Gravis Ultras

While researching 1990s sound cards with wavetable synths, I came across an interesting resource called Rich Heimlich’s Patch Set Overview, namely issue #5 from July 1995.

When I tried to unearth older issues of same, I stumbled upon a curious PDF file. It looks like an Ensoniq Soundscape manual, but it’s in fact mostly filled with hundreds of posts from, capturing a major flame war in response to the innocently named Patch Set Overview. The flame war was basically Rich Heimlich vs. the united forces of Gravis Ultrasound fandom.

Gravis UltraSound (late 1992)

With the distance of 20+ years, it’s fascinating to read. And it’s hard to escape the conclusion that Mr. Heimlich was right and the GUS fans were wrong, and often extremely rude about it (although Mr. Heimlich certainly did not hold back).

Rich Heimlich ran a games QA company called Top Star. That put him into a fairly unique position of someone who was familiar with the then-current crop of mass-market sound cards, how they sounded, and what flaws they had, especially when used with games (the #1 use for sound cards by far, at least at the time).

He had the idea of publishing a rating of popular wavetable sound cards since most people only ever heard one or two and had no basis for comparison. That sounds sensible, but on  the overview was immediately attacked by GUS fans. The responses ranged everywhere from “I don’t care what you say”, “you’re wrong”, “you’re lying”, “you don’t know anything”, to “you’re a paid shill of Creative Labs”.

Gravis UltraSound MAX (late 1994)

The discussion is fascinating. One valid criticism brought agains the overview was that there is “pure” and weighted scoring, but no explanation of the weighting. That was taken into account for the 5th issue and the weighting was better explained.

But the biggest thrust of the most incensed posts was that the whole thing is an attack on the GUS and that the GUS is unfairly rated worse than terrible, horrible, no-good Creative Labs products. No one seemed to mind that the GUS was rated lower than Roland or Ensoniq products, but rating below the AWE32 was clearly a capital offense. Mr. Heimlich did not calm the situation by claiming that the GUS was a failing product selling poorly and with a high rate of returns.

That the GUS ultimately failed is clear enough. Knowing the sales numbers would be rather interesting. Based on how rare GUS cards are nowadays (especially compared to Sound Blaster 16 and AWE32), it is obvious that GUS sold much, much worse than the Sound Blasters.

Reading the raging discussion it is apparent that Mr. Heimlich had a much better understanding of the sound card market than the ardent GUS supporters. He also knew about problems that specifically game developers had with the GUS (no MIDI interpreter, no ROM, resulting high memory requirements).

GUS and I

Back in the 1990s, I only kept hearing about the Graves Ultrasound and how great it was. I only got my hands on one around 2003 when it was long obsolete. And it was… well, underwhelming.

The mostly-software Sound Blaster emulation was a real dog, even with the latest & greatest software. The sound quality was a bit meh, not because the synth as such was bad (although the output sample frequency dropped like a rock with lots of channels) but because 1 MB just isn’t that much. And 1 MB was the maximum the GUS could be upgraded to (prior to GUS PnP); the base model came with just 256 KB RAM.

Gravis UltraSound PnP (early 1996)

In hindsight it’s obvious that the GUS came at the wrong time. It was a sound card with good potential, but the Sound Blaster and AdLib standards were already firmly established, and register compatibility with those standards was a market requirement.

Basically the GUS needed to come either earlier, heading off the Sound Blaster, or it needed to wait until a platform like Windows 95 arrived where hardware compatibility wasn’t a big problem (but at that time, it would have had to be much, much better).

It would appear that Gravis was so certain they had a winner on their hands that they dismissed the reality of Sound Blaster compatibility requirements. While there is no doubt that the GUS was an excellent tool for music composition (it was basically a sampler), that market was tiny compared to the market for a games-compatible card.

It is instructive to compare the GUS with Ensoniq boards (especially with the Ensoniq Vivo 90). Neither had Sound Blaster compatible hardware, but Ensoniq had samples in ROM (no RAM) and considerably better software providing Sound Blaster and AdLib emulation. The Vivo 90 was a fairly cheap card (like the GUS) and likely outsold the GUS by one to three orders of magnitude, in no small part thanks to numerous OEM design wins—because it much better satisfied market demand. Such is life.

This entry was posted in Creative Labs, Internet, PC history, Sound, UltraSound. Bookmark the permalink.

18 Responses to Gravis Ultras

  1. MiaM says:

    At the time Creative had to sponsor events such as The Gathering (or was it The Assembly) to force the organizers to require Sound Blaster compability for the demo competitions. Up to that, the whole demo scene were oriented around GUS.

    My impression is that up to that time the only usage of wavetable synthesis were whatever the games used and what booooooring men in suits and ties used to try to impress other boooooring men in suits that never had seen “multimedia” before.

    As I recall the GUS had hifi quality on the analogue parts and sounded good when connected to a serious hifi stereo setup or a large PA system. At the same time both Creatives own Sound Blasters and especially the SB clones had all kinds of disturbing background noise. On your pictures it’s clear that all versions hade voltage stabilisers and separate IC’s handling the analogue inputs and outputs.

    Also it seems like the GUS were the king of playing back “tracker modules” with good sound quality which was what the demo scene were interested in.

    Listen to some music (not related to the computer scene) from the time and you will hear that synthesis simulating accoustic instrumets were not a thing at that time. Who cared if the Sound Blasters could sound like a piano if you actually wanted your sound card to sound like a synthesizer?

    P.S. I’m not trying to defend the GUS, I’m more trying to express how crappy the sound blasters and clones actually sounded at the time. At the time I still were an Amiga user and I’m into high quality hifi stereo systems since long before that time. I just recall how different sound cards sounded at different friends.

  2. Michal Necasek says:

    It wasn’t called a Creative Noise Blaster for noting. Of course the sad truth is that Creative understood the market and knew that selling noisy cards didn’t matter when most users had crappy desktop speakers that wouldn’t know HiFi if it bit them in the ass. The cheap clone manufacturers knew that too, and sold even worse junk. Those who did care about quality bought something else (maybe a Turtle Beach board or later an AudioTrix).

    BTW many cards were definitely susceptible to noise inside the PC. When the PC was extra noisy (I mean EMI rather than mechanical noise), even a good card could only do so much.

    Yes, for tracker music, GUS was very good. It was much less good for MIDI because while most every MOD easily fit into the GUS’s RAM, a halfway decent General MIDI instrument bank did not. That was a problem because gamers wanted General MIDI, and gamers were the biggest market for sound cards by far.

  3. Yuhong Bao says:

    It is also unfortunate that it came just before PCI was introduced, given that it wasn’t Sound Blaster compatible anyway.

  4. Michal Necasek says:

    The Gravis Ultrasound? It showed up in 1991/1992, long before PCI was available and years before anyone (Ensoniq, 1996) dared to build a PCI sound card.

    But yes, the GUS might have benefited from PCI, and it would have been a smart design. But alas, the GUS came out at the wrong time.

  5. Yuhong Bao says:

    October 1992 was months before the Pentium was released.

  6. Thomas says:

    Hello People!
    A few thoughts regarding “got my hands on one around 2003 when it was long obsolete. And it was… well, underwhelming”. It is important to view it in it’s time context.
    The top capability of GUS was the playback of music using sampled sounds, doing this without speed penalty on the CPU … this greatness was gone just a few years later, you could have music with sampled sounds downmixed and played back through any cheap SB clone without any wavetable capabilities and most important without great speed penalties on the CPU.
    What made GUS great could be done on any sound card and that killed the GUS (I think).

    About GUS competition, indeed there were other (more or less expensive) cards with General Midi …
    For in-game music, if you want Mozart, GM is unbeatable but if you want that era’s techno/rock style music, standard GM instruments set is pathetic. Starting in 80’es throughout 90’es it was the time of the samplers, you cannot play that music with a symphonic orchestra instruments set.

    Back in 90’es I was ecstatic about my SB clone with OPL synth until I’ve seen (and heard) an Amiga 500 … from that moment on, for me game music on PC (being Adlib or GM) sounded from mediocre to horrible. And Amiga was not even half the GUS.
    In the PC world GUS was the only thing that brought the magic of Amiga … unfortunately a little bit late. Had Gravis done it when the 386 came to market, it would have been a blast.

  7. Michal Necasek says:

    I’m pretty sure there were PC MOD players with software mixing in the early 1990s. Not the same as the GUS, but not nothing. The huge problem that GUS had was lack of content. For SoundBlaster-oriented games the GUS was overkill with sub-par compatibility, for General MIDI it was not that great due to limited RAM and steep frequency drop-off with more channels. For trackers the GUS was terrific, but that was never a mainstream thing. Yes, the GUS was an okay sampler, but a sampler isn’t something every PC user would want.

    I disagree about General MIDI and techno/rock music. One just needs to listen to the soundtracks of games like DOOM or System Shock.

    In a way the UltraSound was like dedicated MPEG decoder cards, great when they came out but soon completely obsoleted by commodity CPUs and graphics chips.

    You have a good point that things are relative. Compared to AdLib, the Sound Blaster was brilliant, and compared to the old Sound Blaster, GUS was brilliant too. I also think that had GUS showed up about 2 years earlier, things could have gone very differently.

  8. Renee Senger says:

    Reading about these vintage ISA sound cards I remenber Crystalizer Tidalwave 128, a Crystal C59236 chip based board. At that time it was considered as a good option for OS/2 too. I never found one to test. Anyone here tested this card?

  9. Michal Necasek says:

    No, would love to have a Crystallizer myself. But finding a card with a CS9236 synth should be a lot easier than finding a Crystallizer card. I have an IBM OEM ISA card, S-16WP1/L, which has CS4237B and CS9236 chips on it (and very little else — these were really highly integrated chips).

  10. Pingback: Trackers vs MIDI | Scali's OpenBlog™

  11. Pingback: How Many Gravis UltraSounds? | OS/2 Museum

  12. Nathan Anderson says:

    I wonder what kinds of games y’all were playing back then? I never had the pleasure (?) / misfortune (??) of owning a GUS, but I *definitely* remember that a whole slew of games from shareware publishers such as Apogee and Epic *heavily* advertised the GUS capabilities in their games, and I remember thinking back then (as an impressionable young thing) that it must be the audio card to have…

    Also, not a judgment call either way on the merits or faults of the GUS, but reading that linked Usenet thread took me back to some comp.os.os2.advocacy flamewars from way back in the day. The, uh, “zealotry” on display here by GUS fanatics definitely calls to mind the same kind of unhinged and emotionally-based defenses that I remember from the OS/2 (and even Apple) camps back in the day which just caused me to, as the kids these days say, “SMH”. (And I was an OS/2 fan!)

  13. Michal Necasek says:

    GUS was not that great for games (OK, I’m not a GUS zealot). It was a capable card and it was great for demos and trackers and for the games that took advantage of it (mixing lots of channels with very low CPU overhead), but as a general gaming card it was not very good, which is what killed it. The #1 problem GUS had was Sound Blaster.

    GUS, Amiga, OS/2, there’s definitely some commonality there.

  14. Michal, thanks for this informative post. Those were the days and that debate certainly felt like a war of science versus feeling.

    Being “right” about this was not difficult if one had the details at their disposal. As a professional in the industry at the time, I did. Consumers had their feelings, but those were not backed by data. The GUS was simply the wrong product at the wrong time and Advanced Gravis knew this and shared that fact with me (a fact that ended up with my being deposed by both sides in a legal case of their stockholders against management). It was, however, the horse they rode to town in and they were going to ride it as far as it could take them. Little did they suspect that it would ultimately sink their entire company.

    My main issue with the company was a willful campaign to lie to consumers about support I knew (and the publishers knew) was never coming. They knowingly put out lists of products with upcoming support and many of those were products we were retained to test. We knew they had no such support months before they shipped or else we’d have been required to have tested such support. Gravis decided that if they lied enough about it the industry would be shamed into supporting them. They were wrong and the impact was fatal.

    The Ultrasound shares much history with the Amiga. It’s a product of what if’s but not really of what was. I wrote the Official Sound Blaster book series, not because I preferred Creative or because it was a technically superior product, but because it was the most successful product of its time.

    There’s also a recent documentary on all of this called Beep. It came out last year and covers the history of audio in the gaming.

    For what it’s worth, I was also a huge OS/2 fan myself and had a very personal, and intriguing experience with IBM’s top brass that you may find interesting. It’s covered in the webisode of my interview on the Beep website (

  15. Michal Necasek says:

    Rich, thanks for coming out of the woodwork 🙂 Now I need to look at that Beep thing, especially because I already have the earlier book by Karen Collins and found it very informative, if not nearly as detailed as I would have liked it to be.

    Not long ago I bought a copy of the Official Sound Blaster book and I can confirm that it’s not any kind of fanboy thing, it’s almost entirely a purely technical description of the cards, what they can do, and what the software does, with no qualitative judgments to speak of. Just stuff along the lines of “it’s better than AdLib” and “it sells well” which is/was difficult to dispute. It’s also obvious that a similar book about the GUS would have had to cost almost as much as the card itself and not $29.95 🙂

    From the flame wars it was painfully clear that too many people could not make the distinction between technically superior (for some definition of that) and commercially successful. Perhaps the geeky soul wants the technically superior to succeed and refuses to believe that the inferior product could possibly sell better.

    So now I understand better why you were critical of Gravis; you had access to non-public information (game betas) which proved that the company was deliberately misleading users. That was very short-sighted of them.

    There’s definitely a lot of similarity between the Amiga and GUS, but I’d say the Amiga was considerably more successful. They stayed in business a lot longer and built a decent platform, but I’m sure a lot of the same things that killed Gravis also killed the Amiga. I’m by no means an Amiga expert, I never owned one and mostly watched it from a distance. What I clearly remember is that the Amiga had a vocal fan base with very strong feelings.

  16. On the first book, I actually had to annoy Creative by sneaking the appendix in after deadline. They were adamant that no developer info should be in it and I was just as adamant that the book had to address developers. I had the right to include that info legally and did so. They responded by illegally not holding up a key component of their end of the contract on the book and it cost me a rather large sum of money. So, if I’m a Creative shill, I have a funny way of showing that loyalty. hehe

    Regarding technical superiority, I have little doubt that I said this many times in those posts of long ago, but it’s as true today as then: technically superiority often has little to do with what products gain the upper hand. Beta was superior to VHS, Amiga was superior to the PC, etc., etc., etc. Creative was a master of marketing beyond everything else they did. Their products were just good enough to allow for exceptional marketing to carry the day.

    I have also taken great pains to correct mythology that tends to settle in now and again in the intervening decades. People continue to put forth their recollections over the facts time and again. I see posts that continue to suggest that the Sound Blaster succeeded because it also sported digital sound support. Flat out wrong. That part of the card wasn’t even touched for more than a year after they trounced AdLib. It won because it was better marketed, was slightly cheaper and included a game port which, at the time, saved you $50 and an expansion slot which was even more valuable for any clone (which often had just three). The Sound Blaster was a no-brainer purchase.

    I’m still shocked someone took the time to put some of that discussion in a PDF… Wow.

  17. Pingback: Rich Heimlich’s Patch Set Overview | OS/2 Museum

  18. MiaM says:

    TL;DR Amiga went down due to a bad in-house semiconductor manufacurer, and the lesson from using Amigas is that everything in computers still sucks.

    As we anyway are on an Amiga side track…

    I believe that what actually both made Commodore rise and Commodore/Amiga fall was that they bought MOS Technology and that MOS fell behind in the later half of the 80’s.

    In mid 70’s up to mid 80’s it was really good for Commodore to own their semiconductor manufacturer with a steady supply of VLSI IC:s at a known and low price. They could sell the VIC-20 and C64 at prices lower than most other computers, and they did (as roumours say mainly to get even at Texas Instruments who previously had started selling their own calculators at a lower price than what Commodore had to pay for the chips), killing of the Texas TI99/4A and most other home computers. (Afaik Ataris and Tandys 8-bit computers survived a few years after the big war in 83, and in Europe Sinclair Spectrum survived a decade, and in the UK their Acorn BBC computer survived. But most other were killed in the wars).

    When Commodore bought Amiga Inc they were able to manufacture the Amiga chipset at MOS Technologies. Some slight improvements were made over the first years. From a hardware point of view the most important of theese improvements were that the dma controller “Agnus” moved from a DIL to a PLCC package (second version of A2000 and A500 and later CDTV used that) (perhaps MOS first PLCC chip?) with more pins and more logic inside, and later capable of adressing more memory (later A500, later A2000, (later CDTV?), A500+, A600). Other improvements were really minor, like there were already DMA for up to 6 bitplanes but only colour registers for 5 bitplanes in the ramdac (called “Denise”) so they just added a feature where the 6th bitplane halved the brightness, called “Extra Halfbright”. (The reason for having DMA for 6 bitplanes but a ramdac only capable of 4 bitplanes was that from the beginning there were a special mode called hold and modify where two bits selected if the other four bits should either use ramdac register 0-15, or load a 4-bit value in one of the RGB channels (and specifying which channel)). The only major IC Commodore had do buy externally where the 68000 CPU.

    But when it was time for the only major overhaul that made it into production, the “AGA” chipset (A1200, A4000) with 32 instead of 16 databits, MOS just couldn’t make advanced enough chips so Commodore had to let out the production to other manufactureres (for example HP iirc). I don’t know how and where the construction of the chips were made, perhaps it was a cooperation between HP and MOS, or perhaps HP made them to specs from Commodore directly. Anyhow, that made Commodore dependant on external manufacturers who of course wanted to make a profit. And since MOS couldn’t manufacture better chips, Commodore saw no reason to rework the IC that handled for example audio and floppy, which IMHO was a mistake. It could still only handle DD disks with normal diskdrives, and the sound were 8-bit, although 4 channels with separate volume control for each channel making it in practice possible to play back an equalient to 14 bits in stereo. And also the audio hardware were limited to two samples per scan line, so when using TV resoultions (15kHz, almost all games used that) you could only achieve about 31kHz sampling frequency. To play back 44,1kHz “cd quality” audio you had to put the graphics hardware in a VGA compatible mode. I believe that if they had made all chips in house they would have made a newer version of that chip too, as double bandwith of the DMA were available with a 32-bit instead of 16-bit data bus.

    Btw, between this the A3000 were released. It had a 32-bit data bus for all cpu and external bus master access, but the chipset were still 16-bit. To gain performance there were multiplexers so that the memory shared between the chipset and the CPU could be accessed in 32-bit chunks by the CPU and 16-bit chunks by the chipset. That had the silly effect that moving blocks of data were faster by using the CPU than using the dedicated blitter hardware in the chipset…

    Commodores last effort to save the company were the CD32 gaming console which was kind of an A1200 without a floppy controller (and supplied keyboard and mouse althoug ports for them were available) but with a cd rom. A sign of how bad their economy was is that the power swich for the CD32 were the same that were produced for C64, and that is known to fail in CD as the switch can’t cope with the inrush when switching on the 12V DC line. (On C64 it would switch 5V DC and 9V AC and the later with far less current available).

    At the end MOS was just an environment disaster waiting to happend, with waste tanks on their way to start to leak.

    Although you could attribute market and other to Commodores and Amigas successes and disasters, IMHO it was ultimately the IC production and it’s lack of improvement that staked out Commodores path. Of course the “toy” image was a problem in general business markets, but even a “serious” company like Apple did struggle in the 90’s when the PC market finally caught up on the nische competition.

    P.S. I would say that the biggest lession I’ve learned from using Amigas for more than a decade is that almost every major operating system sucks in some way, and many parts of modern hardware suck too.

    I still struggle to understand why a modern OS seem to run for example the file system code at a higher or equal priority to the GUI code. AmigaOS only ran short interrupt handlers at higher priority than the GUI. Even if an appliaction itself were slow, atleast the gui elements would respond immediatly showing that your mouse click actually made it all the way to the computer. I beleive that the lack of such GUI performance in other OS:es is the reason why mice for most other computers use noisy micro switchs for the mouse buttons – the noisy click sound and the tactile feeling tells the user that the click actually took place. On the Amiga anything equal were never required, you saw with your eyes that the click took place.

    Although the hardware and thus the software didn’t support memory protection, we must remember that the OS had all other modern features even back in 85. Everything that didn’t have a hardware limit were only limited by the amout of memory (two kinds, one shared with the video/audio/io hardware, one only available to the CPU and expansion cards using DMA). Everything you could allocate were in linked lists that coud expand indefinitely (except for how much memory you had). Interprocess communication took place by passing messages, not only for the GUI but for all parts of the OS (although not as visible to the casual programmer). Expansion cards were automatically configurated by a combination of software and hardware. There were no conflicts between different cards. If you temprarily wanted to add an extra hard disk and didn’t jhave a cable with enough connectors it was perfectly normal to add an extra disk controller as that would always work flawless.

    Sorry for posting a wall of text like this, but I think it’s worth knowing a bit about what Amiga was 🙂

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.