AMD Acquiring Another Company To Bolster Its AI Play

[Type44Q]

The days of dGPU are now numbered as people move towards lighter and smaller laptops that are based on low power iGPUs, and with AI and NPUs those low power APUs will be even more feasible for producing AAA quality graphics.

I couldn't disagree more: wait till VR/AR goes fully mainstream; it'll be the most addictive drug the human race has ever gotten a taste of - and that's before we even take into account this increasingly-dystopian world that the urban masses will be wanting to escape from.

Household heaters, at least in northern climes, will be the VR racks themselves.

As the saying goes, "Gradually, then suddenly."

[NeoMorpheus]

Damn, i’m late.

But anyway, glad to see the typical negative shit posted from the usual suspects just because its an AMD related article.

Moving on, AMD it’s going to sell the hardware part of it since they dont want to compete against their partners, unlike Ngreedia.

Personally, observing how their partners (cough microcrap cough) and people (phoronix certified haters) keep betraying them, i’m not sure if such considerations should be respected.

[coder]

AMD,
Just fix your damn drivers (which includes fully supporting devices older than last-gen), and re-instate ZLUDA. Those aren't too much to ask. In fact, proper driver support is the bare minimum. Your hardware is plenty good enough for AI, but nobody wants to buy it because of the abysmal state of your drivers and the looming threat of prematurely dropping support of devices. People are looking for an alternative to Nvidia, and you're not offering one.

AMD can only keep dumping money down this pit for so long, before seeing a return on investment. The "if you build it, they will come" is woefully simplistic and I daresay naive.

I think AMD came to the conclusion that they lack the sales channel and perhaps supporting infrastructure, which limited their uptake at a time when people were willing to wait like 8 months and pay much more for roughly comparable Nvidia hardware. Buying a big integrator gives them the ability to put their products front-and-center in front of customers, right along side competing solutions (i.e. Nvidia). If they can successfully gain some momentum in these markets, they will no longer need to hold this asset and can probably just spin it back out.

[coder]

AMD has been on a spending spree, they spent something like 125 million on that one AI company, then they spent another 665 million in cash on another company and now nearly 5 billion on this.

They bought ATI in 2006, 18 years ago, and despite all this time, for the most part, they are still a few steps behind NVIDIA.

I see no reason to expect that any of these acquisitions are going to be any different.

That's because you don't understand business.

[qlum]

I couldn't disagree more: wait till VR/AR goes fully mainstream; it'll be the most addictive drug the human race has ever gotten a taste of - and that's before we even take into account this increasingly-dystopian world that the urban masses will be wanting to escape from.

Household heaters, at least in northern climes, will be the VR racks themselves.

As the saying goes, "Gradually, then suddenly."

May be wrong here, but I believe neither will take over.
Just like books still exits and tv did not kill radio.
AR may sound nice, but most of the time even if it works flawlessly, people don't always want to use.
Eye / handtracking probably also an inferior input device.
VR while more immersive is also not something people always want, being able to walk away, aware of your surroundings, ignoring the fundamental problem of motion sickness.

I personally feel that while these techniques have a place they are not something to outright replace what we use now.

[schmidtbag]

AMD can only keep dumping money down this pit for so long, before seeing a return on investment. The "if you build it, they will come" is woefully simplistic and I daresay naive.

I would argue you're being naive if you think they're going to make money leaving it as-is and spending their money on assets that depend on a trustworthy platform, which ROCm isn't. The product is broken and incomplete. Despite the fact the hardware is very good for a better price than Nvidia, people aren't buying it. It's not hard to figure out why.
Maybe instead of dumping money down the pit, they should just simply spend it wisely. That would include ditching more managerial positions and hiring more/better developers. AMD has real talent on their team but they've got too much catching up to do.

I think AMD came to the conclusion that they lack the sales channel and perhaps supporting infrastructure, which limited their uptake at a time when people were willing to wait like 8 months and pay much more for roughly comparable Nvidia hardware. Buying a big integrator gives them the ability to put their products front-and-center in front of customers, right along side competing solutions (i.e. Nvidia). If they can successfully gain some momentum in these markets, they will no longer need to hold this asset and can probably just spin it back out.

This isn't the first time they've bought out a 3rd party like this, and it doesn't seem to work. It's just another pit they're throwing money into.

[coder]

I would argue you're being naive if you think they're going to make money leaving it as-is and spending their money on assets that depend on a trustworthy platform,
which ROCm isn't.

It's not either/or. They need to keep investing in their software stack, but they also need the other pieces in place to see a good return on the investments they're making.

This isn't the first time they've bought out a 3rd party like this, and it doesn't seem to work. It's just another pit they're throwing money into.

The reason why this company was so expensive is that they have good revenues. That means they'll be contributing to AMD's bottom line and that AMD could get a decent price if/when it decides to spin them back out.

It's not the same kind of gamble as when they acquire a company for its technology. Like sophisticles, your understanding of business seems extremely simplistic.

[duby229]

AMD can only keep dumping money down this pit for so long, before seeing a return on investment. The "if you build it, they will come" is woefully simplistic and I daresay naive.

I think AMD came to the conclusion that they lack the sales channel and perhaps supporting infrastructure, which limited their uptake at a time when people were willing to wait like 8 months and pay much more for roughly comparable Nvidia hardware. Buying a big integrator gives them the ability to put their products front-and-center in front of customers, right along side competing solutions (i.e. Nvidia). If they can successfully gain some momentum in these markets, they will no longer need to hold this asset and can probably just spin it back out.

The lack of uptake isn't because of system builders or integrators won't sell it's because their customers aren't interested. This buyout isn't going to help AMD sell more products.

I think AMD needs to rethink their software stacks for compute and whatever that happens to be needs to be backported all the way back to GCN1 and then it needs actually tested on each and every single possible product that can support it.

Then people are going to go home and find out that AMD just released a massive GPU compute update that works on everything and supports all the functionality you need. At which point people are going to test it at home when they will over the course of personal experience decide whether they like it or not. When they decide they like it that's when it will blow up in the professional spaces....

AMD had a chance to do this with ROCm, but they blew it. All AMD had to do was enable it on everything and test it everywhere. Get everyone that they possibly could exposed to it. But they didn't and now I think it's too late.

[coder]

I think AMD needs to rethink their software stacks for compute and whatever that happens to be needs to be backported all the way back to GCN1 and then it needs actually tested on each and every single possible product that can support it.

That's silly. The few people still using AMD GPUs older than Polaris are mostly folks not very serious about GPU compute. I'd even say they only need to support Vega and newer, and that's mainly because they kept using VEGA iGPUs in some of their APUs until fairly recently.

The rest of your post is just wishful thinking that having a product which could compete with Nvidia is all it takes. Nvidia has virtually locked up an entire industry. You can't break that stranglehold just by having a couple generations of competitive technology, even at lower prices. To get real market traction, you need to break the doors down and ensure that comparable, fully-integrated datacenter product offerings can reach customers and key decision makers, in these organizations.

[duby229]

That's silly. The few people still using AMD GPUs older than Polaris are mostly folks not very serious about GPU compute. I'd even say they only need to support Vega and newer, and that's mainly because they kept using VEGA iGPUs in some of their APUs until fairly recently.

The rest of your post is just wishful thinking that having a product which could compete with Nvidia is all it takes. Nvidia has virtually locked up an entire industry. You can't break that stranglehold just by having a couple generations of competitive technology, even at lower prices. To get real market traction, you need to break the doors down and ensure that comparable, fully-integrated datacenter product offerings can reach customers and key decision makers, in these organizations.

I totally disagree with you... Top tier management at corporations aren't the ones building the infrastructures and applications that people actually use. Right there, that's the flaw in thinking. AMD seems to think that if only executive managers could force this on their employees and customers then it will just work.

But it won't...

EDIT: AMD employees and customers will be the ones to decide what they want. And that will never happen if they don't have personal exposure and experience to decide it! AMD buying a corporation to force the decision is going to blow up on them! The very people that are expected to be the ones to use it are the very same ones that don't have exposure or experience with it!!

EDIT: It's like going to work and being told that you can't use what you know and love, but instead this half assed buggy software that you've never seen before that only works on one specific hardware configuration that sucks ass.

EDIT: It only works on hardware that virtually nobody has and it obviously hasn't been tested beyond getting it to compileand even then only one specific old version Ubuntu....Seriously, that's like a quarter of a percent of all people....wtf did AMD think was going to happen!?