• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

AMD Announces Radeon Rays and Radeon GPU Profiler 1.2 at GDC 2018

That's the problem people are not realizing. Especially those saying "a sold card is a sold card". It doesn't work like that. Without cards in the hands of developers and gamers, AMD's ecosystem will start dying off. And when you have no ecosystem, who's gonna bother developing anything for you? And then you just fall into a neverending downfalling spiral of doom. Cryptomining is a nonsense that will not last for long. Gaming and development has been on the rise from year to year regardless of all the doomsday freakouts news outlets have over PC sales. I'm just shocked that AMD is apparently not aware of this. Cryptomining is a nice short term financial injection, but long term, it'll harm them more than it'll benefit them.
lets not get carried away, 6 months of high prices has had the same affect on nvidia but they both have enough gpus in gamers hands to warrant devs time.
 
FUD aside, what we have now is AMD with ray tracing implemented on top of Vulkan and Nvidia with with ray tracing implemented on top of DX12.
The perfect setup for developers to ignore the technology for the time being, because whatever API they use, it won't work on all cards. Hopefully that will change soon.
 
Open Source is the way to go. Look for example at the Open Source Radeon drivers for Linux -> that was the best decision AMD could have made. FGLRX was ok, but often had weird glitches that dragged on, because of the lack of resources nobody fixed them. The current RadeonSI / AMDgpu is pure poetry -> you can bisect the bug and then report a bug or even propose a fix yourself. This way you can learn something and also contribute to the community -> and that feels great.



Well, invest bucket loads of money into nVidia then . But please, don't come crying here when they decide to discontinue and drop the support for their proprietary code.
Not everytime. Cuda is the best example, in VFX most of the major player are using tech that are exclusive to nvidia because it's so much easier to get support for it. Even though open Cl was made by apple, the support for it was (or is still) so bad that some developers just gave up, or used metal instead.
 
FUD aside, what we have now is AMD with ray tracing implemented on top of Vulkan and Nvidia with with ray tracing implemented on top of DX12.
The perfect setup for developers to ignore the technology for the time being, because whatever API they use, it won't work on all cards. Hopefully that will change soon.
Both of those APIs share a common theme though and I am definitely of the opinion AMD will be working on both dx12 compatibility and consumer level support, they said so ie working with Microsoft, their hardware Is rather important to Microsoft.
 
I'm still cranking away with my GTX 780.... Refusing to pay for these overpriced cards. I see the 1060s dropping in price--$280ish for a three gig, and almost $100 more for a six gig.... Not much of an upgrade though, coming from a 780, definitely not worth the price--even at MSRP.

So, I'll wait til next season I guess.

#Notgivingitup
20180310_175653.jpg
 
Both of those APIs share a common theme though and I am definitely of the opinion AMD will be working on both dx12 compatibility and consumer level support, they said so ie working with Microsoft, their hardware Is rather important to Microsoft.

In what way? Besides being both about graphics, that is.
 
In what way? Besides being both about graphics, that is.
Both expose a similar feature set in a similar way and both are closer to the metal.
They're not cross compatible per say but the code required for the jump is not as extreme as in prior cases.
 
I recently moved a graphics house away from Nvidia to AMD GPU's because of the serious support AMD has been giving to the cinematography and vfx (more the cinematography community tbh) community as of late, and the vastly cheaper prices of GPU's seeing significant drops of TCO.

Those who pah pah this fail to see the difference between Nvidia's stranglehold in GPGPU compute (which is difficult to move away from because your code stops working) from creative industries (who most don't give two shits as long as the specific plugins they need are supported, which has seen a heavy shift away from CUDA due to Apple dropping Nvidia a few years ago).
 
That's the problem people are not realizing. Especially those saying "a sold card is a sold card". It doesn't work like that. Without cards in the hands of developers and gamers, AMD's ecosystem will start dying off. And when you have no ecosystem, who's gonna bother developing anything for you? And then you just fall into a neverending downfalling spiral of doom. Cryptomining is a nonsense that will not last for long. Gaming and development has been on the rise from year to year regardless of all the doomsday freakouts news outlets have over PC sales. I'm just shocked that AMD is apparently not aware of this. Cryptomining is a nice short term financial injection, but long term, it'll harm them more than it'll benefit them.

Here I am for you to see. (Vega 64 liquid from the beginning) And now that the CCs started to waver, all our markets are filled with mildly used Radeons from the panicking miners.
 
Vulkan is open source (successor to OpenGL), Mantle was proprietory hence why it failed.

Revisionist history at its finest, folks. I'm sure the delusionals believe it, too. Is a green logo hypnotising or something?
 
Revisionist history at its finest, folks. I'm sure the delusional's believe it, too. Is a green logo hypnotizing or something?
Not sure what you mean. Are you trying to say Vulkan is not open source or that Mantle wasn't proprietary or a failure?
 
Not sure what you mean. Are you trying to say Vulkan is not open source or that Mantle wasn't proprietary or a failure?

Mantle is/was not proprietary. That was fabricated by Nvidia lovers. Hell, what does AMD use that is? Trueaudio, b/c it's hardware based?

Just b/c AMD developed it with limited partners doesn't make it proprietary. It was clearly stated that Nvidia could develop for it...they sneered (for obvious reasons). Did Nvidia lovers expect AMD to add support themselves? Lololololol
 
Last edited:
Sure it was. Mantle was released only as a part of the AMD Windows driver set, thus not compatible with Nvidia GPU's. https://en.wikipedia.org/wiki/Mantle_(API) See support section.

Incorrect. It can support non-GCN architectures. Nvidia rejected it. Why would Nvidia say they won't support Mantle if it can't be compatible... Lol.

Or I guess AMD just went on stage at events and lied about Mantle being an open API.
 
Incorrect. It can support non-GCN architectures. Nvidia rejected it. Why would Nvidia say they won't support Mantle if it can't be compatible... Lol.
Or I guess AMD just went on stage at events and lied about Mantle being an open API.
Technically, Mantle was built for GCN first. Adopting that to different architectures was not as simple as you would think.
Also, it was never open source and never open. AMD said it would eventually be but things never progressed this far.

Although, I guess Vulkan could be considered what Mantle was supposed to be. But that owes a lot to OpenGL legacy rather than Mantle's technical aspects.
 
Incorrect. It can support non-GCN architectures. Nvidia rejected it. Why would Nvidia say they won't support Mantle if it can't be compatible... Lol.
Please show a reference that backs up your claim please. You are literally the first person I have ever seen state this.
For further reference;
https://linustechtips.com/main/topi...ot-open-source-compatible-only-with-amd-gpus/
https://www.amd.com/en-us/innovations/software-technologies/mantle
https://www.extremetech.com/gaming/...ls-an-open-source-answer-to-nvidias-gameworks
 
Technically, Mantle was built for GCN first. Adopting that to different architectures was not as simple as you would think.
Also, it was never open source and never open. AMD said it would eventually be but things never progressed this far.

Although, I guess Vulkan could be considered what Mantle was supposed to be. But that owes a lot to OpenGL legacy rather than Mantle's technical aspects.

Technically correct, it wasn't ever really finished. It was not locked down or incapable of non-GCN, though. Obviously, if they spent the cash to develop it, then it's not going to be PnP for a competitor, but the API was there for all to use if they wanted to. Even if it were easy Nvidia wouldn't be caught dead with their criminal behavior and arrogance.
 
You are mistaken. Mantle was never open source.

It was never finished. The public SDK was to be released at the end of 2014 ALLOWING NVIDIA AND INTEL TO WRITE DRIVERS FOR IT. AMD already said that many times previously... IDK why there's this huge denial that AMD only made it for themselves. Intel didn't sign on and it was dead (do I have to point out that nvidia wouldn't add support).

If your argument is that it wasn't open, b/c it was still in beta....that's pretty flimsy.

I would link the AMD presentation video, but the link is dead.

I'm tellin ya, revisionist history.
 
Last edited:
The public SDK was to be released at the end of 2014 with the official announcement of it being an open API.
Citation please. You made a claim that is not supported by the information found by a simple search. EDIT; I'll be happy to accept such information from a credible source. Not willing to just take your word for it when the information you offer is contradicted by common knowledge and information easily found.
IDK why there's this huge denial
Factual information. Mantle was never open sourced and was never developed for anything other than a specific range of Windows based AMD GPU's. That makes it closed source and proprietary. It's replacement, Vulkan, is open source and non-proprietary. Perhaps you are getting the two confused?
 
Citation please. You made a claim that is not supported by the information found by a simple search.

Factual information. Mantle was never open sourced and was never developed for anything other than a specific range of Windows based AMD GPU's. That makes it closed source and proprietary. It's replacement, Vulkan, is open source and non-proprietary. Perhaps you are getting the two confused?

No confusion...am I the only one that can read and remember press releases?

Since the video is gone, I guess this will have to do https://linustechtips.com/main/topi...courages-nvidia-and-intel-to-use-it-for-free/
Too bad it links to wccftech LOL, but I can't seem to find the vid.

I guess you can win on a technicality, since it wasn't released, but that wasn't the intention. The industry killed it.
 
No confusion...am I the only one that can read and remember press releases?

Since the video is gone, I guess this will have to do https://linustechtips.com/main/topi...courages-nvidia-and-intel-to-use-it-for-free/

He confirmed that AMD will be releasing a public Mantle SDK this year.
They never did. And SDK is still a far cry from either open or open source.
You are correct in that they promised to have an open API and open source. Eventually. This never materialized.
 
They talk much but i haven't seen anything good on Vulkan who works great.
 
It was never finished. The public SDK was to be released at the end of 2014 ALLOWING NVIDIA AND INTEL TO WRITE DRIVERS FOR IT. AMD already said that many times previously... IDK why there's this huge denial that AMD only made it for themselves. Intel didn't sign on and it was dead (do I have to point out that nvidia wouldn't add support).

If your argument is that it wasn't open, b/c it was still in beta....that's pretty flimsy.

I would link the AMD presentation video, but the link is dead.

I'm tellin ya, revisionist history.
Try digging up the open Mantle spec that everyone could implement. That should clear things up a bit for you.
 
Back
Top