Aztech PCI 368DSP Sound Card Review

This sound card is based on VLSI ThunderBird 128 chip. For a considerably small money you get a good sound card supportingsound playback through the headphones and 2/4 speakers using Q3D and QEM via API DS3D and EAX 1.0.

by Ilya Gavrichenkov
01/10/2000 | 12:00 AM

A thunderbird in the sky does not necessarily signify the worsening of the weather and the comingstorm...


In the field of 3D sound we still have the opportunity to enjoy a very exciting battle between Aureal andCreative. In fact, this topic as well as the "show" itself is a very interesting thing. However, Creative andAureal are not the only members of the PC 3D sound market: quite a lot of other companies represented there canboast their own products and corresponding achievements. Unfortunately, the world market is now dominated mostlyby Aureal and Creative products. However, their competitors don't give in and do their best to get a guaranteedplace under the sun. Today we are very anxious to tell you about PCI 368DSP soundcard from Aztech. Why are we soexcited about it? It's very simple. Some time ago, somewhere in 1992, if we are not mistaken, one of our colleagueswas looking for a sound card for his 386 computer. He went to the shop and was offered a choice of SoundBlaster 16from Creative and SoundGalaxy Orion16 from Aztech. At that time nobody, including the soundcards manufacturers, hadany ideas of a real 3D sound. So, our friend decided on a sound card from Aztech. And one of the determining factorswas the possibility to also buy a daughter card with an FM-tuner for a small extra payment. In general, he was verysatisfied with the purchase and enjoyed his new Orion. You are probably wondering why we are telling you all this? Andthe reasons that guided us come from the situation in the sound market, which had long been beloved by Aztech. And wehave to admit that Aztech sound cards cost much cheaper than those by Creative and could boast some unique featuresabsent by the cards from SoundBlaster family, such as the already mentioned FM-tuner, for instance. However, Aztechlittle by little left the territory and one day totally disappeared from the stage. However, the company reshuffledits structure and started dealing with not only sound cards but also acoustics, modems and graphics adapters. And todaywe will speak about a graphics card designed and sold by Aztech, which is relatively new for the sound market. As we havealready mentioned above, the card possesses a puzzling industrial name - PCI 368DSP, and is based on ThunderBird 128 chipfrom VLSI. This chip is especially interesting due to the provided hardware support of QSound Labs sound technologies.


PCI 368DSP sound card is supplied in a small box, which contains the card itself, a CD-disk with the drivers and theaccompanying software, CD-Audio cable and a tiny user's manual.

The supplied software includes demo programs, software sound synthesizer from Yamaha and Clef Studio 3.0 packageintended for sound playback, editing and recording.

On the one hand, this package seems to be rather poor, although the soft synthesizer is still an indisputableadvantage. On the other hand, if we put up with the absence of games among the accompanying software, then a moredetailed user's manual could be just irreplaceable. From the manual, which goes together with the card you can onlylearn to install the driver and that's it. You won't find even the basic sound card parameters. If you are notsatisfied with the contents of the manual, you may take a look at a more detailed version provided on a CD-disk inpdf-format, which has everything you may need. Note that this card is sold in retail for $25, in other words thepackage fully corresponds to its price.

The heart of the card is ThunderBird 128 (VT21702) DSP chip from VLSI. This DSP comes from the second generationVLSI ActiMedia architecture.

Here we would like to draw your attention to one pretty interesting thing. The utilized STAC9704T codec is designedfor one stereo out, which serves to connect 2 acoustic speakers. On the sound card, however, there are 2 stereo line outs.At the same time ThunderBird 128 chip supports 2 channels played through 2 speakers each, but it also allows usingadditional music DAC for another 2 channels, which can be played through 2 rear speakers. 368DSP sound card has onlyone single DAC. The thing is that STAC9704 is capable of sending to 2 line outs a stereo signal encoded for the playbackthrough 4 speakers using a line out and the so called true line level out. If you are curious about more details on thematter, you may download the corresponding .pdf file from SigmaTel web-site. This feature of the codec to provide twostereo signals is beautifully fit for the end-signal produced by QSound technologies. And as for some particular soundtechnologies, we will return to them a bit later in our review.


The sound card got installed very quickly and without any problem.

Yamaha software synthesizer was installed as default, and in the system tray appeared a new icon for sound card controlpanel quick start.

It seems highly attractive that all the drivers and software can be completely and very easily removed, which is a verygood form.

368DSP card control panel looks very familiar if we compare it to those of other sound cards, such as Vortex2. There areseveral properties pages where you can change some settings for your sound card.

Mixer settings and CD Player

The type of the connected acoustics and special effects

QEM settings

And Now a Bit of Theory

Let's dwell on the main technologies provided by QSound Labs and supported by 368DSP card on ThunderBird 128 chip.

The present day situation in the PC sound market looks as follows. QSound can now boast strong and stable positionsonly in the field of sound playback through headphones and 2-speakers. QSound composes its algorithms for playing 3Dsound through headphones basing on the results obtained during tests with real people listening. In other words, puremath1ematics is no longer enough for QSound: they are now focusing on the human perception of the sound. All in all,they carried out over 55,000 listening sessions. In case of headphones, all the companies usually resort to an alreadytraditional approach: HRTF filters, amplitude and time differences between the two signals reaching the listener's ears.However, as soon as it comes to 3D sound playback via speakers, QSound take their own unique way.

To play back the positional sound in 3D environment QSound engineers worked out two algorithms: Q1 and Q2, which are includedinto the sound engine under the code name Q3D. Q1 algorithm is aimed at processing sound streams, which will be played throughthe speakers, and Q2 algorithm is intended for the sound playback through the headphones. Both of them make use of absolutelydifferent filters and sound processing algorithms. Each algorithm takes an arbitrated monophonic signal and applies to itcertain 3D sound source localization filters. As a result, the end stereo signal acquires the info about the location of thesound source represented as time, phase and amplitude differences. Operating the given parameters in a certain way you canchange the sound as if the sound source was moving in the environment. In most cases it is possible to change the filterparameters in real time regime. Besides, the algorithms can be also applied to a few separate signals, and then theresulting stereo signals have to be summed up into one united signal, which contains all the info about the location ofeach sound source taken separately. Moreover, the algorithms can be applied during the signals mixing, which definitelycontributes to the overall efficiency increase and reduces the requirements to the available CPU resources. Q2 algorithmsapply HRTF filters using time and amplitude signal differences. Q1 algorithms can create 4 separate resulting signals,which allows using four-component acoustic systems. Q1 doesn't involve cross-talk cancellation algorithms and hence savescomputing capacity. And since Q1 is based on the results obtained during real listening experiments, the user is free tochoose the speakers type and their location. That's why the rear speakers are usually provided with ordinary panoraming.

Besides the basic algorithms used to create positional 3D sound, there are also some extra parameters, which help tomake the sound closer to reality. Among these special parameters we can list sampling rate (for instance, to create Dopplereffect), selected sounds volume regulation for correct mixing. And in order to avoid some hearable artifacts when thesampling rate is changed, there are special interpolation algorithms.

To add more realism to the created acoustic model, the developers used such environmental effects as controllablereverberation. QSound designed a special QEM engine for the realization of this reverb effect.

QSound aimes at using DirectSound3D interface from Microsoft that's why Q3D engine bases on DS3D. Q3D 2.0 is alreadyprovided with reverberation support. ThunderBird 128 chip is a DSP and supports Q1 and Q2 algorithms as well as signalmixing and extra effects on the hardware level.

QEM (QSound Environmental Modelling) is a sound rendering engine, which allows reproducing realistic reverberationwith the help of EAX 1.0 from Creative. QEM works via EAX 1.0 and is controlled by Q3D 2.0. Besides, QEM permits addingreverberation to games without EAX 1.0 support running via API DS. QEM reproduces environmental acoustics andreverberations of a certain quality, because it takes into account such factors as room size and construction materialsforming the environmental objects. In fact, we can expect QEM 2.0 to be EAX 2.0 and I3DL2 compatible.

QXpander allows increasing the sound field when playing an ordinary stereo sound. This technology is based on Q1algorithm. The sound field grows because the sound sources are moved as far as possible from each other. In this caseall mono signals are duplicated in both sound channels.

QSurround allows playing multi-channel sound, like Dolby Digital, through 2 speakers with surround sound effect thanksto the used Q1algorythm. In this situation there appears a phantom central channel, the front channel sound gets muchbetter due to the added low frequency channel ".1" and the system creates virtual speakers for the side and rear signals,so that to make all the signals sound as if they were played through multiple speakers and not just two. And the subwoofermay also be working in this case. So, since the info about the side and rear signals remains, QSurround provides not justa stereo sound with surround effect. By the way, QSurround possesses a Dolby Labs certificate for Dolby Logic and DolbyDigital. We would also like to point out that there is absolutely no multi-channel sound decoding.

To play the sound through 4 speakers QSound uses QMSS technology, which allows playing the ordinary stereo soundthrough 4 (and more) speakers due to the ability to create separate sound channels. Note that QMSS doesn't simplyrepeat the front speakers signal in the rear ones. It simply creates 4 (or more) separate channels instead, andeach of these channels carries the info required to correctly model a surround stereo. In this case the rearspeakers may also get an additional reverberation effect, which makes the sound heard easier to perceive. It ispeculiar of QMSS technology that it uses 4 speakers for the sound playback if the sound card is equipped withcodec of STAC9704 type.

Unfortunately, QMSS technology is unfit for games. However, QSound is now actively developing a new 3D soundtechnology, which will work for 4 and more speakers and will be intended for 3D positional audio. A new technologywill be supported in ThunderBird's successor - a new ThunderBird Avenger DSP chip.

We would like to stress once again: QSound Labs believe that it is not that important for sound modeling how andby means of what algorithms it is actually done. The main thing is how the listener perceives the modeled sound. Andin fact, QSound is very famous for successful algorithms and reasonable involvement of the available resources (by theway, the resource manager licensed by Microsoft and included into DirectX was developed by this particular company).

And Now a Bit of Practice

The testing system was configured as follows:

First let's say a few words about MIDI playback. ThunderBird 128 chip supports 64-voice polyphony on the hardware level.However, the standard control panel doesn't provide you with the opportunity to upload any external sound bank. As defaultyou will get tbird128.dls, which is supplied together with 368DSP.

As you can see, the setting is severely limited to a very vague term - Studio Quality. However, you can also add sucheffects as reverb, chorus, Game and Max Polyphony. We failed to distinguish all these effects by hearing, but it is verysubjective and it may be our bad ear. Simple as that...

Besides, 368DSP card is also accompanied with a software synthesizer by Yamaha - SoftSynthesizer S-YXG50 v.,which allows playing MIDI files with a 256-voice polyphony. Yamaha synthesizer offers you a choice of 676 differentinstruments and 3 types of effects (reverb, chorus and their combo). You may also choose the level of CPU utilizationwhen using the synthesizer.


The sound provided by Yamaha software synthesizer is very rich. That is why if you are fond of listening to MIDI,you won't be disappointed. Moreover, in our system the CPU utilization was not exceeding 30% while the quality and theamount of voices were set to their maximum.

We tested this sound card with synthetic tests as well as in real applications. Here we want to note right away thatthe sound perception is a very subjective process. Every person hears the sound in his unique way. It is very difficultto describe verbally what you actually feel when you hear this or that sound and besides, the impressions provided by anumber of experiment participants are very unlikely to coincide. That is why you'd better take everything you will readbelow only as something to think of in the meanwhile. Ideally, each of you should listen to it himself and make his ownconclusions.

You should bear in mind that all the tests were run with the headphones on and through the 2 speakers with a subwoofer.We used the configuration with 4 speakers and a subwoofer only when listening to CD Audio and watching DVD video. The thingis that you can use all the 4 speakers to the full extent only in these two cases, because in other situations the sound issimply duplicated and the rear speakers play the same thing as the front ones. Besides, the volume level of the rear speakersis very low and cannot be regulated, and the sound quality is very bad, so that you can even hear background noises similar tothe white noise.

In older games, which use only DirectSound interface without EAX 1.0 support, you will be able to activate QEM engine andhence to reproduce reverberation effects. In this respect, you have to go to 368DSP control panel and choose the desiredreverberation type for your game. Reverberation really works in Quake2, for instance, however, it remains unchanged forthe whole game and if you are interested in our personal opinion, the sound there leaves much to be desired. Besides, youmay also activate 2D to 3D regime, which is in fact very similar to QXpander because the algorithms used are the same Q1.When listening through headphones, it is possible to use QEM only.

If the game supports EAX 1.0, then you should enable QEM support in the control panel. However, you don't need to selectany particular reverberation effect, but leave it set to Generic, so that the game acoustics could change dynamically fromone level to another and even when you move from room to room.

Minerva 1.08 test didn't show anything particular. It only set the number of sound streams supported (2D and 3D) equal to256. However, the reproduced mixed sound was not very clear and we could even hear some crackling noises and notice the tonedeformation. We believe it has to do with the imperfection of Minerva test.

We took the card's amplitude-frequency response (hereinafter called AFR) with the help of Sound Analyser test. Especiallyfor this purpose we connected the line in and out with a loop. All the measurements were carried out five times and then wepicked out the best result.


As you can see, the AFR graph of 368DSP card is rather level. The card owes it mainly to high quality of thecodec by SigmaTel.

ZD Labs Audio Winbench99 test served to get the CPU utilization and to subjectively estimate the sound quality withDirectSound and DirectSound3D interfaces.

The tests aimed at subjective sound perception finished successfully. We don't have any specific remarks here. Thesound positioned well, although even with the headphones on we could feel that the sound was slightly moved forward.

All the sound cards tested support 32 hardware DS3D streams. Here are the results obtained in the CPU utilizationtest taken for the sampling rate equal to 44.1KHz with 16bit precision:

 MX300SB Live! PlayerPCI 368DSP
Static: Voice 82.413.225.07
Static: Voice 162.533.495.01
Static: Voice 322.675.116.04
Streaming: Voice 82.593.45.01
Streaming: Voice 162.593.525.08
Streaming: Voice 322.715.115.99

It is clear that the card on Vortex2 shows very low CPU utilization, while the card on ThunderBird 128 showed thehighest CPU utilization of all. Nevertheless, even a 6% CPU utilization is still very-very little. We have to mentionthat all the tests using API DS showed absolutely the same results.

With the help of demo programs included in the supply we can evaluate the sound perception with the help of Q3D engineby QSound. Note that all demos are started from a CD, i.e. they don't get copied to your disk during installation.

Space Junk demo presents a virtual world where you can move. In this world you meet some objects, which can also moveand emit various sounds. You can come across 8 sound sources at a time. Moving in this virtual world you can sometimesclose your eyes and check if you are detecting the type and the location of the sound source correctly. You can alsodisable hardware sound support and listen to the software emulation.

The second demo program is called Wumpus. This program is also intended to show sound sources positioning with the help ofQ3D. However, here the developers chose a very extraordinary approach: you don't have to close your eyes to evaluate thepositioning quality. This demo is none other but an arcade game imitator and you have to shoot all these wampuses, whichyou can't see but can hear. In fact, you aim your rifle guided only by what you hear.

The third demo is a "free" version of a music editor. This idea is also rather unusual because the user doesn't need tohave any specific skills in composing music. All you should be able to do is just move your mouse or joystick.

In order to estimate the CPU utilization level in real applications we turned to Heretic II game, which supports A3D 2.0and DS3D+EAX 1.0.

We used ntdemo.hd2 (120KB) and Heretic II with 1.4 patch (with the older patches it doesn't work). The sampling rate wasset to 22KHz on the game settings page (High).

At first we ran the game in Software Rendering mode at 320x240@16bpp, i.e. when all the graphics is calculated by the systemCPU. Here is what we've got:

 No sound368DSP EAX 1.0SB Live! EAX 1.0MX300 EAX 1.0MX300 A3D 2.0MX300 A3D 2.0 w/o WTr
Software Rendering 320x240@16bpp39.8 fps37.5 fps37.5 fps34.7 fps32.6 fps34.9 fps

Wavetracing (WTr) support was disabled with "s-geometry 0" command.

We would like to point out that when we used A3D 2.0 with the old a3dapi.dll (v.2.10) library, we got 18.5fps. The newa3dapi.dll - version 2.25.

Now let's enable hardware graphics output acceleration through API OpenGL and take another look at the results:

 No sound368DSP EAX 1.0SB Live! EAX 1.0MX300 EAX 1.0MX300 A3D 2.0MX300 A3D 2.0 w/o WTr
OGL 640x480@16bpp82.1 fps74.6 fps73.8 fps66.5 fps63.3 fps67.4 fps
OGL 1024x768@16bpp58.8 fps58.1 fps58.1 fps56.9 fps56.9 fps57.0 fps

So, what can we conclude taking into account the results obtained? All sound cards tested more or less utilize the systemCPU. In general, it is not a discovery for us. We can see that the most CPU consuming appears Wavetracing. On the other hand,there are not so many people playing games in Software Rendering mode. Hardware graphics acceleration frees some CPU resources.Besides, at 640x480 the graphics card performance is restricted by the CPU abilities and that's why the difference in CPUutilization for the sound is very tangible no matter how insignificant it may seem. However, at larger resolutions, such as1024x768, it appears not the CPU but the graphics card, which limits the system performance, so, the choice of a sound cardis no longer so dependent on the CPU utilization level.

Note that SB Live! cards and 368DSP utilize the system CPU on almost the same level.

Of course, we used the games in order to get a clear picture of the features of 368DSP sound card. Just in case you forgot:ThunderBird 128 chip provides hardware support for the 3D positional audio engine - Q3D working through API DS3D. Besides, wehave also involved QEM reverberation engine working through EAX 1.0 interface.

We used the headphones and speakers in the following games: Heretic II, Half Life, Unreal Tournament, Baldur's Gate, NeedFor Speed: HS, Thief: The Dark Project and some more. QSound utilizes its own peculiar solutions during the sound playbackthrough Q3D engine. We have every reason to state that the sound playback through 2 speakers was very good. We think that368DSP proved there even better than SB Live! and almost as good as MX300. Well, QSound deserved our congratulations for awell done Q3D engine. As for the sounding through the headphones, it yields a bit to MX300, but easily surpasses Live!.

Speaking about the reverberation quality, we can't recall any difference in the way the games sounded with 368DSP and SBLive!. Although sometimes the reverb effect provided by 368DSP seemed a bit harder than that provide by Live!. And the maindifference lied not in the reverberation quality but in the general impression made by the sound. We would like to stress onemore time that every listener hears all sounds differently: if we miss something, then you may probably notice it and vice versa.

368DSP sound card supports A3D 1.0. This purely software support is provided by Q3D engine. It converts A3D 1.0 calls intofunction calls in DS3D. We checked A3D 1.0 support in Thief: everything was just perfect. We wish it also supported A3D 2.0,however, Aureal doesn't let its competitors introduce this feature in their sound chips. In fact, no one can guaranteeimpeccable performance of 368DSP sound card in all games supporting A3D 1.0. Take for instance, Heretic II. We failed to runthis game with sound through API A3D 1.0 unless we installed the patches adding A3D 2.0 support.

We feel like reminding you that the support of A3D 1.0 only doesn't allow 368DSP card to compete with those based on Vortex2chip, which can boast A3D 2.0 hardware support. In fact, another version - A3D 3.0 is already appearing and probably Aureal willchange its position in terms of A3D 2.0. Well, we'll see...

If we compare the sound produced with Q3D (with QEM) to that produced with A3D 2.0, we would probably vote for dynamicallychanging sound sources obtained basing on the environment geometry and for early sound reflections rather than for staticreverberation and Q3D positioning. However, we do not impose our opinion: tastes differ.

QEM 2.0 is expected to support EAX 2.0 and I3DL2, which will be improved towards higher dynamics of reverberation effectsby means of added occlusions and obstructions. Besides, EAX 2.0 and I3DL2 will be reversibly compatible with EAX 1.0. If youare curious to see how it all works, you may take a look at EAX 2.0 demo from Creative. Now it isn't yet clear if QEM 2.0support will be introduced into 368DSP drivers. If it is made there, then 368DSP sound card will undoubtedly gain inattractiveness, because its service time will get much longer. And in the meanwhile we can be only sure that 368DSP isjust a good modern sound card, which provides beautiful sounding in games when played through 2 speakers and competitivesounding quality in headphones.

Anyway, every technology has its cons and pros. We believe that one day all sound chips manufacturers will introducereverberation and wavetracing support into their products. All we can do is just wait. To tell the truth, Aureal seems toturn out the first company supporting all the technologies listed above. New API A3D 3.0 allows using reverberation effectstogether with wavetracing. Actually, QSound is also very unlikely to be keeping hands in pockets that's why we can hope tosee something fresh new pretty soon. By the way, QSound web-site sells various utilities, which allow us for instance, toplay mp3 files with surround sound effect.

We listened to the sound played by 368DSP from a DVD and a CD disk. It was very nice indeed. When all 4 speakers are active,QMSS technology is used, which makes the sound perception undoubtedly better. The card doesn't support any multi-channel sounddecoding, however, the sound quality is nevertheless very high and can be compared even to Dolby Pro Logic quality. In the nearfuture QSound is planning to modify QMSS, so that it will be possible to reproduce all types of the sound through 5.1 acousticsystems.

With QXpander the sound is very good, however, 2 speakers will hardly be enough after you've heard the 4. But if youare hard up or lack room, you will by more than pleased with this solution from QSound, which provides an excellent soundthrough only 2 speakers.


The retail price on 368DSP sound card is about $25. For this considerably small money you get a good sound card supportingsound playback through the headphones and 2/4 speakers using Q3D and QEM via API DS3D and EAX 1.0. Besides, it also supportsA3D 1.0. In fact, you will be able to hear 3D sound in most games available in the market today. Since 368DSP card is anabsolutely fully-fledged PCI card with hardware sound playback and mixing support, you will also get the whole bunch of suchpleasant trifles as simultaneous reproduction of different sound streams, including 2D and 3D. For instance, you can run a CDplayer, an mp3 player, FM radio and a favorite game, and all the sounds will be played at one and the same time. All you haveto do is just to set the volumes according to your preferences and that's it.



We mentioned the absence of digital out (like S/PDIF) among the drawbacks of this product, because today this functionalfeature is practically becoming a standard. Today you can easily buy 5.1 acoustics accompanied by multi-channel decoders. Andsince there is what to connect then it should be possible to connect it.

It is also worth mentioning that reverberation effects cannot be applied to the sound during CD Audio and mp3 playback, whichis possible with SB Live! cards. However, we didn't list this item among the card's lows on purpose. Frankly speaking, we don'tthink that CD sound needs any extra effects. On the other hand, however, some of you may like it. That is why we decided simplyto state the fact and to let you make your own conclusions.

Considering the card's features, it is undoubtedly better than the cheaper sound cards on Vortex (AU8820) and Vortex Advantage(AU8810). In games supporting EAX 1.0 368DSP and SB Live! run neck and neck, with the only exception of 4 speakers output. Nowthere is no info about EAX 2.0 and I3DL2 support. However, it is very likely to be introduced especially since it has beenalready announced for the new ThunderBird Avenger, which will soon replace ThunderBird 128.

In terms of 2 speakers output, 368DSP card is an indisputable leader. That is why if you prefer to listen to the soundthrough 2 speakers and a subwoofer, then you should definitely decide on Aztech. And trust us: you won't be disappointed.

Now there are sound cards on HSP ForteMedia FM801-AS chip, which also support Q3D and QEM and are equipped with 2 lineouts. Note that at first sight they seem very similar to 368DSP on DSP ThunderBird 128 chip. However, sound cards on HSPForteMedia FM801-AS cost less than $15! What's the difference then? And the thing is that ThunderBird 128 is a DSP, whichimplies that all algorithms of Q3D are processed on the hardware level. ForteMedia FM801-AS chip is a HSP, i.e. Host-basedSignal Processing. It means that Q3D engine works due to the system CPU and the sound is modeled on the software level. So,if you decide to save 10 bucks, then you have to bear in mind that there will be fewer fps in games in this case than witha ThunderBird 128 based card.

So, if you are looking for a low-cost sound solution for the nearest 6 months, then 368DSP sound card based on ThunderBird128 chip and supporting QSound Labs technologies is exactly what you need. Besides, if you are tired of arguing about thecompetition between Vortex2 cards and SB Live!, then you will get a beautiful opportunity to stay aside and to watch thewhole thing passively.


Since we have mentioned ThunderBird Avenger chip, let's say a few words about it too. You probably know that Philipsbought VLSI and that VLSI keeps working in the field of DSP sound cards development. In 2000 the market should see a newDSP chip aka ThunderBird Avenger. This chip fully corresponds to the main tendencies of the PC sound technologies.ThunderBird Avenger chip has Q3D 2.0, QSurround and QMSS hardware support. It also supports the conversion of differentmedia types (CD Audio, MP3, MIDI, gaming sound) and further 6-channel output for playback through 5.1 acoustic systems.The chip has S/PDIF out to connect external receivers. In fact, ThunderBird Avenger can boast new version of QMSS, however,as for the AC-3 streams hardware decoding, there won't be any. The chip will work through API DS3D, EAX 1.0 and A3D 1.0; besides,it will also support EAX 2.0 and I3DL2. Well, we can only patiently wait for the first sound cards on ThunderBird Avenger to appearin the market. And among some other interesting features, we can mention up 96 DS3D sound streams at a time and in case of DS sound -up to 256 streams.