Apple To Be Parting Ways With Imagination Graphics

[starshipeleven]

I knew Apple was big, but I just noticed their revenues have almost caught up with those of the entire global semiconductor industry

And that's off iPhone sales alone (for like more than 70%, anyway). SUre is nice, but imho that's a bit risky. What if demand drops or market is saturated?

[starshipeleven]

Wow, what an extremely unprofessional response.

I'm so biased that I was totally expecting them to act like dicks. I really hope Apple nukes them in court, and then buys them out.

[starshipeleven]

And they'll opensource the drivers. Or not.

Would it really matter anyway? That's going to be only in Apple's SoCs.
Hacking an iPhone to run Android is completely retarded on so many levels that it's not even funny.

There could theoretically maybe be in a possible future that Apple will be making their own GPUs for their own x86 products, but I somehow think it's a bit too soon to even think about that.

[juno]

Yeah, so I don't think Apple abandoning PowerVR in favor of creating their own GPU has anything to do with Apple wanting to get rid of OpenGL and Vulkan.

Me neither, but I already said that 3 pages ago.
It's just a lucky coincidence and fits in their whole concept.

And that's off iPhone sales alone (for like more than 70%, anyway). SUre is nice, but imho that's a bit risky. What if demand drops or market is saturated?

The Market has already been saturated for years at least in industry nations. Everyone has a smartphone. It's just that almost everyone wants a new one every 1-2 years. A paradigm shift from that won't happen overnight and Apple has huge amount money reserve so they will be able to react.
But I would also like to see more enthusiasm towards the MacOS world. The quality is suffering, there are no outstanding software or hardware features and market share is dropping. Personally, I'm not a fan of MacOS but a stronger Apple would keep reminding people that there are alternatives to WinDOS.

[starshipeleven]

The Market has already been saturated for years at least in industry nations.

FYI, the big sale boost started when they made the larger iP6 that everyone knew would sell like hot cakes in Asian markets.

Everyone has a smartphone.

iPhones aren't just "smartphones", the cater to a specific market segment, afaik that segment in Asia isn't saturated yet. Yes all people have some smartphone, but there are enough people changing it to buy iPhones.

It's just that almost everyone wants a new one every 1-2 years.

Sales of iP in europe or USA aren't anywhere near high enough to justify that boost of revenue.

A paradigm shift from that won't happen overnight and

Market saturation happens relatively fast, it took a few years for normal smartphones in first world countries. If you don't have a plan for it NOW you can't usually cobble together one in time.

Apple has huge amount money reserve so they will be able to react.

Money is necessary but alone it is not sufficient, what worries me is the lack of innovation. They didn't make any particularly noteworthy hardware after iPhones. Even their iWatch things aren't terribly different from any other.
And again you don't come up with a new cash cow overnight.

But I would also like to see more enthusiasm towards the MacOS world. The quality is suffering, there are no outstanding software or hardware features and market share is dropping. Personally, I'm not a fan of MacOS but a stronger Apple would keep reminding people that there are alternatives to WinDOS.

Yeah, this is part of what I was talking about above. For example I sometimes get the odd Apple fan that whines to me about the lack of a Mac Mini with more modern hardware. (last one is from 2014 and the best one has a i7 4xxxU with a HD5000)

This is because it's rather secondary to their main business that is mobile now.

[name99]

And they'll opensource the drivers. Or not.

Way to miss the point.
Apple probably have no interest in creating a traditional GPU --- that's backward looking. What they are likely interested in is creating us a throughput engine to complement their existing latency engine in their Cyclone and subsequent cores. Think something like Larrabee or Cell.
Now obviously Larrabee and Cell both failed in their own ways, but this does not, IMHO, mean that the idea they had was bad, more that their specific implementations were lousy and/or that they were trying too soon, before appropriate transistor density was available. Presumably Apple will drop the lousy parts of these two predecessors (Cell's horribly memory model and ISA that's kinda sorta the same but not the same as the main PPC ISA, Intel's high x86 decode overhead for when what you want are LOTS of small cores, and use of a ring that couldn't scale well to so many cores) but retain the vision --- the ability to move code (same ISA, same address space) as necessary from a latency core to many of the throughput cores and back as appropriate, somewhat like one moves code from the big to the LITTLE core and back as appropriate.

As such, the concept of "drivers" make no sense. Apple may augment the ISA with a few instructions that are especially necessary for GPUs (most obviously texture lookup), but as far as developers are concerned, the app will, if it's low level, call into the Metal library just like today; if it's high level it will call into SpriteKit or SceneKit. Other apps will call into the Accelerate library which will presumably implement appropriate functions (large matrix manipulation, large FFTs, large neuro computations) on top of this sea of throughput cores. Finally APIs like OpenCL and OpenMP will (via LLVM modification) likewise run their kernels on the sea of throughput cores.

[name99]

Wow, so they can finally enforce their "WebMetal" and never mention Vulkan or OpenGL again.

Apple has already published a WebGPU prototype that you can experiment with TODAY.
Meanwhile this mythical WebVulkan that you keep insisting Apple should be using instead remains utterly nonexistent...

[name99]

Clearly Imagination is making a statement that Apple has been poaching their talent and therefore should be put on notice that any internally developed graphics engine should have a healthy dose of skepticism with regards to IP.

On Apple's side, they are saying that "we have been paying you royalties for a product our engineers are providing you more IP on then we get in return" so they are going their own way. This is not unusual since they are poaching Imagination's talent.

Since Apple has been moving more and more to internal integration and shifting away from their reliance on external IP sources, this isn't surprising.

But Apple will have serious uphill battle in graphics IP. Intel ran into the same issues trying to develop their own technology and was forced to cross license.

Can we quit using this inflammatory language of "Apple has been poaching their talent"?
What exactly is your mental model here? That management OWNS engineers and so has a right to be aggrieved when someone else offers them a better job or a higher salary?
Were you up in arms when Tesla "poached" Chris Lattner from Apple? When AMD "poached" Jim Keller from Apple?

[bug77]

Way to miss the point.
...

I'm sorry, I must have given the impression that I care. I know Apple is the right thing for some users, I have no problem with that, but there's no iDevice in my house. And I like it that way.
I was just poking at users that shame PowerVR for being open source hostile.