Technical investigation into PS4 and XB1 audio solutions *spawn
- Thread starter Brad Grenz
- Start date
Highly doubtful that audio processing will take any significant amount of GPU or CPU resources, insofar as impacting the graphics or physics/AI fidelity (at least in the vast majority of games, as it depends on the complexity of the particular audio implementation in each game). Carmack talked about how little of the CPU needed to be dedicated to output in-game 5.1 audio during the development of DOOM 3 (this was back in 2003). One can imagine how it's even easier to calculate 5.1 game audio now, with the much, much more powerful modern CPUs (and GPUs, if devs really want to go that route) we have nowadays.
The worst-case scenario I can envision for the PS4 is that multiplatform titles have more-or-less worse audio fidelity and/or complexity, in some scenarios, than the Xbox One. These audio differences will go unnoticed by nearly every gamer without an audiophile's ear, and it will go completely ignored in almost every multiplatform game comparison--just as audio differences were completely ignored between the PS3 and 360. This is of course assuming that the PS4 actually does have a significantly inferior DSP for audio.
According to Cerny:
So, "the principle thing it does," doesn't mean "the only thing it does.' If we knew the part number of this DSP chip, we'd know just exactly what the totality of its functions are.
The worst-case scenario I can envision for the PS4 is that multiplatform titles have more-or-less worse audio fidelity and/or complexity, in some scenarios, than the Xbox One. These audio differences will go unnoticed by nearly every gamer without an audiophile's ear, and it will go completely ignored in almost every multiplatform game comparison--just as audio differences were completely ignored between the PS3 and 360. This is of course assuming that the PS4 actually does have a significantly inferior DSP for audio.
According to Cerny:
So, "the principle thing it does," doesn't mean "the only thing it does.' If we knew the part number of this DSP chip, we'd know just exactly what the totality of its functions are.
Last edited by a moderator:
Silent_Buddha
Legend
Doom 3 does virtually nothing with audio. It's most definitely not something to hold up as an example of well processed game audio. It does positional, very few environmental effects, virtually no environmental interactions/permutations, no wavetracing at all, etc.
The audio in Doom 3 was about as basic as you could get for positional audio, most AAA console games now days do more with audio than Doom 3 did.
Carmack did a little bit with EAX, but not very much as he was very public about being angry that he was forced to use it for any hardware accelerated audio effects. Hence a far older game, Unreal Tournament, did far more than Doom 3 does, when used in conjunction with hardware acceleration of EAX (not possible with just standard audio hardware of the time) which would have brought games to a crawl if they tried to emulate it in software (CPU).
Considering how much Sony likes to go to lengths to expound on their wonderful hardware and the custom alterations they have made to the standard GCN building blocks, the fact that they say nothing about extra features is very telling. It in fact, most likely means there is nothing more than what was stated. Especially when asked directly about X hardware's capabilities. For anything that does more than standard, they always point out that it does more.
Regards,
SB
Cerney is not a hardware guy, he often talks from a very high level, never technical details. People also have very little attention span for audio and audio related tech, they can't see it, they don't care.
The last part is just a fallacious argument "Evidence of absence"...
It frees up the CPU (which isn't powerful) or GPU to do other things. As far as I've read, the new audio processor uses the same audio API that was used on 360. So if you write audio code as normal, it will run on the audio processor without you having to do anything special.
On top of that, there are some people that appreciate high quality audio. I do agree that most people will probably not have a great audio setup, but the people that do may appreciate having a good sound mix.
We have an audio engineer on this board who told us XB360 could consume the better part of a whole core, and the XB1 audio block does more than Xenon could manage in its entirety. Pointing to a lack of devs using demanding audio because of a lack of resources needed to implement it is like pointing to a NES and saying, "we manage graphics just fine, so there's no need for 3D hardware."
The fact audio has stagnated can be attributed to a lack of available resources more than it's reached a dead-end and cannot be improved upon (although saying that, I doubt it'll make a huge difference to most gamers even if XB1's audio was leaps and bounds improved).That's unreasonable. Firstly, it's not a DSP chip but a section of silicon on Liverpool (and if it was a DSP chip, it'd no doubt do more than just decode audio). Secondly, Cerny was asked if there's hardware, and all he cared to mention was that it mostly does audio de/compression. If it could do a lot more (decompression is a fraction of what SHAPE does) he'd have said, "it handles a lot of the audio tasks, such as decompressing audio, filtering it, and resampling." Or somesuch. He'd no more ignore those abilities than say, "our GPU mostly processes vertices ready for drawing." Taking his response as open-ended and entertaining the possibility of it holding significantly more power is ignoring the wealth of common-sense evidence to the contrary - leaks that don't mention anything beyond audio de/compression and the system's designer not saying it does anything more than de/compression and a boat-load of details on every other facet of the hardware such that it's obvious Sony aren't keeping anything back (for no good reason).
bkilian, do you think it is possible to transform your voice in real-time into something like the voice in this almost 20 years old game, Killer Instinct, just using the processing capabilities of SHAPE alone?
It would be nice to hear something like this changing parameters with the use of your imagination.
It would be nice to hear something like this changing parameters with the use of your imagination.
Hmm, I don't see how it's "unreasonable" for one to interpret Cerny's wording in that way. If he had said, "it only does decompression," or merely, "it does decompression," then I would take it that to mean the dedicated hardware only does decompression. But when he says, "its principle function is decompression," it should stand to reason that it doesn't just do decompression, but also does other does "other stuff." If my interpretation of his wording here is off in some way, then please explain.
I think it's entirely possible that he simply thinks most people don't care about every single (mostly uninteresting) technical function the sound hardware in the PS4 serves, and thus decided not to delve too deeply into it. It's also possible that audio is not one of his areas of expertise, and that's partly why he might have glossed over what the PS4 can do audio-wise.
Yes, he did express some enthusiasm about audio raycasting by the GPU, but that was likely just another way for him to illustrate his point about the general processing potentials of the GPU (something he's enthused about in almost every interview or presentation on the PS4).
So there are leaked documents showing that the sound hardware on the PS4 is only used for decompressing audio streams? If so, I put way more credence in that than any hearsay, conjecture, or speculation by you and I. If there really are leaks saying as much, then I can close the book on whether or not the PS4 sound hardware can do much more.
I still maintain that any difference in audio quality or complexity will go entirely unnoticed by 90%+ of gamers. Hell, I consider myself a semi-audiophile, and I have a difficult time distinguishing the difference in 5.1 fidelity between games like DOOM 3 and Splinter Cell: Chaos Theory on Xbox (back in 2004), and any Call of Duty game, for example. Some games that I do appreciate the sound for its exceptional clarity (Amnesia: The Dark Descent, and Battlefield 3) I attribute more to a combination of higher bitrates and better output format (e.g., uncompressed LPCM) than complexity of the audio engine. Incidentally, this extra fidelity and detail means larger sound files, which comes at the cost of RAM, and that happens to be an area where the PS4 has more of at its disposal. So, *shrugs*, who really knows the end result of which console will "sound better."
Last edited by a moderator:
Voice transforms are very cheap. We had a bug on 360 at one point that made everyone into Darth Vader. I was sorely tempted not to report it.bkilian, do you think it is possible to transform your voice in real-time into something like the voice in this almost 20 years old game, Killer Instinct, just using the processing capabilities of SHAPE alone?
It would be nice to hear something like this changing parameters with the use of your imagination.
Basically a simple SRC and an EQ/CMP would do everything you'd need. So yes, SHAPE could handle it.
Based on what's been explained its not even a case of "good or better" quality audio, it's that the basic requirements of accurate audio can be very taxing on a CPU. Shifty mentioned games taking more than a whole core of CPU, IIRC that was Forza and it wasnt about audiophile quality sound per se but just the amount of processing required for all those cars, the crowd, doppler effect, etc.
Well, the 360 has a CPU from 2005, so let's keep that in mind. I'm assuming that a CPU (or GPU) from 2013 would have a much easier time handling the same load of audio processing. My PC is currently equipped with an AMD quad-core (ca. 2010, I believe), which is hardly state-of-the-art by today's standards, but it can run pretty much every high-end PC game I throw at it, surround sound and all, without hardly breaking a sweat.
Whatever the case, if the audio processing on the Xbox One version game really would impact the graphics and/or physics/AI processes that much on the PS4 (which I doubt), I'd venture that most developers--especially multiplatform developers--would simply pare down the complexity of the audio processing for the PS4 version, if need be. And I doubt anyone would even notice.
It'd be disappointing for the audio on the PS4 to actually be less accurate than last gen, is that what youre proposing would happen?
And you dont have any evidence of that a pink winged elephant don't exists, therefore it exists.
Edit: just in case people out thinks I believe in pink winged elephant, I don't. What trying to say is we take in all facts and form a theory. Then test the theory...and if it holds up to new discovery of facts.
So far the PS4 doesn't seem to have any dedicated hardware to advance audio processing, however I agree with Statix (somewhat) that the majority of the gamers will not care or could tell the differences without a/b testing. Hell, the majority of gamers could barely tell about sub-hd resolution of this gen.
Last edited by a moderator:
As mentioned a few pages back this is apparently a ~100 GFLOPS part, 8 extra CPU cores weren't enough to emulate its functionality in alpha kits.
At the least it should take a load off XboxOne's CPU/GPU.
Last edited by a moderator:
I never asserted anything existed, I was pointing out the poor logic that just because they don't publicly disclose something is not evidence it is not there.
You see this is the kind of logic that Xbox fanatics use to validate their secret sauce...don't play that game.
I don't think I inferred that, and it's extremely doubtful that would be the case. The PS3 had no dedicated hardware for sound, and relied solely on the Cell processor, and it was by most accounts as good or better than the 360 in terms of in-game sound quality this generation. There's little reason to the believe that the PS4, with its better CPU and dedicated sound hardware (ambiguous as its capabilities may be), would fare any worse in the sound department than the PS3.
I understand the concern that Xbox One may possibly possess an "advantage" of sorts (again, unconfirmed) with its dedicated hardware for sound as a way to free up CPU/GPU resources, but I highly doubt this advantage will amount to anything that will make up for its _reported_ deficit in GPU capability in real-world, multiplatform graphics performance by a significant degree. For various reasons, I think this processing coup will be minimal at best, insofar as impact to graphics performance.
Last edited by a moderator:
Yet again, surround sound has nothing to do with the complexity of sound design in a game. Your output format and your audio effects graph are only tangentially related.
For audio work, which generally involves highly optimized vector operations, the 360 and the Jaguar CPU in the new consoles are roughly equal.
Similar threads
- Locked
