The question came up a while ago. Just how commercially unsuccessful was the Gravis UltraSound? There appears to be no public information about the sales volume of the UltraSound. But now, looking at a sample of 3 (three) classic GUS cards, it occurs to me: Could the serial number be exactly that? I know that’s extrapolating a lot from very little.
Lots of folks put pictures of their GUSes on the Internet. Not too many show the reverse side where the S/N sticker is usually located. So let’s start with mine:
- GUS 2.4, PCB week 47/92, S/N: K33579
- GUS 3.7, PCB week 50/93, S/N: K103633
- GUS 3.73, PCB week 16/94, S/N: K119146
If the serial number is a true serial number, it would mean there were somewhere upwards of 120,000 GUS Classics made, perhaps around 150,000 (in 1994, the GUS MAX took over). The 2.4 is an early revision (2.2 was the first production rev), so it has a relatively low serial number. The 3.7 and 3.73 are fairly close to each other in age so makes sense their serial numbers wouldn’t be very far apart. The 2.4 and 3.7 are a year apart so there’d be a much bigger gap.
GUS MAX does not appear to have a real serial number. The S/N looks to be a production week/year (mine says “Serial Number 0695”).
What serial number does your GUS Classic have? With enough data, it should become apparent if the serial number looks like a simple counter incremented for each card, and it should be possible to estimate how many there were. Please list the version, the date code etched on the reverse of the PCB, the serial number, and other identifying information if it seems relevant.
Update: Older GUS MAX cards appear to have six- or possibly five-digit serial numbers in the same format as GUS Classic cards.
Update: And here is the complete list so far:
- GUS 2.4, GF1 43/92, S/N: K15404 (batyra.pl)
- GUS 2.4, GF1 43/92, S/N: K17728 (vogons.org)
- GUS 2.4, GF1 41/92, S/N: K20617 (eBay Kleinanzeigen)
- GUS 2.4, GF1 47/92, S/N: K20695 (eBay Kleinanzeigen)
- GUS 2.4, GF1 47/92, S/N: K24466 (eBay listing)
- GUS 2.4, GF1 week unknown, S/N: K25871 (RK)
- GUS 2.4, GF1 week 47/92, S/N: K33400 (eBay listing)
- GUS 2.4, GF1 week 47/92, S/N: K33579
- GUS 2.4, GF1 week 47/92, S/N: K39582 (hardforum.com)
- GUS 3.4, GF1 week 45/92 , S/N: K48027 (reddit)
- GUS 3.4, GF1 week 45/92, S/N: K48246 (auction listing)
- GUS 3.4, GF1 week 51/92, S/N: K53728 (dosdays.co.uk)
- GUS 3.4, GF1 week 51/92, S/N: K55883 (auction listing)
- GUS 3.4, GF1 week 51/92, S/N: K56240 (auction listing)
- GUS 3.4, GF1 week 45/92 , S/N: K57605 (wavetable.nl)
- GUS 3.4, GF1 week 47/92, S/N: K60420 (auction listing)
- GUS 3.4, GF1 week 47/92, S/N: K60624 (retronn.de)
- GUS 3.4, PCB week 36/93, S/N: K67727 (eBay listing)
- GUS 3.4, GF1 week 25/93, S/N: K68146 (oboroc.com)
- GUS 3.4, GF1 week 25/93?, S/N: K71451 (audiofanzine.com)
- GUS 3.4, GF1 week 25/93, S/N: K71891 (vogons.org)
- GUS MAX 1.7, CS4231 week 29/93, S/N: K102816 (phantom.sannata.org)
- GUS 3.7, PCB week 50/93, S/N: K103633
- GUS MAX 1.8, CS4231 week 12/94, S/N: K104862 (twitter/lazygamereviews)
- GUS MAX 1.8, CS4231 week 12/94, S/N: K106776 (eBay listing)
- GUS MAX 1.8?, GF1 week 02/94, S/N: K107427 (Wikipedia)
- GUS MAX 1.8, CS4231 week 14/94, S/N: K110046 (auction listing)
- GUS MAX 1.8, CS4231 week 29/94, S/N: K114492 (imgur.com)
- GUS MAX 1.8, CS4231 week 15/94, S/N: K116188 (linustechtips.com)
- GUS 3.73, PCB week 16/94, S/N: K119146
- GUS MAX 1.8, CS4231 week 34/94, S/N: K120306 (vccollection.ru)
- GUS 3.74, GF1 week 52/93, SN: K125809 (vcfed.org)
- GUS MAX 1.8, CS4231 week 34/94, S/N: K126321 (oldschooldaw.com)
- GUS MAX 1.8, CS4231 week 35/94, S/N: K128008 (twitter/lazygamereviews)
So far the theory is holding, but more data is needed.
Update: Some amount of serial numbers was almost certainly skipped around the beginning of 1994. Without doing a statistical analysis, I believe that with 25+ samples, finding nothing at all in the 75k-100k range is highly improbable.
On Gravis UltraSound and PCI, I wonder how much the lack of this was the PCI SIG’s fault (for not realizing early on that they needed PCI sound cards).
I think it was the other way around, the sound card vendors didn’t think they needed PCI. Sound cards were one of the last ISA hold-outs, with ISA sound cards being manufactured in volume until 2000 or so. If Intel didn’t kill ISA, that probably would have kept going even longer.
It’s hard to say how much the PCI SIG was at fault, but you are right that the fact that sound cards were ignored in the PCI specs made it sooo much harder to make PCI sound cards later on. Ensoniq and others had to jump through all kinds of hoops to get Sound Blaster compatibility on PCI. Heck, even ISA PnP is terrible with sound cards and vendors put a lot of effort into effectively defeating PnP.
For the UltraSound, PCI came too late anyway. By the time PCI went mainstream, the GUS was effectively gone.
The point would be to get a PCI version of UltraSound early on when PCI was new so that games can support it.
Yeah, and why would Gravis care about something only a few people would buy? By the time PCI was widespread, Windows 95 was unleashed and everything changed anyway.
Of course, DOS gaming was still common after Win95 was released. It was in 1997 that Intel and Microsoft would start trying to get rid of ISA and hopefully games would have been supporting the PCI version of for example UltraSound for a few years by then. Of course, it would be likely that Gravis would not be the only one in the PCI sound card market.
I think that more than the lack of ISA connectors (a lot of chipsets supported ISA internally long time after) what changed the game for the sound cards were the arrival of Windows 95, games being developed for Windows instead of DOS (Sound Blaster Compatibility) and the release of AC’97.
Don’t get me wrong, I think the GUS would have been a great fit for PCI. But the timeline just doesn’t work at all.
But it reinforces my thought that the GUS came at the wrong time, both too late (to beat the Sound Blaster) and too early (to take advantage of Windows 9x or PCI).
You mentioned the example of Sound Blaster eventually removing the C/MS chips before. This kind of process generally takes years.
I like the way PC-98s handled sound PnP. NEC defined I/O port A460H as a “sound hardware identifications register”, which explicitly denotes which capabilities are present (YM2203 OPN or YM2608 OPNA FM synth, PCM support, etc)
PCI was a rough transition for sound cards simply because the SoundBlaster “standard” relied on classic ISA DMA to perform digital audio playback Emulating ISA DMA over the PCI bus wasn’t too reliable or easy to do. PCMCIA (and other exotic devices like parallel port sound devices) had similar problems with SoundBlaster compatibility for the same reason. Heck, even though classic DMA was present, it might also be a reason why Microchannel sound cards were rare. Creative’s offerings relied on jumpers still even though they should have been using programmable option select to soft configure the cards.
I think the lack of MCA cards was not a technical problem but simply an issue of insufficient demand. PS/2 machines were typically used in business settings where audio was not a priority.
As for PCI, I think there were two separate problems. One was ISA DMA, which had several standardized solutions (DDMA, PC/PCI) and a number of home-brew vendor-specific solutions. Mose of these worked pretty well. The much nastier problem was resource allocation, and especially interrupts. That’s a problem which the PCI SIG solved for VGA and IDE early on, but SB-compatible audio was either rejected or (more likely IMO) not considered at all. And in 1996 it was really too late to think about that. ISA PnP cards were similarly problematic but vendors were able to solve it in various ways; for PCI there was no sure-fire solution.
My experience with PCI sound cards is that Sound Blaster compatibility is not worth the effort on boards which have ISA slots, and a lost cause on newer systems (for example if PCI slots are not on bus 0, forget it). There is only a relatively narrow range of boards from the early 2000s which had no ISA slots but SB-compatible PCI cards worked.
I’ve tested well over a hundred various ISA and PCI sound cards and my experience is as follows: Manually configured ISA will always work. ISA PnP is anywhere between working great and a complete disaster, depending on how the card implements it. PCI is a horror and much depends on the motherboard and chipset; a given card + board combo will either work well or not at all.
Which is why it would have been so important to do PCI sound cards early on when PCI when new so we would be prepared when we got rid of ISA years later.
I think part of the problem was that 8237-style DMA was considered highly x86-specific.
I believe Gravis was a company in panic, and they weren’t as big as they appeared with individual engineers making board designs around common chips, with each rev introducing fantastic incompatibilities, and errors like Left/Right swap issues.
As mentioned above the real killer was PCI. For whatever reason having PCI take over or bridge into ISA space was seen as something that didn’t need to happen, so we never could get a PCI SoundBlaster card. Like it or hate it, the SoundBlaster defined what gaming on the PC was. There were a few PCI cards that had “SoundBlaster” compatibility, however they may have been register compatible at a PCI address space, they used EMM386’s v86 mode to re-map the IO onto the ISA space, and do the ISA DMA through the VM.
Although WinG was ultimately a disaster, DirectX proved to be the point where Game developers could finally get something better than 320x200x8 video and 8bit mono sound without going crazy over ‘standards’. And we all remember how iD felt about Paul Radek & DMX.
So do you have some GUS serial number or not? 🙂
To me it looks like there were somewhere under 150k GUS classics made in 1992-1994. By the end of 1992 or so, there were already over 1 million Sound Blasters out there (says Sound Blaster: The Official Book), and who knows how many clones. So the general reluctance of ISVs to spend effort on GUS support is quite understandable.
There were some PCI cards that were completely SB compatible with no need for EMM386 or similar, but may have required PC/PCI or the right chipset or had other restrictions. There was clearly a lot of effort spent on making SB-compatible PCI sound cards, and PCI really made it difficult. I still think it’s because back in ’93 or so, no one needed a PCI sound card. By the time PCI sound cards became a useful thing to have, it was too late to define a standard.
Not to mention that it was a totally brain-damaged design that was perhaps good for an 8-bit CPU but not a PC. 64K boundary problems, 16MB addressing restrictions, poor performance, no wonder the PCI SIG didn’t want to touch that with a 10-foot pole.
Sales volume can be estimated from the reports in Gravis’s Form 20-F filings to the SEC. The 1996 filing was the only one I could locate. Page 6 lists sound card sales by year as being $5.9 million in 1994, $10.2 million in 1995, and $8.8 million in 1996. Sales in 1993 and 1992 should be considerably smaller and I think Kensington rapidly killed the Gravis sound card lineup after buying Gravis in 1997.
Interesting notes: Almost $2 million had to be written off because of the failure of Phoenix joystick and Ultrasound Ace which was part of $8 million loss in 1996.
IMHO Microsoft should have made a SB emulator for DOS apps as a part of Win95, and maybe do it as some kind of joint effort with the existing and upcomping sound chip makers. Like an additional part of PC-98 but available earlier than 98.
But it seems like Microsoft is really afraid of scaring away any hardware manufacturer by making it simpler for other manufacturers to do what has been hard to to for the existing manufacturers.
That’s a good source of information. Clearly in ’96 Gravis was on its last legs. But as for the shipment estimates, the UltraSound MSRP was “$119.95 to $199.95” as noted in the SEC filing. How much does it translate to as a unit sale price in terms of the total revenue? $100? $150? Something else?
And yes, the filing has the total sales for previous years, $23.7M in ’94, $13.9M in ’93, and $5.8M in ’92. Assuming the sound card revenue percentage held steady at 20-25% or less, let’s estimate that the sound card sales were around $3M in ’93 and around $1M in ’92.
Also interesting that the given the R&D figures, Gravis must have had a very small development team.
Anyway, assuming Gravis was selling the cards for around $100 apiece, I think that pretty well matches my estimate of around 150,000 or fewer GUS Classics made. So the serial numbers probably are real serial numbers, or something very close to it.
The GUS MAX probably sold a bit less, and the PnP much less, although there were 3rd party manufacturers of GUS PnP compatibles (using the AMD InterWave chip). Compared to the sales of Sound Blaster alone (millions of cards), it was not a lot, and compared to the sales of all SB compatibles it was nothing.
That sounds like something which would have been good for Microsoft but much less good for companies like Creative, greatly diminishing the value of providing SB-compatible hardware. Microsoft had no experience in that area (the WSS was not SB compatible at all) and the people who did might have been very disinclined to help Microsoft. There were other companies like Ensoniq who had the know-how but again probably wouldn’t just give it to Microsoft. And Microsoft probably wasn’t too keen on paying for it when their story line was WinG/DirectX. So… yes, it would have been great for users but probably could never happen 🙂
Didn’t Microsoft add some SoundBlaster support to NTVDM in Windows XP? I know it wasn’t perfect, otherwise VDMSound wouldn’t have been developed. I know it came in handy for MPU-401 emulation and my Roland MT-32. Out of the box, the Soundblaster Live! couldn’t run any external MIDI devices from a Windows 9x DOS box or using the DOS emulator.
I think there was something, but arguably at a point in time when it didn’t matter anymore because SB-compatible hardware didn’t really exist.
I thought that horrible”Microsoft business sound card” thing had a sound blaster emulation mode for Windows 95… I can vaguely recall playing DooM on our work Compaq Pentiums and actually having sound…
Microsoft were big enough to be able to buy a company that has the right knowledge. But how hard could it have been to make the dos box emulate SB? The only hard thing about this would have been the decision to actually hide more of the real hardware from the dos applications, and decide where to draw this line. Perhaps they kind of wanted there to be a big enough difference between Win95 and NT and thus not only hid all hardware in the dos box in NT but also wanted to expose all hardware in Win9x?
(As a side note there are third party aftermarket hacks to be able to actually access some hardware from DOS applications in NT. I use one of those for running the DOS application for an old ALL-03 eprom programmer on a XP box).
I think they were mainly afraid of being known as a company who at random destroys the market for existing hardware companies.
But they kind of did this for all software companies who focused on DOS applications, and they didn’t get any bad reputation for that. Windows kind of killed most DOS office applications and almost all DOS utilities.
I really don’t see a big conflict between this for existing games and DirectX for new games, except that it might had slowed down the transition somewhat.
Do you perhaps mean “Compaq business sound card”? Quite possible that they had some Windows-based SB emulation. Microsoft only sold Windows Sound System and that was definitely not SB compatible, and I think it was basically meant as a reference design rather than a product.
Sure, Microsoft can buy some random company. That doesn’t mean they want to.
And yes, Microsoft was wary of upsetting OEMs, because they knew they depended on them. They were much less worried about upsetting ISVs.
The WSS Wikipedia entry indicates that although the hardware was definitely not SB-compatible, MS did provide SB & AdLib emulation in software for WSS owners; it was something built into the software drivers. Although I never owned a WSS myself, I do also recall seeing a demo from waaaaaay back when of this working on a machine equipped with a WSS card. I want to say that you actually had to pick a second set of I/O / IRQ / DMA addresses for SB compatibility mode that were masked/intercepted by the software driver, separate from the ones used by the WSS for native access to it. So similar in practice (from the user’s perspective) to how MediaVision PAS16’s SB emulation worked, except the SB implementation was 100% in software, rather than (as in MV’s case) essentially having 2 hardware sound interfaces on a single board.
It’s probably also worth mentioning that the WSS was still an ISA device, and since I don’t know 100% how the SB emulation worked under the hood, it’s possible that the driver still relied somehow on a card under its control being present on the ISA bus in order to intercept calls being made to the phantom “SB” hardware.
Re: GUS MAX serials, other than the different format which could be interpreted as a date, how do you know it is a date? Have you run into MAX cards with non-unique numbers (e.g. another serial “0695”)?
The problem with the idea of having SB emulation on Windows 95’s DOS box is that it would mean that many users would need to have their software configured differently depending on whether they were running it under Windows or in full “MS-DOS Mode” (i.e. users without a different non-SB sound card like a Gravis UltraSound).
Since games were the primary users of sound hardware on DOS and gamers and game developers were both initially very wary of Windows 95, having a situation where the game’s sound doesn’t work (assuming the game doesn’t just crash) when run under Windows (due to it being configured for the real non-SB hardware) would only have harmed Windows 95’s reputation for gaming.
Yea, the OS/2 2.0 debacle probably don’t help here, especially as it was part of what made DOS gaming last longer than it should in the first place.
Re GUS MAX serials: 4 digits is just not enough. So far I haven’t seen one where the serial number couldn’t be a date, but I don’t have enough data.
Re WSS, I guess I need to dust off the original Windows Sound System 2.0 disks I have and see what that card really does. It does look like there’s supposed to be some kind of Sound Blaster emulation, but I suspect that it’s nothing to write home about (at minimum restrictive, requiring EMM386, and apparently a specific EMM386 version). There would be no AdLib emulation needed because WSS did come with an OPL3 chip.
Two hardware interfaces on a single board was essentially standard for most Sound Blaster clones circa 1993-1997. Typically they would offer Sound Blaster as well as WSS compatibility (OPTi, Crystal, ESS, many others). SB Pro compatibility was 8-bit only, designed for games, and WSS compatibility was 16-bit, with higher sampling rates, designed for everything else.
I think it is probably based on http://www.delorie.com/djgpp/doc/rbinter/id/69/48.html
Very probably. “WSSXLAT requires EMM386.EXE version 4.47 or later. No other memory managers are supported.” (from MS’s Q121259). The bit about no other memory managers being supported would have been quite a drawback.
Super:
It’s probably now that I have to confess that I’ve never used a sound card under dos (except for maybe running the manufacturers software for some configuration thing or so). (I come from the Amiga world).
I assumed that if you had those “SET BLASTER=…” stuff the games would use a SB with those parameters, and if you didn’t have those parameters set the game would not try to use sound.
Requiring Win95 to be started to be able to have sound in existing games could perhaps actually been a way to push game developers into the Windows world. Especially if there were an easy way to use the Win32 API even from a dos app (i.e. recompile/reassemble the game with some changes to run as some kind of full screen win32 “with hardware graphics compatibility” app as some intermediate solution.
—-
Re MV and other sound cards:
I’ve used MV PAS 16 in Linux. The card turns up as two devices, one capable of playing 16 bit sound and one capable of playing 8 bit sound, with different sliders in the mixer. My memory is that MV PAS 16 did sound rather goot, but not really great. I replaced mine with one of those Turtle Beach cards that were 20 bit and could set the hardware sample rate at 44,1 and 48kHz. That really sounded great, as far as I can tell about as good as my current M-Audio Audiophile 24-96 card.
(Sorry for going off-topic, but after all we are discussing pc sound cards 🙂 )
It was much more complicated than SET BLASTER. Some games needed that, others didn’t care at all. Some supported just Sound Blaster, others supported a variety of cards in their native mode. Some supported a wide variety of resource configurations (ports, IRQs, DMA channels), others very few. Sound was like graphics in that every game did things differently, but much, much worse because there was no equivalent of VBE and Sound Blaster was barely a VGA equivalent. It didn’t help that VGA was well established when DOS gaming got really going, but audio was evolving more or less the whole time (AdLib/OPL2 – 1987, SB – 1989, SB Pro – 1991, OPL3 and SB 16 – 1992, WSS – 1992, not even mentioning MT-32 and General MIDI).
Did you have some TB MultiSound Fiji or similar card?
What I remember very well as a gamer in the 1990s is that Windows was poison. Too many DOS games ran worse or not at all. Configuring everything properly was not trivial, and Windows just made everything even harder. RAM was a significant factor too, with games requiring 4 or 8 MB and if you had that much RAM and Windows loaded, there wasn’t enough memory left for the game.
It also didn’t help that some game developer’s coding intentionally prevented their games from running under Windows. Source: Raymond Chen’s “The Old New Thing”
As for WSS, Gravis actually used a WSS compatible codec on the Ultrasound Max (Crystal CS4231), but apparently didn’t use WSS standard ports, so the generic drivers wouldn’t work.
Oh, that sounds like a real night mare 🙁
(unintentional pun)
Yes, it must be the Fiji card that I used. (I still have it in my last computer ISA based computer, but don’t use it any more).
P.S. my experience of graphic cards on PC’s were more like that you had to swap bios between EGA cards because some bios worked on XT’s and 286’s but not 386’s while some other bios worked on 286’s and 386 but not on XT’s. It seems like it were far worse with sound than with video…
Game developers of course didn’t want to deal with Windows, because they knew that for DOS games, Windows didn’t solve any problems but created more than a few. So that’s understandable (and it’s understandable that Microsoft didn’t like that).
Especially when many of the problems was not difficult to fix like using proper DPMI functions for virtual interrupts.
A bit late to this party, but I’d just like to toss in that MCA soundcards were less than common for two other reasons not covered above–
1. There was a HUGE backlash against the PS/2. A lot of people saw that platform as IBM’s attempt to take control of the market again after losing it in the 80s.
2. For whatever reason (likely multiple reasons), MCA Sound Blasters were typically $100 more than the ISA version. Indeed, MOST cards were at least $100 more expensive on MCA, and that definitely didn’t help the platform.
It wouldn’t be too much longer before VLB became a thing, and despite VLB’s serious flaws, it was still a lot cheaper to work with than anything IBM was putting out.
Theoretically there were about five years between MCA and VLB, 1987 to 1992. But for the first 3-4 years there were no sound cards to speak of, so it didn’t matter 🙂 I think the price differential was to a lesser extent because MCA was more complex and to a much bigger extent because the sales volume was much smaller.
Best of all, in a VLB system with fast graphics and storage, you could still use the same old ISA sound card. In fact I don’t think I’ve ever seen a VLB sound card.
Trivia: Did you know that VLB uses the 16-bit MCA connector? MCA wasn’t selling much so the connectors were readily available and cheap 😀
Wasn’t aware of the VLB using the MCA connector, actually! That’s an interesting footnote.
As for the soundcard thing, I’m going to have to *partially* argue that point. I ran into my first soundcard advocate around mid-1987. A local computer store was pushing the Creative Music System pretty heavily and ran demos in the store. That card would later be renamed to the Game Blaster, but Adlib had a head start in the third party support that forced Creative Labs to take a completely different angle with the Sound Blaster. I find that particularly interesting because the CMS was actually stereo (unlike Adlib or SB pre-Pro) and had superior audio quality to the Adlib..
I ran into a MCA Sound Blaster card in a Babbage’s around late 1991. They were there, but they weren’t selling well at all. They had the Sound Blaster 2.0 for sale for $130 and the MCA version was going for $250.
Yes, the AdLib came out in ’87 but really, sound cards IMO became a big thing only around 1990 after the Sound Blaster came out and when CD-ROMs became more widespread.
The CMS was indeed stereo but opinions differ on whether the audio was higher quality than the AdLib (Yamaha OPL2). The OPL2’s FM synthesis was fairly advanced for the time, definitely more capable than the (two) Philips SAA-1099 on the CMS/Game Blaster.
A Sound Blaster 2.0 at $250 sounds like a rip-off 🙂 What it comes down to is that completely removing ISA slots from PS/2 (MCA) machines turned out to be a big mistake which made PS/2s even more expensive than they already were, and more difficult to use. Technically ISA was crap compared to MCA (and IBM knew that better than anyone else), but in reality ISA was with us until about 2000 and supported for years after that, simply because for some/many devices ISA was perfectly adequate.
I think PCI used a similar if not the same connector too.
As a sidenote, while we’re at it: Not only did VL use one of the MCA connectors. Another MCA connector were used for the expansion port on Commodore Amiga CD32.
For that particular reason I have saved an old PS/2 motherboard with MCA connectors. The long term plan was to try to unsolder the relevant connectors (with a gas burner or similar on the solder side).
Gravis was way too smart to be so public about their numbers. IF this number is truly sequential they would have no doubt started it at some number well beyond zero and likely jumped numbers between versions. See if anyone can produce a card with serial numbers below 20,000 or 10,000. None of this addresses the bigger issue of how often Gravis had to buy back their inventory.
As noted elsewhere, Electronics Boutique brass shared their figures with me back in the day and at that point they were the biggest single buyer of sound cards out there. The most telling thing they shared was that the Ultrasound was the most returned product in their history up to that point. They finally had to inform Gravis that they would no longer carry the product in their stores due to the overwhelming number of dissatisfied customers that returned it time and again. They’d keep repackaging them and reshinking them (as was common back then) and the cards would just keep coming back. District managers told me it was a product they knew all too well as store managers were complaining about having to even stock them.
This was all due to simple expectations that were wrongly set by Advanced Gravis themselves. Mainstream consumers thought they were buying a Sound Blaster for less than the real thing, and with some better features. Remember, these buyers had not that long ago had lived through the experience of buying a Sound Blaster that was an AdLib card for less money with better features. Sadly the reality was that the GUS was not fully SB compatible and that’s what most consumers thought they were getting. It really is that simple.
Yes, the poor SB compatibility really hurt. There were people who didn’t care (they only cared about trackers or MIDI or whatever), but they were a distinct minority. I think the SB compatibility improved over time, but that was too little, too late. Others (like Ensoniq) did much better.
It’s totally possible that the serial numbers started somewhere north of 10,000. I don’t know if there was any information about GUS sales figures back in the day. The single public document is a SEC 20-F filing that Richard Wells dug up. It’s from 1996 when it was all over, but from the numbers it’s obvious that Gravis was a small company which probably spent way more on sound card development than it could afford, and the total sound card sales must have been somewhere in the 100k-200k unit range (I don’t have a good sense of how much Gravis was selling the cards for, only the retail pricing is well documented).
And I don’t know how many cards Creative moved (perhaps you have some idea?), but I’m pretty sure when the GUS came out there were already more Sound Blasters in use than the GUS ever sold, and SB probably sold at least 1-2 orders of magnitude more each year. And that’s not counting the SB clones. Which is why these days a SB 16 is worth very little, and a GUS easily sells for $100 🙂
Remember that the 100k-200k number is for number of cards produced and sold into the market to distribution. That doesn’t in any way represent the number of cards actually purchased and retained by owners. Creative sold countless millions of cards during their prime years. During the GUS period Creative was selling between 30,000 and 80,000 cards every MONTH.
The saddest fact of this sad chapter is that Advanced Gravis was a top-flight peripherals manufacturer at the top of their game. This folly cost them everything.
The 100k-200k figure should be considered an upper bound; as you say, even if they did sell 200k units it would be laughably little compared to what Creative sold. I don’t know if the returned cards ended up in a landfill or something, I assume they did get sold eventually. Maybe not in the US/Canada, and possibly at a loss. So basically Gravis burned the retail channel, as if they needed more problems than they already had 🙂
Reading the SEC filing it’s not entirely clear but there’s a pretty strong suspicion that the sound cards took down the Gravis peripheral business which seems to have been doing pretty well.
Michal, that’s 100% correct (as I alluded too elsewhere). The GUS sunk an otherwise successful company that was on the verge of becoming Logitech. All that vanished with their folly into sound cards.
little late, but I have a 2.3 serial number K10515. Bought it brand new 🙂