NVIDIA to create another SoundStorm

SAN JOSE — Here at NVIDIA’s Editor’s Day event today, NVIDIA CEO Jen-Hsun Huang addressed questions about the company’s plans for a possible second spin of its popular but ill-fated SoundStorm audio solution. Apparently, a reprise of SoundStorm will happen. Huang told the assembled press types, “We’re gonna build SoundStorm 2. It’s gonna be awesome.” He was less clear on what form the next SoundStorm would take, saying that NVIDIA was still trying to figure out how to deliver SoundStorm as a product. Huang said, cryptically, that the new SoundStorm “will come in a way that you won’t expect.”

In the same Q&A session, Huang spoke about NVIDIA’s history with the NV1 “media processor” chip, which had built-in sound as well as graphics. He disputed that notion that NVIDIA is primarily a “graphics company,” and then suggested that the GPU is going to evolve and expand beyond graphics to include other kinds of digital media.

Based on all of that, it doesn’t sound to me like the next SoundStorm is likely to find its way into a future nForce chipset. Instead, some future GeForce seems like a more likely place for SoundStorm to appear, as strange as that sounds. Of course, it’s all speculation at this point, but in one form or another, SoundStorm will someday return.

Comments closed
    • rika13
    • 15 years ago

    i rather like the gfx card idea, slap some more transistors on the GPU for sound, let the thing render to uber-fast gddr3 (memory isnt the limit anymore; and with turbocache, you wont need the full 256MB anyway) and put a daughtercard on it for your audio jacks/codecs (like dfi and abit are doing)

    this also opens up a BIG door for an attack on ATi sacred ground, the AIW cards, by having a tuner, sound, and gpu on one card it eliminates the patch cables, being nvidia means multiple companies will make them (i hope), it also means an HDTV tuner may be possible

    i wanna see the look on their faces if nvidia makes a personal cinema card with a built-in sound card

    • hmmm
    • 15 years ago

    I too would prefer a discreet card. I might very well opt for an ATI video card, thank you very much.

    • RyanVM
    • 15 years ago

    Here’s to hoping they release a standalone PCIe x1 card, since rumor has it that Creative will be releasing the Sound Blaster Zenith next year with PCIe and Dolby Digital Live support.

    • LoneWolf15
    • 15 years ago

    I for one, do not want a sound/video card in combination. If a company makes a great sound solution but a poor video solution at any given generation, or a good video solution but an average sound solution, I’m stuck with either not choosing either piece of hardware, or choosing a video card but disabling its onboard sound and asking why I paid the extra cash (assuming I can disable it). If nVidia offers SS2 as both an on-the-mainboard solution AND as a discrete sound card I can put with any sound board, I like it. This way, SoundStorm 2 could be marketed much as the GeForce chips are, to any company willing to build a board. It would also offer the option for a high-end cards, if a company was willing to put quality DACs on them.

    • BeowulfSchaeffer
    • 15 years ago

    Soundstorm 2 as part of a USB 2.0 internal card reader!!!

    • ieya
    • 15 years ago

    I’d far rather see a discrete PCI-E x1 card, than integrated onto the graphics card.

    For one, I upgrade graphics cards rather more often than I upgrade soundcards, and don’t particularly want to pay the price premium for an ‘extra’ soundcard every time.

    For another, once you put two DVI ports onto a graphics card backplate, there’s not a lot of room left for all the ports that soundcards need! 🙂

    Unless, of course, they go for double-width cards as standard …

    Or even a low-profile x1 soundcard that would dovetail with a x16 graphics card, sitting underneath the exhaust vents from the graphics card’s heatsink?

    • Herman
    • 15 years ago

    Digital Life article is pretty good (and in english).

    §[<http://www.digit-life.com/articles/nvidianforce/<]§ Looks like pre-DD processing (mixing, 3D, DSP effects, etc.) is done in system memory by the APU directly, then fed to DD encoder which (speculation) may need a pass in mem as well to stay a block or two ahead. All processing is "24-bit internally" but I dont know if that means 24-bits in the buffers, or just 24-bit in the registers (ala 3dfx), so its hard to say what hits mem back and forth. The nVidia HT "Packet Timing" stuff probably makes this work at all. If thats true, then SS might actually need HT and its own hacked mem controller to work well (and cheaply) and maybe a reason why if PCI was not happy. Who wants SoundStorm or a $180 6600 GT with a HT interface? :) I do! Scr*w PCI-Express :)

    • Herman
    • 15 years ago

    For whatever its worth, here is a nVidia whitepaper on its APU:

    §[<http://www.nvidia.com/attach/7136<]§

    • Herman
    • 15 years ago

    From a regugitated nVidia docs:
    §[<http://www.inqst.com/articles/nforce/article.htm<]§ Notice the render in memory part.... And remember soundstorm was born out of the unified memory arch. xbox. nForce’s APU: An Industry Leading Audio Processor The Xbox audio processor is widely viewed as being the best audio processor to be used in a gaming console. Not only is the Xbox’s audio highly sophisticated, it requires a minimum of CPU resources -- in gaming consoles the audio processor must pull its own load. The nForce has the same Dolby Digital 5.1 audio engine native to the Xbox. The nForce APU (Audio Processor Unit) is a multi-DSP unit capable of four billion operations per second. It can simultaneously process 192 2D voices and 64 3D voices -- all fully hardware accelerated. Like the Xbox itself, the APU was designed with DirectX 8 in mind. This full DirectSound hardware implementation supports occlusions, reflections, reverberations, Doppler shifts, and other real-time hardware accelerated sound effects. The APU supports full DSL2 accelerations. The APU also enjoys benefits from having HyperTransport paired with the IGP’s generous memory interface. With so much bandwidth available to it, the APU can render effects directly and reliably to system memory. While a few existing high-end sound cards provide hardware enhanced Dolby Digital decoding, only the nForce provides real-time Dolby Digital AC-3 encoding certified by Dolby Labs. This unique capability means that the APU can encode a DirectSound compliant game in real-time into a 5.1 Dolby Digital stream to be sent to an external Dolby Digital decoder for playback through a home theater system. For access to some of the APU’s more sophisticated features, NVIDIA will be providing a breakout box similar to what is bundled with high-end consumer level audio cards such as the “Sound Blaster Live! Platinum 5.1”. The history of integrated audio is a sad one. Up until now, nearly all integrated sound solutions have been feeble software driven engines. Not only do these implementations commonly slow even the fastest gaming rigs by ten percent or more, but the quality of sound and the flexibility of sound effects are executed to the very lowest standards. The nForce makes a jarring break from this legacy by providing audio capability that is comparable to the most expensive consumer level add-in cards. By comparison, all other integrated sound solutions look pathetically underpowered. If the nForce achieves significant market penetration, the integrated APU could pose a real threat to the future of standalone audio cards.

    • Zenith
    • 15 years ago

    Krogoth255 – What I read said they cancelled them both.

    • RoninGyrbill
    • 15 years ago

    I believe the new soundstorm will come in pill form first, followed soon after by liqui-gels. If the Inquirer is to be believed.

    Get an M-Audio Revo.

    • Zenith
    • 15 years ago

    Ahah! So this is why nVidia canceled the NV50.

      • Krogoth
      • 15 years ago

      IIRC, Nvidia didn’t cancel the NV50, it was actually the NV48. Then again SLI offers a better performance and -[

    • Sargent Duck
    • 15 years ago

    I too hope it’s a descrete card. I think the potential is there to make some money, considering it’s really only Creative with add-in cards.

    • Zenith
    • 15 years ago

    Audio and video on the same card in PCIe? I LIKE! If there aren’t problems with noise, I’m all for it!

    • Consolidated
    • 15 years ago

    SoundStorm2?! Hmmm, the bitter pill of the Bush re-election doesn’t taste so bad now!

      • GodsMadClown
      • 15 years ago

      Speak for yourself. I’d rather have Creative for 4 more years than Bush. I’ll put my Live! back in my machine if Bush doesn’t take office. Are you listening Creative? Get that marketing department to work. You have until the inauguration.

        • hmmm
        • 15 years ago

        ROFL. Hell, I’d go out and *buy* another Live!

    • Toasty
    • 15 years ago

    …the new SoundStorm “will come in a way that you won’t expect.”

    And the article directly below this one is about Nvidia graphics technology being used in the PS3. Uhh, can we connect the dots here? Nvidia provided Soundstorm for the Xbox, I’m pretty sure they will provide an updated version for the PS3, as surround sound has been the biggest limitation of the current PS2, with games only being able to support Dolby Surround, and in some cases DTS.

    Toasty’s 2005 Nvidia prediction: Brand spankin’ new Soundstorm technology in the PS3.

    It’s official.

    • nerdrage
    • 15 years ago

    It’s like deja vu… only 10 years later
    §[<http://www.firingsquad.com/features/nvidiahistory/page2.asp<]§ NVIDIA's first product wasn't only about graphics. The NV1 also integrated a playback-only sound card, something quite popular at the time. With 32 concurrent audio channels of 16-bit CD-quality audio and hardware phase shifting for simplistic 3D sound, the NV1 was actually more impressive than many first generation PCI sound cards.

      • d0g_p00p
      • 15 years ago

      So strange. I was beta testing games and hardware and I tested the NV1. I tested the Phillips NV1, because nVidia partnered with them at the time. Ahh the days.

    • SpotTheCat
    • 15 years ago

    I just hope whatever it has will be consistant between manufacturers, and a ton cheaper than creative. if anyone can compete with creative, I’de say nVidia can with their market hold and history in this reletivly un-touched market.

    • blitzy
    • 15 years ago

    I really hope they don’t go with integration to a gfx card or mobo, a fully featured discrete sound card would be preferable for me

    • IamSpazzy
    • 15 years ago

    There was a paper a while back about using the GPU for sound processing. I’m thinking that nvidia is probably going this route. §[<http://www.bionicfx.com/.<]§ Given the fact that work has already been done and it's in software, Nvidia probably just needs to write a driver abstraction layer, and probaby create a break out card out actually get the sound out.

      • Freon
      • 15 years ago

      Yeah I think this is actually been a good idea since T&L engines were first introduced. I’ve been waiting for it. A T&L engine would be a good start on your way to bouncing sound.

        • SXO
        • 15 years ago

        It’s about time someone said it. It’s the *[

    • Captain AMD
    • 15 years ago

    I hope this one isn’t as dissapointing as Soundstorm. As far as integrated solutions go it was the best but it didn’t deliver half of what it promised and continues till this day to have problems.

    • spworley
    • 15 years ago

    I think #1 is correct… combine the sound processing with the GPU.

    And if you’re really clever you can even share processing between the graphics and the audio silicon.. A pixel pipe is a big floating point DSP. What does Soundstorm need for encoding 5.1 Dolby with greater than 16 bit precision? A big floating point DSP.

    • barich
    • 15 years ago

    PCI has enough bandwidth for real time DD encoding. There were rumors that it didn’t, but nVidia never confirmed them. Realistically, there’s no way that PCI couldn’t have enough. 8 uncompressed 24-bit 192khz audio channels only add up to 36 MB/s. SoundStorm does substantially less than that.

      • Herman
      • 15 years ago

      If it was only passing this stuff through in the way a normal audio card would do (mixing and other relatively light DSP work) that would be ok. But there is more to it.

      Plus you have to consider that somethign has to keep streaming this stuff in chunks into memory as well, and somehting has to be reading it from HD / CD / Network.

      • Seawolf(crp)
      • 15 years ago

      The APU, running at full steam, can use 200MB/sec. PCI offers 133MB/sec.
      This was confirmed by NVIDIA.

      Seawolf of NFHQ, who (Can’t ever Remember his bloody Password for this place)

    • Shining Arcanine
    • 15 years ago

    Personally I’d prefer to see this as a stand alone PCI or PCI Express card rather than a component intergraded with the graphics card. Doing so lowers upgrade costs when you don’t want to upgrade everything.

    It would also be more of an attack on Creative’s marketshare as an add in card than it would be as an intergraded component. Therefore forcing Creative to innovate (particularly in their drivers).

    • sherfh
    • 15 years ago

    All I can say is I hope they fix the audio drivers this time…

    • Loie
    • 15 years ago

    i hope and pray nvidia finds success with this, just to crush Creative and their crappy sound blaster drivers once and for all!

    • DukenukemX
    • 15 years ago

    Sounds to me like Nvidia is going to have SoundStorm2 built into future graphics cards. Would make sense because PCI-E 16X would offer enough bandwidth for Dobly Digital Encoding.

    Only problem is that I really want it as a add-on card or built into my motherboard.

    I also hope they don’t charge an arm or a leg for SS2 in their graphics cards.

    • 5150
    • 15 years ago

    Bring. It. On.

    • Freon
    • 15 years ago

    How about just a PCI-E 1x slot, and bundle it with the boards?

    I have a feeling Nvidia will be bringing us some more interesting things on the front of bouncing and propogating sound.

      • Spotpuff
      • 15 years ago

      Does PCI-E x1 have enough bandwidth for real time Dolby encoding?

      I know HT does… but I am not so sure about PCI-E. if it doesn’t, then this isn’t an option.

        • _Shorty
        • 15 years ago

        well, 48KHz 16bit x 5, plus sub .1
        2 bytes x 5 x 48000=480000 bytes per second
        plus sub channel, which is much less than a full bandwidth channel

        so, it’s nothing to worry about, really. Even without the sub channel, it’s only 468.75KB of data per second. That’s nothing.

          • Herman
          • 15 years ago

          Its not the streaming bandwidth for the sound data, its the bandwidth to memory for processing it.

          Back in the day when they had the mem controller to themselves, they could do whatever horendous hack they needed to to pull it off. Think of it in terms of what video cards need, where they have their own memory to play with.

          Once the mem controller went on the processor the old design just didn’t fly. So many times, like for network cards or hard drive controllers, they hardware is designed around the bus it is sitting on. Each controller doesn’t have all the power and flexability to run like software does on the processor. And to get the speed needed, much of it is “real hardware” logic married to the busses. Its very expensive to just start all over, and each bus has its own tricks to get up to speed needed. This is one reason, as an example, why there is no cheap hardware based raid 5 right now, its just too much processing at the speeds these things have to run to be practical and economical.

          Also, I think alot of it was, and just a guess, that whatever license they got for xbox carried over, but only for those chips. They couldn’t afford it by them selves, and either intel or ms helped them set up the original deal for a minimal cost, or subsedized (sp) it for them.

            • _Shorty
            • 15 years ago

            the dolby encoding process never touches system memory, so, don’t know why you even brought it up. All the sound processing takes place on the chip itself, including the dolby encoding. All raw sound info gets piped into it, it does all the processing it needs to, and finally encodes it to DD, and that gets piped out on the SP/DIF. Only thing that’s ever in system memory is the raw sound info.

            • Herman
            • 15 years ago

            Thats just not true. Encoding dolby digital is not just some nowadays simple DSP issue where you mutate a stream as it goes through. Doing it in “real time” is not cheap in lots of ways. If it was that simple and “only” a license issue, there would be relatively lots of reasonably priced solutions.

            §[<https://techreport.com/ja.zz?comments=6409<]§ This is from a Hexus article that got yanked. BTW, all modern audio cards use system memory at least to read the stream, which has to be written by the app/driver to memory first.

            • _Shorty
            • 15 years ago

            um, audio encoding is audio encoding. You give the seperate audio streams to the codec, and it encodes it. There’s nothing magical about it. You feeling ok?

            • Herman
            • 15 years ago

            In a perfect world yes, if by codec you mean a regular hardware dsp / dac before it hits whatever transport medium (jacks on the back).

            But in this case there is the compression on five channels to deal with just for starters… Getting 10/12 to 1 on sound is not cheap in processing time or memory, especially if you are starting with proably 24-bit source from the regular audio processing. If this were *decompressing / decoding* it would probably be more like what you say, but would probably still have to buffer the incomming somewhere to work on it before it could be uncompressed and useful. I dont imagine there is room on the south bridge for gobs of memory.

            Dont confuse software codecs which have a whole PC system behind them to run on, with what the APU has to work with, which is basically its relatively slow running controller and luckily a mem controller just down the hall.

            See some othe rh other posts, like #52, and read some of the docs I posted. If you are still convinced that the APU never touches memory, then I dont know what to say.

    • PRIME1
    • 15 years ago

    l[

      • atidriverssuck
      • 15 years ago

      hah!

      • SpotTheCat
      • 15 years ago

      lol! that sounds like something they would do… everyone gets all hyped up and then it’s in a bad form nobody would want to use.

    • MaceMan
    • 15 years ago

    Yeah, I can see SoundStorm 2 on the next generation of video cards, whether as an optional chip that sits in its own socket, or in a line of GPUs. Either way, it is a vendor option for top-of-the-line stuff. It would certainly be a feature distinction and help SFF boxes that already sacrifice space to a video card (for gaming).

    I just wonder how they’ll roll the drivers for those bad boys. A seperate line of drivers for StormForce video cards?!?

    This is tweaking my geek meter through the roof. I like it.

    • atidriverssuck
    • 15 years ago

    Would be good if they got everything right this time, especially EAX support. Could this be a PCI Express soundcard? A firewire one? An external model? All I know is we need competition here, and it would be good to see some.

    • Spotpuff
    • 15 years ago

    *[< Huang said, cryptically, that the new SoundStorm "will come in a way that you won't expect."<]* Like, in the next Nvidia chipset hopefully? "It wasn't in nforce 3,4 or 5... but just wait 'til nforce 8!"

    • PRIME1
    • 15 years ago

    Hmmm… Soundstorm 2, SLI, Nforce 4 & 5, Playstation 3, Geforce 6xxx. Next year looks to be a good year for NVIDIA.

    • zurich
    • 15 years ago

    Heh more PS3 news? 😉

      • Freon
      • 15 years ago

      Good point. I have a feeling they broke this news because it is going to come out soon anyway. Distinct possibility.

    • GodsMadClown
    • 15 years ago

    Why do I want my audio near my video? I thought you were supposed to put the soundcard as far from the vid card as possible. Educate me.

      • UberGerbil
      • 15 years ago

      Only if you’re worried about interference in analog components. If your sound and video are all digital, that doesn’t matter. And last I checked, most kinds of video (movies, games, etc) include audio, so when you’re processing one you’re usually processing the other.

      From nVidia’s point of view, it gives them a feature ATI doesn’t have, and possibly a way to sell (the chips for) add-on cards.

    • adisor19
    • 15 years ago

    Hmm, interesting development indeed 🙂
    I can only expect great things to happen with the next SoundStorm. Whether it comes on the GPU or as a discreet chip, it’s better then nothing. And it’s just what Creative needed : a good reality check.

    Adi

    • UberGerbil
    • 15 years ago

    I’ve actually been thinking they might put SS onto the GPU for a while. With PCI-E you don’t have to worry about the bandwidth.

    There’s already a thread about this in the forums
    §[<https://techreport.com/forums/viewtopic.php?t=24701<]§

Pin It on Pinterest

Share This