• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Adobe and AMD Enable Brilliant Experiences

AtenRa

Lifer
Today marks an exciting moment with the announcement of AMD and Adobe’s collaboration on Adobe® Photoshop® and Premiere® Pro CS6. The collaboration between AMD and Adobe brings the first implementation of OpenCL™ heterogeneous compute within the Adobe Creative Suites family to optimize new and existing features in Adobe Photoshop CS6 and Premiere Pro CS6.

This is wonderful news for the millions of creative professionals using Adobe products. Adobe is launching Adobe Premiere® Pro CS6 which now includes OpenCL™ accelerated features in the Mercury Playback engine as well as Adobe® Photoshop® CS6 with breakthrough performance enabled by industry-standards with OpenCL and OpenGL acceleration in the new Mercury Graphics Engine. Benefit users can look forward to include an ever-growing array of hardware accelerated functions able to offer amazing performance and productivity.

Source
 
it's just some cuda-related stuff... nothing really important here (exept for another bad news for nvidia)
 
This is good news. These kind of steps are what will bring gpu compute mainstream and out of proprietary niches.
 
Well...then it's time to benchmark CUDA vs OpenCL in Adobe...and get the full picture.

Because that is what matters right?

Performance.
 
Including IGPs? A distasteful as they may be...

Not sure if serious.
IGP performance in GPGPU is very limited, the biggest strides I have seen would be Intel's QuickSync...hence why it would be nice to see benchmark...I'm not going to yell:

"Wauv OpenCL!!!"

...if the performance is slower than the CUDA code...

Or joy over IGP's...when the performance data is missing.

Hence why I need benchmarks to make an informed decision.
 
Not sure if serious.
IGP performance in GPGPU is very limited, the biggest strides I have seen would be Intel's QuickSync...hence why it would be nice to see benchmark...I'm not going to yell:

"Wauv OpenCL!!!"

...if the performance is slower than the CUDA code...

Or joy over IGP's...when the performance data is missing.

Hence why I need benchmarks to make an informed decision.

2/3 'rds of all computers sold are CUDA capable? That is what I was asking for clarification on.
 
No, you are being mentatious now IMHO.

Lets wait for the reviews...before praising PR.

What? Seriously? You called 2/3 rds not a niche? What are you talking about? Honestly, I am trying to figure out what you were trying to say.

If you meant CUDA capable GPUs represent 2/3 rds of the market, OK. I was asking if that included IGPs in the mix. That's it. That seems high to me, so I was hoping you would clarify.

It's worth noting that with IVB (once drivers enable it) IGPs from both AMD and Intel will be OpenCL capable. Which means that Adobe gets to code once (hopefully, or the model is broken) and it will provide benefit to everyone.

I've bought (and likely will buy more) GPUs for CUDA exclusively...
 
Not sure if serious.
IGP performance in GPGPU is very limited, the biggest strides I have seen would be Intel's QuickSync...hence why it would be nice to see benchmark...I'm not going to yell:
While QuickSync is a great feature, it's more like a stride in the opposite direction: it mainly relies on a fixed function hardware developed for a specific purpose, while the GP prefix in GPGPU means 'general purpose'.
 
What? Seriously? You called 2/3 rds not a niche? What are you talking about? Honestly, I am trying to figure out what you were trying to say.

If you meant CUDA capable GPUs represent 2/3 rds of the market, OK. I was asking if that included IGPs in the mix. That's it. That seems high to me, so I was hoping you would clarify.

It's worth noting that with IVB (once drivers enable it) IGPs from both AMD and Intel will be OpenCL capable. Which means that Adobe gets to code once (hopefully, or the model is broken) and it will provide benefit to everyone.

I've bought (and likely will buy more) GPUs for CUDA exclusively...

He meant ~45% of the non-Intel IGP market, being all NV and AMD GPUs, including integrated, per shipping quantities in the most recent quarter.
That doesn't reflect existing marketshare, just currently being sold marketshare.

http://www.xbitlabs.com/news/graphi...ts_Estimate_in_Q2_2011_After_Adjustments.html

And since IvyBridge contains hardware support for OpenCL, going forwards when they give it software support as well, it will mean 100% of future shipping products from Intel and AMD and NV, when Intel stop selling things which don't support OpenCL.
So it will go from 20% with CUDA (NV marketshare), to 100% in the future.
At which point you wonder if some possibly extra performance (if it's faster with CUDA) is worth being limited to 20% of the potential market. YAY OPEN STANDARDS!

Not sure why we have to care about reviews when it's almost certainly faster than CPU, and almost certainly more useful to 80% of the population than CUDA support.
 
Last edited:
Well...then it's time to benchmark CUDA vs OpenCL in Adobe...and get the full picture.

Because that is what matters right?

Performance.

Quality with the performance matters.

Has CUDA gotten better there? When I used CUDA in the past for video encoding, it was bad to put it bluntly.

Excellent performance, terrible encoding. Geniunely curious about that.

(Considering this generation from nVIDIA)
 
Quality with the performance matters.

Has CUDA gotten better there? When I used CUDA in the past for video encoding, it was bad to put it bluntly.

Excellent performance, terrible encoding. Geniunely curious about that.

(Considering this generation from nVIDIA)
CUDA can give you the exact same result, and should give the exact same result. It does not, because devs want to sell people fast broken software. The problem isn't CUDA, it's that GPUs are only blazing fast for the exact same work at certain parts of the process. In other words, x86 CPUs are still prettyawesome.
 
Many people in the world do not have dedicated graphics cards. They only have IGP or APU so expanding support to those people is a GOOD thing.
 
While QuickSync is a great feature, it's more like a stride in the opposite direction: it mainly relies on a fixed function hardware developed for a specific purpose, while the GP prefix in GPGPU means 'general purpose'.

Thats very misleading. Since the GPUs FP units can barely handle 5% of the code that the CPUs FP units can. Hence also why its so relatively easy to make fast. The bar is simply low enough.

This is also why GPGPU technologies like CUDA and OpenCL is a niche and will always be so.
 
So any OpenCL capable GPU. Hardly an AMD coup.

They are the main backer behind OpenCL in the GPU territory, Intel only added it so that there wasn't an opportunity for their APU's to become all around better CPU's because of the integrated GPU's.

As for Adobe, sure it is. Nvidia has put barely any effort into OpenCL performance as a few companies and sectors picked up on Tesla and Cuda. Adobe being one of those companies. So adding OpenCL support means that finally AMD gets GPU assisted support in two of the biggest multimedia editing suites in Photoshop and Premiere.
 
They are the main backer behind OpenCL in the GPU territory, Intel only added it so that there wasn't an opportunity for their APU's to become all around better CPU's because of the integrated GPU's.

Apple demanded, Intel obliged.

As for Adobe, sure it is. Nvidia has put barely any effort into OpenCL performance as a few companies and sectors picked up on Tesla and Cuda. Adobe being one of those companies. So adding OpenCL support means that finally AMD gets GPU assisted support in two of the biggest multimedia editing suites in Photoshop and Premiere.

Dont forget Sony Vegas Pro and Cyberlinks Powerdirector 10.
 
AMDs APP SDK is seriously ahead of Intels software stack. It even supports AVX in compatible CPUs.

2e3a5c9.gif
 
Back
Top