AMD TrueAudio support in games?

[MtSeldon]

we read new games that support Mantle all the time, but i don't see any news about TrueAudio.

is there any new game on the horizon? or is TrueAudio doomed?

 

[AtenRa]

Thief has TrueAudio with Catalyst 14.3 and later(Mantle as well). As of now i dont know another game with TrueAudio.

http://community.amd.com/community/...talyst-143-beta-the-ultimate-driver-for-thief

 

[n0x1ous]

Lichdom: Battlemage I think was one. You can get it now with early access on steam (not sure if true audio feature is implemented yet)

 

[Flapdrol1337]

True audio has far fewer people that can actually run it.

 

[NostaSeronx]

True audio has far fewer people that can actually run it.

It will be more widely supported as Realtek is implementing the same Tensilica IP.

http://www.cadence.com/cadence/newsroom/press_releases/pages/pr.aspx?xml=082113_rtt

 

[MtSeldon]

I know about thief's Trueaudio support . But its said to be not a proper implementation. Also it had low scores from game reviews.


I dont understand why AMD doesn't give enough support to it as they do with mantle. DSP adds a lot to the atmosphere of the game.

 

[motsm]

I know about thief's Trueaudio support . But its said to be not a proper implementation. Also it had low scores from game reviews.

All TrueAudio does in Thief is offload the already existing convolution reverb to the GPU. So AFAIK, it's not improving anything outside of performance.

I dont understand why AMD doesn't give enough support to it as they do with mantle. DSP adds a lot to the atmosphere of the game.

I would assume because gamers as a whole don't really care about sound. It's part of the reason games no longer use EAX or any form of hardware audio processing, software has gotten to a point where it's "good enough" for 99% of gamers.

 

[Headfoot]

TrueAudio in Thief is a completely different algorithm, and unfortunately it sounds like absolute garbage. They just slapped a convolution reverb on everything that's turned to 11 at all times. It's like the whole world is in a cathedral... I have a 290 and I turn it off on purpose. Which is disappointing because I'm very much looking forward to improved sound.

Lichdom looks significantly more promising. They have a youtube video up which demos it. It is a subtle but noticeable effect, they use the TrueAudio dsp to run headspace tracking style algorithms, so you can hear the direction something is coming from with a great deal more accuracy. Instead of being able to hear left/right, you can hear left/right/up/down and how far away roughly.

 

[motsm]

TrueAudio in Thief is a completely different algorithm, and unfortunately it sounds like absolute garbage. They just slapped a convolution reverb on everything that's turned to 11 at all times. It's like the whole world is in a cathedral... I have a 290 and I turn it off on purpose. Which is disappointing because I'm very much looking forward to improved sound.

I just looked up a YouTube video, and it looks like they do change the reverb a bit for the hardware version

The convolution reverb Thief uses is part of the middleware API and not TrueAudio, it existed before TrueAudio came along actually. It seems however that when you switch it over to hardware mode, the dev's cranked up it's volume (or gimped the software version). There is no technical reason in Thief's implementation of TrueAudio that should make the convolution reverb change the sound drastically from the software mode like it is, so I'd say they were trying to make TrueAudio a bit showy and went overboard.

https://www.youtube.com/watch?v=jpcfppVazi0

 

[chimaxi83]

TrueAudio has been disappointing vaporware so far. Meh. I was looking forward to a change in the audio world.

 

[Gloomy]

I just looked up a YouTube video, and it looks like they do change the reverb a bit for the hardware version

The convolution reverb Thief uses is part of the middleware API and not TrueAudio, it existed before TrueAudio came along actually. It seems however that when you switch it over to hardware mode, the dev's cranked up it's volume (or gimped the software version). There is no technical reason in Thief's implementation of TrueAudio that should make the convolution reverb change the sound drastically from the software mode like it is, so I'd say they were trying to make TrueAudio a bit showy and went overboard.

https://www.youtube.com/watch?v=jpcfppVazi0

Software setting is/was bugged, it didn't do anything. Not sure if it's been fixed.

 

[motsm]

Software setting is/was bugged, it didn't do anything. Not sure if it's been fixed.

The video I linked it seems non existent like you said, another video it just seemed quieter than the TrueAudio version, so maybe that was after a patch?

 

[Headfoot]

I just looked up a YouTube video, and it looks like they do change the reverb a bit for the hardware version

The convolution reverb Thief uses is part of the middleware API and not TrueAudio, it existed before TrueAudio came along actually. It seems however that when you switch it over to hardware mode, the dev's cranked up it's volume (or gimped the software version). There is no technical reason in Thief's implementation of TrueAudio that should make the convolution reverb change the sound drastically from the software mode like it is, so I'd say they were trying to make TrueAudio a bit showy and went overboard.

https://www.youtube.com/watch?v=jpcfppVazi0

Yeah I'd agree. It's like early PhysX games where suddenly everything exploded with debris and everything had banners on it

 

[dailow]

I would assume because gamers as a whole don't really care about sound. It's part of the reason games no longer use EAX or any form of hardware audio processing, software has gotten to a point where it's "good enough" for 99% of gamers.

I disagree... the original Half Life when it was released with Aureal A3D 2.0 (wavetracing) support was amazing and unfortunately the state of 3D audio has regressed since then. Part of the reason is due to Creative Labs suing Aureal into oblivion, then acquiring Aureal's assets and doing nothing with them.

The other reason has to do with Microsoft completely rewriting the audio stack starting with Windows Vista, which prevented developers from directly accessing audio hardware in the API. The new audio stack combined with Creative Labs' incompetency (even their drivers are terrible) led to a dramatic decline in the state of 3D audio.

It's a shame too, because proper 3D Audio really does dramatically improve the FPS experience.

 

[3DVagabond]

I think that AMD might have to do more development work for it to take off (and see it use it's real potential). It doesn't seem to be a priority right now though. Maybe because so few sku's support it?

 

[motsm]

I disagree... the original Half Life when it was released with Aureal A3D 2.0 (wavetracing) support was amazing and unfortunately the state of 3D audio has regressed since then. Part of the reason is due to Creative Labs suing Aureal into oblivion, then acquiring Aureal's assets and doing nothing with them.

The other reason has to do with Microsoft completely rewriting the audio stack starting with Windows Vista, which prevented developers from directly accessing audio hardware in the API. The new audio stack combined with Creative Labs' incompetency (even their drivers are terrible) led to a dramatic decline in the state of 3D audio.

It's a shame too, because proper 3D Audio really does dramatically improve the FPS experience.

Studios develop all of their games as console lead these days and we get the port of whatever audio API they used, so I don't think what Creative or Windows did makes any difference today.

 

[imaheadcase]

Because devs know that what you can do with trueaudio you can already do in software so not a huge deal. Trueaudio would be nice if we all ran dual core CPUs under 2ghz.

 

[NTMBK]

Studios develop all of their games as console lead these days and we get the port of whatever audio API they used, so I don't think what Creative or Windows did makes any difference today.

Apparently the PS4 has the same TrueAudio DSP, so there's some hope there at least.

 

[Headfoot]

Because devs know that what you can do with trueaudio you can already do in software so not a huge deal. Trueaudio would be nice if we all ran dual core CPUs under 2ghz.

Not true.

TrueAudio is 2+1 Tensilica DSPs that can run very particular runtimes significantly faster than a general purpose CPU could. That's literally the purpose of a DSP.

You can do -the general algorithm- in software. You can't do anywhere near as many iterations, with as much resolution, from as many sources, or otherwise increase the scope of the effect, when you run it on CPU. So, you can't do the same thing in software. You can only do a low res approximation of the same thing in software. A parallel example is that the 360 can run Crysis 3 sure, but it's not anywhere near the same visual fidelity as on the PC version. Same basic idea here

In real life it doesn't matter that TrueAudio is computationally superior in specific contexts. Developers would rather spend developer time and money on making the graphics better, adding levels, content, optimization, 95 times out of 100 instead of adding advanced sound. They must not see enough demand for it, so no one takes the time to program the more advanced effects. So the discussion is pretty much theoretical since the economics of it don't work out. Chicken and the egg

 

[Headfoot]

Apparently the PS4 has the same TrueAudio DSP, so there's some hope there at least.

Here's to hoping!

So far it seems the only games to really make the sound a priority (the new Killzone seems like it does though not very explicitly) are PS4 exclusives. I imagine most cross platform devs will program to whichever audio API is the lowest common denominator between Xbone/PS4/PC which surprisingly for once will likely be non-TrueAudio PCs. Hopefully though if the end up allowing the PS4 exclusives to get ported over eventually they drag the DSP code with it and port it to TrueAudio

 

[monstercameron]

Here's to hoping!

So far it seems the only games to really make the sound a priority (the new Killzone seems like it does though not very explicitly) are PS4 exclusives. I imagine most cross platform devs will program to whichever audio API is the lowest common denominator between Xbone/PS4/PC which surprisingly for once will likely be non-TrueAudio PCs. Hopefully though if the end up allowing the PS4 exclusives to get ported over eventually they drag the DSP code with it and port it to TrueAudio

AMD did mention that TA could be effortlessly shipped with audio middleware like fmod
http://www.anandtech.com/show/7370/amd-announces-trueaudio-technology-for-upcoming-gpus

 

[motsm]

Here's to hoping!

So far it seems the only games to really make the sound a priority (the new Killzone seems like it does though not very explicitly) are PS4 exclusives.

Yeah, the new Killzone has some very simplistic ray tracing, if I recall correctly it uses ray projections from the players gun only. I would think if they were using the DSP they could beef up the ray count and include other sounds, so I'm not really sure why it's so limited, but at least it's something "new".

 

[3DVagabond]

I think AMD needs to develop some libraries that devs can use to add Trueaudio to their games. pre programmed environments. I think expecting devs to program their own environments is a losing effort.

 

[motsm]

I think AMD needs to develop some libraries that devs can use to add Trueaudio to their games. pre programmed environments. I think expecting devs to program their own environments is a losing effort.

They don't have to, AMD teamed up with a couple popular API dev's that are already widely used in software mode. That's why Thief has a software mode for it's convolution reverb, the API is handling it as it always would, while the hardware mode offloads the processing to the DSP chip. From what I have read the API uses some impulse responses from AudioEase, and likely many of their own. It can be a relatively expensive and time consuming effort and requires an expert to get proper results, so I'd be surprised if EM had any involvement with the recordings.

 

[3DVagabond]

They don't have to, AMD teamed up with a couple popular API dev's that are already widely used in software mode. That's why Thief has a software mode for it's convolution reverb, the API is handling it as it always would, while the hardware mode offloads the processing to the DSP chip. From what I have read the API uses some impulse responses from AudioEase, and likely many of their own. It can be a relatively expensive and time consuming effort and requires an expert to get proper results, so I'd be surprised if EM had any involvement with the recordings.

I think more could be done though by sampling real 3D environments that you could simply program the objects into. It's the way DSP's have handled it in the past. I'm not an expert though. I could be wrong.