I changed some info in the 'History' part. IIRC Protected Mode is not the same as EMS386, even if EMM386 uses a Protected Mode function to switch the processor into Virtual x86 mode. This article is also in need of a separation of technical info from history... 22.214.171.124 19:11, 13 October 2005 (UTC)
Anyone got a GFDL image of the GF1 to use instead? I'll take a pic soon and use that unless another GUS owner beats me to it. The current pic is effectively unlicenced. Riflemann 13:07, 9 January 2006 (UTC)
Hi, I'm rmentat from Spain. There are some incosistences on the text. As my english is not very fluent I ask someone else to make the changes. Thank you very much.
First GUS used the GF1 wavetable chip for playback instead of a codec. The wavetable channels can be used for MIDI or tracker instruments, or as standard playback channels for games. Each channel is 16 bit mono; two are needed to playback a stereo sample. Up to 14 channels can be mixed at CD quality (16bit, 44.1KHz); more channels degrade the frequency up to 19KHz when using all 32 channels. The direct competitor of the GUS when it reaches the market was the Sound Blaster 16, wich was capable of stereo playback and recording at 44.1KHz, 16bit. The GUS was capable of recording only at 44.1KHz stereo 8 bits, so it has an important weak point in front of SB16, because 8 bits recording has much less quality than 16.
Later Creative Labs reacted launching the AWE32 soundcard, wich was very similar to the GUS or slighty better on some aspects (has 32 channels mixed at 44.1KHz, cut-off filters and chorus/reverb, for example).
For its part, Gravis launched the mentioned daughterboard with a Crystal codec wich gives the GUS the ability to record at 16 bit,48KHz. The next soundcard of the serie was the Ultrasound Max wich comes with the same Crystal codec.
Last GUS was the Ultrasound Plug'n'Play, wich came with the Crystal codec but replaced the GF1 with an AMD Interwave chip. It's 100% compatible with original GUS and has very little improvements (32 channel playback at 44.1KHz, basic reverb).
The GUS was not easiest or more difficult to program than other soundcards. It was just very well documented. That's the reason it has so many followers on the demoscene. The other really great reason was Fasttracker II. On the ages of the 386 it was a complete revolution.
Audio CD and MP3 finally ended with MIDI and modules on games, and with non-professional wavetable soundcards. --126.96.36.199 01:01, 3 February 2006 (UTC)
The history section is ridiculously long-winded. Ultrasound doesn't even enter the picture until past the fourth paragraph or so.
I think parts of the history section should be fork off into its own page on the evolution of PC sound capabilities (if such a page doesn't already exist). The history section then needs to be summarized on the parts that are not directly related to Gravis.
I'm busy at the moment but will proceed with said changes over the next few days. I will also attempt to incorporate other suggested changes mentioned above.
188.8.131.52 02:45, 8 February 2006 (UTC)
This article sure needs some restructuring. I owned three consecutive cards from Gravis (GUS Classic v3.74, GUS MAX v 2.1 and GUS PnP Pro) so I will try to rewrite it with some more technical details and clarifications, but I need to refresh me memory because I've sold the last card back in 1998 as you may imagine, after Gravis got out of the market and it was clear that no further developments will be made. --DmitryKo 00:23, 12 March 2006 (UTC)
- Reference to ICS11614 is right on the photo of the chip
- Hardware analog mixer is intended for analog inputs; you could control the volume of digital sources, i.e. wave and MIDI, because the driver worked around it
- GF1 is not "sample mixer", it's digital sample-based synthesizer, so no need to explain the entire concept of sample-based synthesis here
- SoundBlaster emulation was important, but there's no need to repeat it so many times - most GUS buyers knew what they get
- Whether or not GUS could fare better if it was released in Windows 95 timeframe is a POV - I personally believe it could have changed much, because the DOS gaming was abandoned almost instantly, but the failure of AMD InterWave speaks for itself.
Yeah the "fare better in Windows 95" part is POV, and ironic. Back in the day I had a GUS classic, (actually, I still have it...) and the #1 reason I eventually stopped using it was the fact that only mostly unusable beta Windows 95 drivers were released, that did not even support the wavetable features, and were never developed past beta. I ended up popping a cheapo SB16 clone in beside it for Win 95 use. I don't know if the reasons better drivers never came about were technical, or just lack of effort. (As noted, later GUS revisions basically came with the SB16 clone "built in"...) Linux was able to use the GUS as a straight DAC just fine. And incidentally, the current ALSA driver stack in Linux, derived directly from the Linux Ultrasound Project. (Which also ironically, once it became ALSA, no longer supported my GUS worth crap, last I checked. Which was sometime around 1999...) The old Linux Ultrasound Project page still seems to be up... http://www.perex.cz/~perex/ultra/ Last updated in 1998!
Hello, Dmitry, I made those edits. I apologize for not double-checking my edits before commiting them. My terminology was probably wrong. I was trying to convey the idea that the GUS mixed PCM pre-recorded samples in real-time -- i.e., it did not 'synthesize' them using mathematical-means (such as the Yamaha DX-7, Roland MT32/LAPC1.)
When you say "GF1 has no codec, so samples need to be downloaded to sample-RAM prior to playback." What does that mean? I was under the impression that the GUS could not play (DMA) direct from main-memory. CPU PIO was possible, but the overhead was horrendous (and sound-quality was terrible, too, because the GF1 wouldn't apply its usual linear-interpolation for sample-rate-conversion.)
You are definitely correct about the analog-mixer. I had the original (v2.2?) GUS Classic, and the line-in/mic-in had NO digital volume-control whatsoever. (On the plus side, the GUS did have stereo MIC-IN -- a novelty for its time!)
- Digital sampler circuitry typically has the ability to control the volume of its voices, and does it in realtime because it's a requirement for a musical instrument. It probably doesn't even deserve explaining here.
- Codec usually reflects a chip that does some coding/decoding. In regard to soundcards, most of them could play precompressed samples, and CS4231 was specifically the chip used in Wndows Sound System sound card from Microsoft. It supported sample rates up to 16 bit 48 kHZ, playback of IMA ADPCM compressed samples, and was able to DMA sounds from system memory just like SoundBlaster.
- The GF1 could also DMA from system memory - if you remember the defaults for SET ULTRASND=220,5,1,11,7 command, the second and the third numbers define DMA channels used by GUS itself and its SB emulation - but it could not play back sample data simultaneously with DMA operations as SB-compatibles did, it was just moving it to local sound RAM (remember, it was just an adopted digital sampler chip) so the developers needed to allocate a buffer in local memory and "stream" sample data from system memory as needed.
- This was the main source of developer's complains, so the CS4231 codec was added to the daughtercard option and the GUS MAX, making it WSS-compatible. It's also the reason why the beta of DirectSound "accelerated" (i.e. "non-emulated") drivers were only released for the GUS MAX) , because DirectSound is all about cyclic wave buffer in the system memory (see WaveCyclic port class in WDM).
- InterWave cards had "accelerated" drivers right from the start, becase the GFA1 had the CS4231-compatible circuitry inside). --Dmitry 11:34, 18 April 2006 (UTC)
Developer Support POV
Anyone care to rewrite the Developer Support section? It's quite POV-biased. --Trixter 05:44, 20 April 2006 (UTC)
Compaq InterWave card has a RAM slot
I had one of the Compaq cards mentioned in the AMD InterWave section of the page. It had a single 30-pin RAM slot on it, despite the article text saying that the clones had no RAM slots.
Tagged fact as dubious. --Powerlord 04:37, 11 June 2007 (UTC)
- Before someone else asks, I've tagged it as dubious because I don't have another source to confirm this with. In fact, I only remember it because it was extremely hard to find a 30-pin 4MB RAM chip. --Powerlord 04:43, 11 June 2007 (UTC)
I have two of the Compaq Interwave cards in front of me right now, and they most definitely have an expansion SIMM socket:  You can even use Gravis' drivers, provided the TEA6330T output is enabled. Whotookplasma (talk) 06:27, 26 November 2007 (UTC)
I'm removing the whole sentence, dubious statement or not. The claim that the boards without RAM expansion couldn't emulate a GUS Classic is wrong, anyway. The non-Gravis Interwave boards, including the Compaq with the SIMM slot, had 512KB soldered onboard. Polpo (talk) 22:49, 15 February 2008 (UTC)
Driver for Windows XP!
- http://come.to/gus_wdm/ — Preceding unsigned comment added by 184.108.40.206 (talk • contribs) 15:18, 3 December 2012 (UTC)
Remarkable for MIDI playback?
The GUS was remarkable for MIDI playback quality with a large set of instrument patches that could be stored in its own RAM, having up to 32 hardware audio channels.
In terms of the MIDI synthesizer portion of this card --
This device may have been remarkable for its price point, but the MT-32 predated this hardware, had downloadable instruments and while not strictly a wave-table synthesizer, the end result of the MT-32 was a superior musical instrument. — Preceding unsigned comment added by 220.127.116.11 (talk) 08:58, 7 November 2012 (UTC)
I just fixed the part about the Ensoniq OTTO. This chip was not used in the Ensoniq ESQ1/SQ80/Mirage and the Apple II GS (those featured the Ensoniq DOC, ES5503, which is an 8-bit beast with rather limited wave size and overall waveform memory addressing capabilities), but in the Ensoniq VFX and later machines. There's the intermediate DOC-II/OTIS (ES5505), which however is also more closely related to OTTO (5505) than DOC (5503) and was used in the Ensoniq EPS sampler series. 18.104.22.168 (talk) 20:13, 10 September 2013 (UTC) (RB)