• Open Menu Close Menu
  • Apple
  • Shopping Bag
  • Apple
  • Mac
  • iPad
  • iPhone
  • Watch
  • TV
  • Music
  • Support
  • Search apple.com
  • Shopping Bag

Lists

Open Menu Close Menu
  • Terms and Conditions
  • Lists hosted on this site
  • Email the Postmaster
  • Tips for posting to public mailing lists
Re: Using the GPU to process floating point samples
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Using the GPU to process floating point samples


  • Subject: Re: Using the GPU to process floating point samples
  • From: Stéphane Letz <email@hidden>
  • Date: Thu, 17 Feb 2011 18:43:59 +0100

>
> Message: 7
> Date: Thu, 17 Feb 2011 07:40:00 +0100
> From: tahome izwah <email@hidden>
> Subject: Re: Using the GPU to process floating point samples
> To: email@hidden
> Message-ID:
> 	<AANLkTimBB6ZuJ_9=email@hidden>
> Content-Type: text/plain; charset=ISO-8859-1
>
> This has been discussed in a rather lengthy thread titled "AudioUnits
> and OpenCL". In a nutshell, GCD and OpenCL are Dont's for realtime
> audio processing. Markus Fritze summed it up nicely:
>
>> Ehm, you know that GCD is running at the main thread level? Which can be blocked by UI operations,
>> etc. That doesn't seem like a wise choice for real-time processing. OpenCL also doesn't have a
>> threading mode, so you tasks will be shared among all the others and your latency becomes
>> unpredictable.
>>
>> Markus
>
> HTH
> --th


We have done some tests in the context of the Faust project (http://faust.grame.fr/), that is compiling GPU code (OpenCL or CUDA) starting from the Faust specification. The main results are:

-  threading or latency are not the most problematic issue. We could achieve a one buffer latency implementation in a RT context that was quite acceptable. Threading issue are probably manageable by using dedicated GPU card for audio.

- the most critical issue is the kind of processing one can do on GPU: GPUs are designed for massively data parallel algorithms, and this is not the case for most audio processes (as soon as you have a "recursion" somewhere).

- correct memory management (layout + access) is also critical, then again redesign audio effect with this constraint is not trivial

My feeling is that GPU for audio can work only in a limited number of cases, and this will take some more time to emerge.

Stéphane Letz _______________________________________________
Do not post admin requests to the list. They will be ignored.
Coreaudio-api mailing list      (email@hidden)
Help/Unsubscribe/Update your Subscription:

This email sent to email@hidden

  • Follow-Ups:
    • Re: Using the GPU to process floating point samples
      • From: Paul Davis <email@hidden>
  • Prev by Date: Re: Bluetooth IO problems (aurioTouch sample app)
  • Next by Date: Re: Using the GPU to process floating point samples
  • Previous by thread: Re: Using the GPU to process floating point samples
  • Next by thread: Re: Using the GPU to process floating point samples
  • Index(es):
    • Date
    • Thread