Login

Adding AMD GPGPU support for Nebula

Nebula feature suggestions

Adding AMD GPGPU support for Nebula

Postby clasam » Fri Mar 23, 2012 9:30 pm

It seems as though AMD has really focused on GPGPU/Compute performance in this round of graphics chips, and NVIDIA has purposely scaled back the 680 compute performance to improve on games (http://vr-zone.com/articles/nvidia-kepl ... 15332.html).

In the Sandra 2012 OpenCL 32 bit scores, AMD 7970 scores a 4270, while the new Nvidia Gtx 680 scores just 3000 (same score as the AMD 7870). Meanwhile double precision 64 bit scores are 1120 and 194, for AMD and Nvidia, respectively. The Gtx680 is even outperformed by last year's Gtx 580...

Please work on adding AMD support...I think these GPU compute support would definitely take Nebula to the next level and help improve the compressor performance as well.

Thanks,

Jon
clasam
User Level 0
User Level 0
 
Posts: 4
Joined: Wed Mar 14, 2012 6:00 am

Re: Adding AMD GPGPU support for Nebula

Postby enriquesilveti » Sat Mar 24, 2012 12:15 pm

Seems that CUDA toolkit till into 32 bits. First part of the code was already translate to CUDA but we need more time to complete the pure CUDA release, beta test , etc. In addition we find some issues with Intel IPP on MAC and OSX Lion so that issues add more delay to pure CUDA release...
Enrique Silveti.
Acustica Audio customer and technical support.

MBP 11.5 (i7-4870 | 16 GB | 512 SDD)
SP4 (i5-6300 | 8 GB | 256 SDD)
RME UFX | PS Lyra2 | SD USBPre2
VM U15 | VM W10 | VM OSX 10.12
N4/NAT4 | SPX3 | RX5 | LN2C | Smaart8 | R5 | PT12 | PX10
User avatar
enriquesilveti
Global Moderator
Global Moderator
 
Posts: 2662
Joined: Sun Mar 28, 2010 9:00 pm
Location: Lodi | Madrid | Buenos Aires

Re: Adding AMD GPGPU support for Nebula

Postby Tim Petherick » Sat Mar 24, 2012 10:01 pm

I can't wait for pure Cuda !!

We have been waiting a while .This will make nebula a beast!! We really need this!
User avatar
Tim Petherick
Expert
Expert
 
Posts: 1349
Joined: Sat Apr 17, 2010 4:07 pm
Location: Bath , Uk

Re: Adding AMD GPGPU support for Nebula

Postby clasam » Sun Apr 01, 2012 10:59 pm

Thanks.

As I understand it, NVIDIA opened CUDA, and is using the LLVM compiler which allows other hardware manufacturers to add their hardware support.

If you write for Pure CUDA, will that automatically work for ARM, Intel, AMD GPUs? Or would they have to add support from their side as well?

Would writing in OPEN CL allow for built in hardware support from the other manufacturers?
clasam
User Level 0
User Level 0
 
Posts: 4
Joined: Wed Mar 14, 2012 6:00 am


Return to Requests

Who is online

Users browsing this forum: No registered users and 3 guests