DigestIToday's motherboards support ISA interface cards less and less. Many cards that were originally ISA, such as network cards, have one after another brought out PCI-interface versions. Even sound cards have appeared in PCI versions. At present, for most people, PCI sound cards are still a product that is too unfamiliar; many people still only half understand them. Why should sound cards change from ISA to PCI interface, what benefits does it have? What limitations does it have? Is it worth buying? Will you be disappointed after buying one? You can get some answers here.
What are FM and Wavetable?
Before getting into the main topic, I must first talk to everyone about Wavetable, so that everyone has some basic concepts. Most early sound cards were FM sound cards. When playing MIDI, the way they synthesized instrument timbres used Frequency Modulation synthesis, using sine waves (SIN) and cosine waves (COS) to synthesize various waveforms. With just these two functions, all kinds of endlessly changing waveforms can be composed. This is one kind of the famous "Fourier transform" function; students in civil engineering, electrical engineering, and other science and engineering departments should all have learned this. This synthesis method needs many oscillators, but oscillators are very expensive, so cheap FM sound cards use only a few simple oscillators to synthesize, and then use approximation methods to synthesize the waveform. Therefore the synthesized waveform is comparatively smooth, and when you listen to it you immediately know it is the tone of "computer music" (music emitted by a computer), not at all like real instruments (the timbre waveforms of real instruments have large rises, falls, and changes).
Therefore later came the so-called Wavetable sound cards. The instrument timbres of this type of sound card do not use frequency synthesis, but come from sound samples of real instruments. The collection of all the instrument samples becomes a set of Wavetable functions. These Wavetable timbres are all stored on the sound card beforehand, and are then processed by the sound chip to synthesize realistic timbres, so the quality of MIDI playback on a Wavetable sound card is closer to the feeling of real instruments being played.
The Necessity of RAM or ROM
Ultrasound with built-in 1 MB RAM (upper left corner)
Wavetable is a set of waveform table functions for instrument timbres. When playing MIDI, it must first be stored in memory on the sound card so that the sound processing chip can process it and synthesize real musical sounds. Therefore the sound card needs ROM or RAM to store these timbres. ROM means the timbres are fixed in the chip; for example, Acer's Magic S23 sound source card has built-in 4 MB ROM to store timbres. RAM means the timbres inside can be dynamically replaced; for example, the Ultrasound sound card or Creative AWE 32. No matter which kind, any sound card that advertises Wavetable needs to provide storage media on it, just like a display card has Video RAM. This is the most typical characteristic of a Wavetable sound card, while ordinary sound cards without memory are the so-called FM sound cards, such as cheap sound cards using the ESS1868 chip.
Why make it PCI?
All right, now comes the question: why should sound cards be made with a PCI interface?? The reason sound cards are made PCI lies in PCI's high bandwidth. The PCI bus is a data bus with a width of 32 bit and a clock of 33 MHz. Its theoretical bandwidth is as high as 132 MB/Sec, over 100 MB/Sec, while ISA has only 8.33 MB/sec, not even 10 MB/sec, so the two differ by more than 10 times.
PCI sound card without RAM
Therefore the Wavetable sampled timbres needed by the sound card can first be placed in system memory, and then sent through the fast PCI bus to the sound chip for processing and synthesis. Therefore the RAM or ROM originally needed on ISA sound cards can be saved on PCI sound cards, because they have already been replaced by system memory. The result is that the cost of the sound card is reduced. Therefore reducing cost is the biggest selling point of PCI sound cards, allowing you to spend less money and get the same sound quality. You absolutely will not see any PCI sound card still equipped with ROM or RAM; otherwise there would be no need to make it a PCI interface, because this violates the design concept of PCI and is no different from "contradicting itself." (But some PCI sound cards still provide RAM slots for expansion, so that you can avoid using or reduce the use of system memory. However, I personally think this is a bit like drawing legs on a snake, because I guess probably not even half a person will go expand it, including myself. I will mention this in detail later.)
Let's look at the RAM for Creative SB AWE64. The expansion cost of just an 8 MB memory card is probably about 2000 yuan; in the used market maybe someone sells it for over a thousand. So you can see RAM really accounts for quite a bit of the cost of an entire sound card (although today's RAM prices have fallen ridiculously). After changing to PCI, these costs are no longer needed, and 2 MB, 4MB, or even 8 MB versions of timbres can be freely provided for the sound card to use, completely unrestricted by the size of RAM or ROM, and with no need for hardware upgrade costs.
Differences from software sound sources
Some people may confuse this characteristic of PCI sound cards with "software sound sources" or "software Wavetable." In fact there are still differences between them. So-called software sound sources, such as YAMAHA's YXG50 or ROLAND's VSC-88, are mainly because the processing and synthesis of Wavetable instrument timbres was originally done by the sound chip, but now this work is completed instead by the CPU. Therefore your CPU must be quite powerful; the recommended basic requirement is a Pentium 166 MHz and 16 MB of memory before it can run. This is just like XING software MPEG: originally the image decompression work was done by an MPEG card, but now it is handed over to the CPU, so it is called software emulation. So to watch Video CD, you do not necessarily have to buy an MPEG card (image decompression card); as long as your CPU is powerful enough, you can emulate it through Xing. By the same principle, to enjoy Wavetable sound quality, you do not necessarily have to buy a Wavetable sound card. The ESS1868 sound card you bought for five or six hundred yuan can also let you enjoy the excellent effect of Wavetable through YAMAHA software sound source emulation and calculation. As for so-called PCI sound cards, the synthesis of instrument timbres is still processed and calculated by the sound chip on the card; now it merely replaces RAM or ROM with system memory. I hope everyone can distinguish this point.
DLS timbres
Because the timbres needed by PCI sound cards are stored in memory by RAM or ROM, the timbres first exist on the hard disk in the form of files. When MIDI is to be played, they are loaded into main memory. Because of this characteristic, PCI sound cards all say they comply with the DLS 1.0 specification; that is, their timbres are so-called DownLoadable Sample, translated into Chinese as "downloadable sampled timbres," stored on the hard disk in the form of files.
Because timbres become files stored on the hard disk, you can use DLS timbre editing software to replace or modify timbres yourself, making the timbres better meet your own needs. However, at present there does not seem to be any DLS timbre editing software; I personally have not seen any either. Perhaps these programs will have practical value in the future. If their interfaces are friendly enough, perhaps everyone will be able to use this little tool to edit timbres by themselves.
Will a PCI sound card sound better than an ISA sound card?
It must be stated again that the biggest selling point of PCI sound cards lies in reducing cost, obtaining the same Wavetable sound quality with less money. So whether a PCI sound card sounds good when playing MIDI in fact still depends on the type of its sound chip, and whether the WaveTable instrument timbres it provides are beautiful. This is the most important point. It is not because it is PCI that its timbres will definitely be more beautiful than ISA and its effect will be better; there is no absolute relationship. Of course, the effect will certainly be better than an FM sound card; that goes without saying.
The Challenge Faced - DOS Game Compatibility
Reducing cost is the biggest selling point of PCI sound cards. This idea is indeed not bad, but changing sound cards from the traditional ISA interface to the PCI interface faces the problem of software compatibility. How so? Because under DOS, the sound program part of every game is written for ISA sound cards. To put it more clearly, the DMA signal needed by ISA does not exist on PCI, so compatibility problems between PCI sound cards and DOS games will inevitably arise, such as no sound, or strange sounds, noise, and so on. Playing games with no sound will of course make playing unpleasant, and you will regret buying it. This is also the point ordinary consumers, especially GAME players, doubt most about PCI sound cards, and fear most. This problem can only be solved through good drivers.
It must be stated that this cannot be counted as a drawback of PCI sound cards; it can only be called a difficulty. Once the difficulty is solved, then in use PCI sound cards are basically no different from traditional ISA sound cards. For games for Win95, the problem is comparatively smaller, because games and sound cards call each other through the DirectSound application interface. Therefore as long as both have very good support for DirectSound, problems are less likely to occur. The biggest problem lies in these games for DOS, including the requirement that they must be able to run normally in both the traditional DOS environment and the DOS window under Win95.
Actually the manufacturers of PCI sound cards also attach special importance to this problem. If a sound card has very poor compatibility with existing games, cannot be driven, produces no sound, or produces noise and strange sounds, then will it still be welcomed? No. It will certainly suffer a crushing defeat and then be neglected. So manufacturers will definitely spare no effort in this respect; they will not joke around with their own products. To put it more plainly, these manufacturers that produce sound chips must provide appropriate drivers; only then will they attract other manufacturers to use their chips to make sound cards. So regarding the compatibility issue between PCI sound cards and game software, I personally feel there is no need to worry too much.
What are FM and Wavetable?
Before getting into the main topic, I must first talk to everyone about Wavetable, so that everyone has some basic concepts. Most early sound cards were FM sound cards. When playing MIDI, the way they synthesized instrument timbres used Frequency Modulation synthesis, using sine waves (SIN) and cosine waves (COS) to synthesize various waveforms. With just these two functions, all kinds of endlessly changing waveforms can be composed. This is one kind of the famous "Fourier transform" function; students in civil engineering, electrical engineering, and other science and engineering departments should all have learned this. This synthesis method needs many oscillators, but oscillators are very expensive, so cheap FM sound cards use only a few simple oscillators to synthesize, and then use approximation methods to synthesize the waveform. Therefore the synthesized waveform is comparatively smooth, and when you listen to it you immediately know it is the tone of "computer music" (music emitted by a computer), not at all like real instruments (the timbre waveforms of real instruments have large rises, falls, and changes).
Therefore later came the so-called Wavetable sound cards. The instrument timbres of this type of sound card do not use frequency synthesis, but come from sound samples of real instruments. The collection of all the instrument samples becomes a set of Wavetable functions. These Wavetable timbres are all stored on the sound card beforehand, and are then processed by the sound chip to synthesize realistic timbres, so the quality of MIDI playback on a Wavetable sound card is closer to the feeling of real instruments being played.
The Necessity of RAM or ROM
Ultrasound with built-in 1 MB RAM (upper left corner)
Wavetable is a set of waveform table functions for instrument timbres. When playing MIDI, it must first be stored in memory on the sound card so that the sound processing chip can process it and synthesize real musical sounds. Therefore the sound card needs ROM or RAM to store these timbres. ROM means the timbres are fixed in the chip; for example, Acer's Magic S23 sound source card has built-in 4 MB ROM to store timbres. RAM means the timbres inside can be dynamically replaced; for example, the Ultrasound sound card or Creative AWE 32. No matter which kind, any sound card that advertises Wavetable needs to provide storage media on it, just like a display card has Video RAM. This is the most typical characteristic of a Wavetable sound card, while ordinary sound cards without memory are the so-called FM sound cards, such as cheap sound cards using the ESS1868 chip.
Why make it PCI?
All right, now comes the question: why should sound cards be made with a PCI interface?? The reason sound cards are made PCI lies in PCI's high bandwidth. The PCI bus is a data bus with a width of 32 bit and a clock of 33 MHz. Its theoretical bandwidth is as high as 132 MB/Sec, over 100 MB/Sec, while ISA has only 8.33 MB/sec, not even 10 MB/sec, so the two differ by more than 10 times.
PCI sound card without RAM
Therefore the Wavetable sampled timbres needed by the sound card can first be placed in system memory, and then sent through the fast PCI bus to the sound chip for processing and synthesis. Therefore the RAM or ROM originally needed on ISA sound cards can be saved on PCI sound cards, because they have already been replaced by system memory. The result is that the cost of the sound card is reduced. Therefore reducing cost is the biggest selling point of PCI sound cards, allowing you to spend less money and get the same sound quality. You absolutely will not see any PCI sound card still equipped with ROM or RAM; otherwise there would be no need to make it a PCI interface, because this violates the design concept of PCI and is no different from "contradicting itself." (But some PCI sound cards still provide RAM slots for expansion, so that you can avoid using or reduce the use of system memory. However, I personally think this is a bit like drawing legs on a snake, because I guess probably not even half a person will go expand it, including myself. I will mention this in detail later.)
Let's look at the RAM for Creative SB AWE64. The expansion cost of just an 8 MB memory card is probably about 2000 yuan; in the used market maybe someone sells it for over a thousand. So you can see RAM really accounts for quite a bit of the cost of an entire sound card (although today's RAM prices have fallen ridiculously). After changing to PCI, these costs are no longer needed, and 2 MB, 4MB, or even 8 MB versions of timbres can be freely provided for the sound card to use, completely unrestricted by the size of RAM or ROM, and with no need for hardware upgrade costs.
Differences from software sound sources
Some people may confuse this characteristic of PCI sound cards with "software sound sources" or "software Wavetable." In fact there are still differences between them. So-called software sound sources, such as YAMAHA's YXG50 or ROLAND's VSC-88, are mainly because the processing and synthesis of Wavetable instrument timbres was originally done by the sound chip, but now this work is completed instead by the CPU. Therefore your CPU must be quite powerful; the recommended basic requirement is a Pentium 166 MHz and 16 MB of memory before it can run. This is just like XING software MPEG: originally the image decompression work was done by an MPEG card, but now it is handed over to the CPU, so it is called software emulation. So to watch Video CD, you do not necessarily have to buy an MPEG card (image decompression card); as long as your CPU is powerful enough, you can emulate it through Xing. By the same principle, to enjoy Wavetable sound quality, you do not necessarily have to buy a Wavetable sound card. The ESS1868 sound card you bought for five or six hundred yuan can also let you enjoy the excellent effect of Wavetable through YAMAHA software sound source emulation and calculation. As for so-called PCI sound cards, the synthesis of instrument timbres is still processed and calculated by the sound chip on the card; now it merely replaces RAM or ROM with system memory. I hope everyone can distinguish this point.
DLS timbres
Because the timbres needed by PCI sound cards are stored in memory by RAM or ROM, the timbres first exist on the hard disk in the form of files. When MIDI is to be played, they are loaded into main memory. Because of this characteristic, PCI sound cards all say they comply with the DLS 1.0 specification; that is, their timbres are so-called DownLoadable Sample, translated into Chinese as "downloadable sampled timbres," stored on the hard disk in the form of files.
Because timbres become files stored on the hard disk, you can use DLS timbre editing software to replace or modify timbres yourself, making the timbres better meet your own needs. However, at present there does not seem to be any DLS timbre editing software; I personally have not seen any either. Perhaps these programs will have practical value in the future. If their interfaces are friendly enough, perhaps everyone will be able to use this little tool to edit timbres by themselves.
Will a PCI sound card sound better than an ISA sound card?
It must be stated again that the biggest selling point of PCI sound cards lies in reducing cost, obtaining the same Wavetable sound quality with less money. So whether a PCI sound card sounds good when playing MIDI in fact still depends on the type of its sound chip, and whether the WaveTable instrument timbres it provides are beautiful. This is the most important point. It is not because it is PCI that its timbres will definitely be more beautiful than ISA and its effect will be better; there is no absolute relationship. Of course, the effect will certainly be better than an FM sound card; that goes without saying.
The Challenge Faced - DOS Game Compatibility
Reducing cost is the biggest selling point of PCI sound cards. This idea is indeed not bad, but changing sound cards from the traditional ISA interface to the PCI interface faces the problem of software compatibility. How so? Because under DOS, the sound program part of every game is written for ISA sound cards. To put it more clearly, the DMA signal needed by ISA does not exist on PCI, so compatibility problems between PCI sound cards and DOS games will inevitably arise, such as no sound, or strange sounds, noise, and so on. Playing games with no sound will of course make playing unpleasant, and you will regret buying it. This is also the point ordinary consumers, especially GAME players, doubt most about PCI sound cards, and fear most. This problem can only be solved through good drivers.
It must be stated that this cannot be counted as a drawback of PCI sound cards; it can only be called a difficulty. Once the difficulty is solved, then in use PCI sound cards are basically no different from traditional ISA sound cards. For games for Win95, the problem is comparatively smaller, because games and sound cards call each other through the DirectSound application interface. Therefore as long as both have very good support for DirectSound, problems are less likely to occur. The biggest problem lies in these games for DOS, including the requirement that they must be able to run normally in both the traditional DOS environment and the DOS window under Win95.
Actually the manufacturers of PCI sound cards also attach special importance to this problem. If a sound card has very poor compatibility with existing games, cannot be driven, produces no sound, or produces noise and strange sounds, then will it still be welcomed? No. It will certainly suffer a crushing defeat and then be neglected. So manufacturers will definitely spare no effort in this respect; they will not joke around with their own products. To put it more plainly, these manufacturers that produce sound chips must provide appropriate drivers; only then will they attract other manufacturers to use their chips to make sound cards. So regarding the compatibility issue between PCI sound cards and game software, I personally feel there is no need to worry too much.
MSN：tiqit2@hotmail.com
