The Potential Disruptiveness of AMD’s Open Source Deep Learning Strategy

[Despoiler]

http://instinct.radeon.com/en-us/th...s-of-amds-open-source-deep-learning-strategy/

Deep learning is a disruptive technology like the Internet and mobile computing that came before. Open source software has been the dominant platform that has enabled these technologies.

AMD combines these powerful principles with its open source ROCm initiative. On its own, this definitely has the potential of accelerating deep learning development. ROCm provides a comprehensive set of components that address the high performance computing needs, such as providing tools that are closer to the metal. These include hand-tuned libraries and support for assembly language tooling.

Future deep learning software will demand even greater optimizations that span many kinds of computing cores. In my view, AMD’s strategic vision of investing heavily in heterogeneous system architectures gives their platform a distinct edge.

AMD’s open source strategy is uniquely positioned to disrupt and take the lead in future deep learning developments.

Spot on article IMO. Nvidia's strategy has been to build tools and support their products (good) to lock customers into their hardware, but at the cost of keeping it proprietary. What they do share is never 100% of the picture even what they label as "open source". AMD learned and has been doing a fantastic job playing catch up and in some cases eclipsing Nvidia at what traditionally has been their game. Open source absolutely will them win the day. The smartest people in the room will always want the flexibility to customize to their needs. Especially in competitive environments and industries. The corporate accountant types will look to companies like Nvidia to hand hold them 100% because they just don't get it. As a platform progresses the number of platform experts increases and pushes out the non-technicals. JHH will only be quoting his partner list for so long before it's ever shrinking.

 

[sontin]

So, a guy paid by AMD writes a AMD love letter about their non existence in the Deep Learning market.

The smartest people in the room have used nVidia hardware to get the job done. Only the dumbest guy in the room will waist time to optimize AMD's unoptimized framework.

AMD should just stop posting such nonsense on their website.

Trolling is not allowed
Markfw
Anandtech Moderator

 

[DooKey]

The magical rise of AMD has been forecast for many moons and the terrible fall of Nvidia has been forecast just as many moons. The prophets have been wrong. I don't understand the devotion to companies since they could care less about any of us.

 

[Kenmitch]

Interesting. Time will tell in the end. Seems like open source would have it's advantages over a canned solution.

See the hatred for all things AMD still grows by the day. Not really sure why some members seek and destroy all things AMD.

 

[nathanddrews]

I don't understand the devotion to companies since they could care less about any of us.

They care as much as it takes to sell us products that we want. Should that devotion be going further than that?

 

[richaron]

I don't understand the devotion to companies since they could care less about any of us.

How much less could they care? At the moment all you're saying is their care factor is above zero.

As far as the open source thing goes: it will be the Universities and top end projects who will want to the most transparency and customizability. These are also the ones willing to put the time and/or money into their software. Since they will basically sandwich the market from top and bottom I can imagine their influence will spread.

 

[tviceman]

Open source, proprietary.... doesn't matter if there is only 1 big dog showing up to a fight.

 

[richaron]

Open source, proprietary.... doesn't matter if there is only 1 big dog showing up to a fight.

Are you talking about how AMD is generally accepted as stronger in compute?

 

[frozentundra123456]

Interesting. Time will tell in the end. Seems like open source would have it's advantages over a canned solution.

See the hatred for all things AMD still grows by the day. Not really sure why some members seek and destroy all things AMD.

Maybe it is just a reaction to the hype.

 

[SolMiester]

Are you talking about how AMD is generally accepted as stronger in compute?

Im sure you know exactly to what he is referring too!

 

[OatisCampbell]

All my gaming life I've been hearing about how "ATi/AMD will bury NVIDIA due to factor X!" or "NVIDIA will bury ATi/AMD due to factor Y!".

What I've noticed:

1. No one seems to care about AMDs open source philosophy.

2. Companies only care about NVIDIA's proprietary philosophies because NVIDIA pays them to.

It's doubtful AMD is coming from a 20-30% marketshare to bury NVIDIA. They might get back up to 40% if they release some good products though, and that is enough for me personally. Would be nice to actually have some choices at the only bracket of performance that doesn't elicit yawns.

 

[DeathReborn]

I think it will be about as disruptive as their open Physics thing they did a while back, about as successful too. Google, Intel & Nvidia all went their own way and AMD got left behind so this is about all they can do. They will sell some yes, but not with a great deal of profit or market share penetration, so a mildly noticeable gnat is how I think it will be seen.

I wish them well with their attempts though.

 

[OatisCampbell]

I think it will be about as disruptive as their open Physics thing they did a while back, about as successful too. Google, Intel & Nvidia all went their own way and AMD got left behind so this is about all they can do. They will sell some yes, but not with a great deal of profit or market share penetration, so a mildly noticeable gnat is how I think it will be seen.

I wish them well with their attempts though.

Disagree, but not because of deep learning, consoles, open source.

AMD simply has to offer products that are performance and price and they will get back market share due to brand loyalty, freesync, and people willing to switch brands for a few games.

Most people DON'T have brand loyalty, so really all they have to do is offer comparable performance for $50 less and they should sell about half the cards.

 

[BFG10K]

Spot on article IMO. Nvidia's strategy has been to build tools and support their products (good) to lock customers into their hardware, but at the cost of keeping it proprietary. What they do share is never 100% of the picture even what they label as "open source". AMD learned and has been doing a fantastic job playing catch up and in some cases eclipsing Nvidia at what traditionally has been their game. Open source absolutely will them win the day. The smartest people in the room will always want the flexibility to customize to their needs. Especially in competitive environments and industries. The corporate accountant types will look to companies like Nvidia to hand hold them 100% because they just don't get it. As a platform progresses the number of platform experts increases and pushes out the non-technicals. JHH will only be quoting his partner list for so long before it's ever shrinking.

Sure, sure. Meanwhile AMD bleeds millions every quarter while nVidia rakes in record profits.

According to what we were told about the opening up of Mantle, AMD should have money pouring in, especially from Android which was allegedly going to "embrace" it and get AMD hardware onto that platform. Yeah, look how that turned out.

Just because something is free and/or open-source, it doesn't mean people want it. Case and point Linux on the desktop where it's still ~1% market share even after decades of being free.

 

[Bacon1]

Android which was allegedly going to "embrace" it and get AMD hardware onto that platform. Yeah, look how that turned out.

Guess you missed Vulkan? Thats just the evolution of Mantle. And yes it is sought after, but no, it was never expected to make AMD money. They gave it away for free.

The sad part about AMD is they are focused on doing better by their customers instead of constantly charging them for new hardware. Thats great for end users, but terrible for profits.

 

[OatisCampbell]

Guess you missed Vulkan? Thats just the evolution of Mantle. And yes it is sought after, but no, it was never expected to make AMD money. They gave it away for free.

The sad part about AMD is they are focused on doing better by their customers instead of constantly charging them for new hardware. Thats great for end users, but terrible for profits.

Are you actually saying that AMD doesn't release parts as quickly because they're nice guys and don't want us to spend too much money on computer parts?!
I'm not Lisa Su, but my guess is parts are released as fast as able for as much as possible and the open source things like free sync are just a marketing strategy. (That works out well for us)
AMD is a corporation, everything they do is for profit.

 

[BFG10K]

Guess you missed Vulkan? Thats just the evolution of Mantle.

Vulkan isn't generating any more Android profits for AMD than Mantle did, which is precisely $0.00. AMD is still losing money. That makes Mantle, Vulkan and DX12 a collective financial failure for AMD. They spent money on it and got nothing in return.

And yes it is sought after, but no, it was never expected to make AMD money. They gave it away for free.

AMD isn't a not-for-profit charity. They're a corporate entity with a legal obligation to make a profit for shareholders. You don't do that by giving away IP for free.

The sad part about AMD is they are focused on doing better by their customers instead of constantly charging them for new hardware.

Nonsense, they're just lead by people with a non-existent business acumen. Mantle/Freesync/Crossfire/HBM all should've been licensed, just as nVidia collects fees from every single gsync monitor sold, and every single SLI motherboard with "certification".

They also should be locking out nVidia GPUs from their chipsets unless nVidia pay "certification" fees, exactly how nVidia locks out paying customers from PhysX when they detect a non-nVidia GPU in the system. Clearly nVidia's customers appreciate such tactics given they continue to make a profit. So now it's time for AMD to start cashing in.

Instead AMD expends resources and then hands the result to competitors for free. Now they're doing the same with this deep learning.

The definition of insanity is to keep repeating the same thing over and over, yet expect a different result.

 

[William Gaatjes]

Sure, sure. Meanwhile AMD bleeds millions every quarter while nVidia rakes in record profits.

According to what we were told about the opening up of Mantle, AMD should have money pouring in, especially from Android which was allegedly going to "embrace" it and get AMD hardware onto that platform. Yeah, look how that turned out.

Just because something is free and/or open-source, it doesn't mean people want it. Case and point Linux on the desktop where it's still ~1% market share even after decades of being free.

Andriod can use vulkan which has its origins from mantle.
https://developer.android.com/ndk/guides/graphics/index.html


AMD does not have a gpu that is low power enough to be used in mobile products like smartphones and tablets.
There are many gpu alternatives already available that can make use of modern shaders techniques.
I do not think it is smart for AMD to start investing into a mobile solution until android becomes a serious gaming platform that needs for example vega derivates.
Nvidia has made some attempts but up till now there is not much traction with the exception of the nintendo switch (Tegra based).
I wonder how much the margins are for a smart phone, to be honest.

 

[Flapdrol1337]

They also should be locking out nVidia GPUs from their chipsets unless nVidia pay "certification" fees, exactly how nVidia locks out paying customers from PhysX when they detect a non-nVidia GPU in the system.

I don't think that would work. And it would destroy the goodwill amd built up.

 

[OatisCampbell]

The definition of insanity is to keep repeating the same thing over and over, yet expect a different result.

That's not the definition of insanity, but might be the definition of "obdurate".

It's hard to say whether AMD's free/open source strategy is "right" for them.

They haven't had the market share/branding power NVIDIA has for a long time, so it's possible that putting a cost on their tech drives more people to NVIDIA. (and that the free tech sways some who would buy NVIDIA to AMD, very possible FreeSync savings shift some to AMD)

Also, they may feel the ROI for the proprietary approach isn't there, that the money NVIDIA has spent on things like PhysX, 3DVision, etc isn't the cause of their leading market share. They might think it's more a matter NVIDIA being first to market with leading performance, and they may just not have the budget to do it.

My guess would be lack of cash is the main reason AMD takes more time to get to market and doesn't pay companies to do things like their adaptive sync, Mantle, etc..

 

[bystander36]

All I see is AMD trying to do the best they can with limited funds. Rather than trying to compete with a proprietary solution, which would take a lot more man power, they are hoping to get enough companies involved to do much of the work for them. It might be a good strategy for them.

 

[frozentundra123456]

Open source sounds noble and good, but perhaps it requires a strong company with its resources and leadership to effectively develop a new ecosystem, even if the end user has to pay for it. After all, most managers are risk averse, and it certainly is safer to use a proven system developed by a single, strong company, even if one has to pay for it.

 

[Illyan]

I hope AMD builds up this area and is able to compete, but right now nvidia is the only viable option for any serious research/scientific computing.

 

[richaron]

... but right now nvidia is the only viable option for any serious research/scientific computing.

I'm going to have to disagree with that. Granted nV got into the market first and therefore has a much greater tool base and user base - both very important concerning viability in certain situations - but I think mind share may be clouding your judgement if you think nV is the "only viable option".

 

[IllogicalGlory]

A lot of AMD's efforts have potential.