkg -revision custom.1.0 kernel_image
If you have another Linux distribution, follow the standard way for
compiling a new kernel. Don't forget
make modules and
make modules_install. Look at the Sound HOWTO and
perhaps the Kernel HOWTO for more information.
After installing the new kernel, you should do a reboot (be sure to have a functional boot disk at hand). Cross you fingers.
If you have a PnP card be sure to launch isapnp either in a boot script (as described above) or manually:
# /sbin/isapnp /etc/isapnp.conf Board 1 has Identity 74 00 00 e3 10 48 00 8c 0e: CTL0048 Serial No 58128 [checksum 74]
Now you can install the sound driver, if you have compiled it as a module:
# modprobe -a sound AWE32 Sound Driver v0.3.3e (DRAM 2048k)
If you think the memory detection was not correct (I have a report
of one who has a AWE64 with 4096k, and ``detected'' have been
28672k), try either to upgrade the
awedrv software or
to specify the amount of memory in the file
#define AWE_DEFAULT_MEM_SIZE 4096 /* kbytes */
Sorry, you have to recompile the kernel then (perhaps compiling the modules will be sufficient, but I don't know for sure).
If it works, you may want to have the sound module loaded
automatically. You can use
kerneld (why this is a bad
idea is explained in section 1.4) or append a single line
sound to your
Debian) or add
/sbin/modprobe -a sound to your
If you have built-in sound support, you will get some useful
information at boot time. If you have sound support installed as a
loadable module, you can get the same information (perhaps after
removing the sound module with
modprobe -r sound
# modprobe -a sound trace_init=1 Sound initialization started <Sound Blaster 16 (4.13)> at 0x220 irq 5 dma 1,5 <Sound Blaster 16> at 0x330 irq 5 dma 0 <Yamaha OPL3 FM> at 0x388 Sound initialization complete AWE32 Sound Driver v0.3.3e (DRAM 2048k)
If you have a
/proc virtual file system, you can look
for the sound device with
# cat /proc/devices Character devices: [...] 14 sound [...]
Next make sure you have the correct devices installed under
/dev/. Please look at the Sound HOWTO for details.
/dev/sndstat about the status of the sound
# cat /dev/sndstat Sound Driver:3.5.4-960630 (Sat Oct 11 19:35:14 CEST 1997 root, Linux flora 2.0.29 #1 Sat Oct 11 19:12:56 CEST 1997 i586 unknown) Kernel: Linux flora 2.0.29 #1 Sat Oct 11 19:36:23 CEST 1997 i586 Config options: 0 Installed drivers: Type 1: OPL-2/OPL-3 FM Type 2: Sound Blaster Type 7: SB MPU-401 Card config: Sound Blaster at 0x220 irq 5 drq 1,5 SB MPU-401 at 0x330 irq 5 drq 0 OPL-2/OPL-3 FM at 0x388 drq 0 Audio devices: 0: Sound Blaster 16 (4.13) Synth devices: 0: Yamaha OPL-3 1: AWE32 Driver v0.3.3e (DRAM 2048k) Midi devices: 0: Sound Blaster 16 Timers: 0: System clock Mixers: 0: Sound Blaster 1: AWE32 Equalizer
If you don't have an output like this, perhaps there is an error in your configuration. Go on and see what doesn't work, then go back to step Getting Started, checking everything.
Try to get an
.au file (Sun workstation) or a raw
sample file, and do
# cat bell.au > /dev/audio
# cat sample > /dev/dsp
You should hear the content of the file via the
0: Sound Blaster 16 (4.3).
If you want to use the OPL-2/OPL-3 FM synthesis to play MIDI-Files
with your sound card, try the program
Appendix B). Start it with
# playmidi -f dance.mid
# playmidi -4 dance.mid
The former will give you OPL-2, the latter OPL-3 MIDI music. If you are embarrassed about the sound, don't blame playmidi: It is the FM synthesis that sounds bad.
Imagine you would only have the OSS/Free Sound Driver: Than this is the quality of MIDI Music you can achieve (apart from software synthesis). Fortunately, you have a SB AWE, and you can use the Wave Table capabilities with the AWE Sound Driver Extension.
The AWE Driver Extension comes with special tools
awesfx) to make use of the EMU 8000 Wave Table
synthesis. First, you have to load a Sound Font Bank on your card -
even if you want to use the ROM samples! You can get the files from
your Windows installation - look for files ending in
The ROM samples can be loaded with
samples are in
SYNTHMT.SBK, as well as in
You can get other Sound Font Banks via ftp or www, try the EMU
Homepage, the web site from Creative Labs, and look out for the
Chaos samples, they are really good (be sure to check out the AWE
Driver Web Site).
Try to load the standard GM (ROM) set with
# sfxload -i synthgm.sbk
and then play a midi file that comes with your SB AWE:
# drvmidi dance.mid
Get your favorite mixer program and start it. Start a raw audio sample and two MIDI files at the same time, and try out the mixer settings. Play a bit, it should be very easy. Here a list of the devices and how they are called:
- Yamaha OPL-3
Synth or FM
- AWE32 Driver
Synth or FM (does anyone know a mixer were the last two are separated from each other?)
- Sound Blaster 16 (4.13)
PCM or DSP
- PC Speaker (still alive)
Other mixer settings refer to the CD ROM, possibly connected to the sound card, to Master Volume, Bass, Treble and Recording Level of the different input lines. You can specify which lines should be recorded.
You can record different sources: a connected CD ROM, a microphone
mic, and any that you can connect to
line in. Set the mixer in the appropriate position.
Play a sound and record into a file, reading from the Raw Audio
Device, for example:
# cdplay # dd bs=8k count=5 </dev/dsp >music.au 5+0 records in 5+0 records out # cat music.au >/dev/dsp
records and plays five seconds of audio from the input device.
Sorry, no information about MIDI port yet!
You can load samples in your DRAM on the sound card with the
sfxload tool. Note that you can only load one sample
file per bank, with bank 0 as default. So, after
synthgs.sbk, the only samples on your sound card are the GS
samples. If you want to load additional Sound Font Banks, you have
to use the
-b option, for example:
# sfxload synthgs.sbk # sfxload -b1 sample.sbk # drvmidi sfx.mid
You can load a default Sound Font Bank automatically with installing the module. Just append a line like
post-install sound /usr/bin/sfxload synthgm.sbk
See the documentation for more details about
drvmidi program, you can use your AWE Driver
to play MIDI files. Just specify the name of your midi file after
# drvmidi waltz.mid
See the documentation for more details about
- The Linux Sound HOWTO
Author: Jeff Tranter,
Last Version: v1.17, 4 August 1997
Please take a good look at the Sound HOWTO (available from sunsite.unc.edu and other sites). It contains a bulk of data about compiling the kernel with sound support, and explains a lot about the sound devices, applications etc.
I understand this HOWTO as an addition to the Sound HOWTO.
- The Linux Sound Playing HOWTO
Author: Yoo C. Chung,
Last Version: v1.4, 13 May 1997
You should also look at the Sound Playing HOWTO. It tells you about the various sound formats and applications to play them.
- The AWE Driver FAQ
Author: Takashi Iwai
If you have problems installing the AWE Driver or using the Wave Table synthesis of your Sound Blaster card, then look here.
- The ISA PnP FAQ
Author: Peter Fox
If you have problems configuring your ISA PnP card, then this is your book of wisdom.
Author: Peter Fox
Last Version: 1.11
Author: Takashi Iwai
Last Version: 0.4.2c
- Linux kernel
Author: Linus Torvald and many other
Last Version: At the moment, you should use 2.0.29
Source: everywhere, where you can obtain Linux
Author: Hannu Savolainen (Please check http://www.4front-tech.com/usslite before mailing me).
Last Version: I have 3.5.5-beta1
Source: with the Linux kernel source package
In the ISA PnP configuration file created by pnpdump, the single devices of your PnP cards appear as sections. In the output below, one card was detected (the sound card), with four logical devices:
LD 0: Audio Device
LD 1: IDE Device
LD 2: WaveTable Device
LD 3: Gameport Device
LD 1 unconfigured, because I have no CD ROM
attached to my sound card IDE port. If you have no IDE port on your
LD 1 will be the Gameport Device and
3 will not appear.
Please refer to section Getting started for further (important!) information about this file.
# $Id: pnpdump.c,v 1.9 1997/06/10 21:37:32 fox Exp $ # This is free software, see the sources for details. # This software has NO WARRANTY, use at your OWN RISK # # For details of this file format, see isapnp.conf(5) # # For latest information on isapnp and pnpdump see: # http://www.roestock.demon.co.uk/isapnptools/ # # Compiler flags: -DREALTIME -DNEEDSETSCHEDULER (READPORT 0x0203) (ISOLATE) (IDENTIFY *) # Try the following if you get error messages like # Error occurred executing request 'LD 2' on or around line... #(VERIFYLD N) # ANSI string -->Creative SB32 PnP<-- (CONFIGURE CTL0048/58128 (LD 0 # ANSI string -->Audio<-- (INT 0 (IRQ 5 (MODE +E))) (DMA 0 (CHANNEL 1)) (DMA 1 (CHANNEL 5)) (IO 0 (BASE 0x0220)) (IO 1 (BASE 0x0330)) (IO 2 (BASE 0x0388)) (ACT Y) )) (CONFIGURE CTL0048/58128 (LD 1 # ANSI string -->IDE<-- # (INT 0 (IRQ 10 (MODE +E))) # (IO 0 (BASE 0x0168)) # (IO 1 (BASE 0x036e)) # (ACT Y) )) (CONFIGURE CTL0048/58128 (LD 2 # ANSI string -->WaveTable<-- (IO 0 (BASE 0x0620)) (IO 1 (BASE 0x0A20)) (IO 2 (BASE 0x0E20)) (ACT Y) )) (CONFIGURE CTL0048/58128 (LD 3 # ANSI string -->Game<-- (IO 0 (BASE 0x0200)) (ACT Y) )) # Returns all cards to the "Wait for Key" state (WAITFORKEY)