We’ve seen Google put out job listings for a position that would indicate they wanted to create custom chips, and we have even seen this backed up by additional reports as well. We received confirmation that Google is indeed building custom silicon, but we aren’t told the extent to which Google will customize their own chips (whether it will be custom a CPU, GPU or both). At least we get an idea as to what Google is working on.
Google is taking this Pixel endeavour quite seriously.
At first glance, this would seem to be only mildly interesting. But it may have serious implications in the areas of privacy and open-source compatibility.
Depending on what functions they are customizing, and how much documentation they provide, this could be a good thing, or a very, very bad thing.
the good old “wait and see”.
I certainly hope this isn’t google trying to follow apple and microsoft with their control of device manufacture.
But with these arm SOC manufacturers being so craptastic with driver (and kernel) support it may well be that google feels they have no choice if they want to be able to reliably release security and OS updates.
We can only hope that this is the case: to punish the incompetence of the SOC manufacturers who love to dump designs on the market with zero intention of providing any continuing support.
Edited 2016-10-06 13:58 UTC
Main question – will they develop open drivers for it (especially the GPU)?
Edited 2016-10-06 06:40 UTC
Si is Google becoming a real competitor to the other smartphone manufacturers? The old saying was they don’t really want to, not to scare manufacturers from their ecosystem. Maybe we will see manufacturers looking at alternative OSes or it’s already too late?
I think they might have just decided that they don’t care about other manufacturers in their ecosystem any more, now that every other alternative platform has effectively died.
And then there’s the whole thing where Europe is going after them for incentivizing the other manufacturers to bundle gapps… a solution to that specific case would be to stop supplying gapps altogether, and take all the market share. (Granted, it’d start a whole new case against Google, but…) Think Samsung can really do a successful Android product without the Play Store?
Well, all the competing alternatives (Tizen, Sailfish, WebOS even Windows Phone) are pretty much dead but there are new players in town, most dangerous of them being a forked Android with gapps replaced by the alternate stack pushed so hard by Microsoft.
But the Pixel line won’t get any significant market share any time soon, those are expensive devices sold only in a few areas. It may take years for Google to get significant market share, enough time for competitors to do something.
Samsung can be successful even without Play Store: they can use Microsoft, they can use Amazon, they can grow an own store in a couple of years… Hell, if they grow an own store, they also can resurrect Tizen.
nicubunu,
Maybe, but did you know they already have their own app store? I first used it five years ago. Unbeknownst to me, it had been re-branded to “Galaxy Apps”:
http://www.pcmag.com/article2/0,2817,2460739,00.asp
It is ironic to suggest this as a viable option when, up to this point, it hasn’t been very viable. There are two issues on the app front: 1) convince developers to join them, 2) convince users to switch. Can they do it? I’m skeptical.
Even stores with low fees/no fees haven’t made a big dent:
https://en.wikipedia.org/wiki/List_of_mobile_software_distribution_p…
The Samsung store has already been bundled for years, so what more can they do? Block google play? It would be extremely risky for samsung to resort to coercion. The circumstances are a bit different, but microsoft used extremely coercive tactics, blocked competitors, and spent billions of dollars, and even then they failed to attract a significant mindshare to their app store.
Even though anything could happen, it seems unlikely that google’s dominance will be matched or even displaced without some sort of intervention.
Edited 2016-10-06 14:41 UTC
I don’t own a Samsung device but I heard once, maybe twice, they have their own store, which people don’t have incentive to use when there’s already the Play store.
Anyway, the importance of having app stores is not that big, very few app matter and most people install very little from them.
If that were actually true, though, BlackBerry 10 would be a success.
It’s got an Android runtime, so it can even run whatever .apk you throw at it, after all.
BB10 OS is really great. But the android runtime is android 4.3. I have a Classic and any android app runs much much slower than a cheap android phone. On top of that most apps use functionality from the playstore so you have to hack all that stuff together.
“…Maybe we will see manufacturers looking at alternative OSes…”
This is about “Market Share Confidence”. When minor OEM are building androids, are also leveraging Google mind share.
What if Google says to Them: Boot belongs to me. And this is the strict hardware profile you have to comply with.
Google doesn’t even have to. Instead, instruct them to use a ‘certain’ chip in their assembly.
Can somebody explain me why Google Assistent works only on Pixel? Does it something to do with hardware? But what kind of other hardware is needed than for Google Now? I’m confused.
Better raw performance is quite down in the list of priorities.
Agreed. But there are other benefits that can be had from new silicon. Lower power consumption is pretty high on my list of priorities.
I had the original moto x with it’s custom silicon. I’d rather not wait for nearly a year again while someone ports the latest version of android to my phone.
“…there isn’t a timeline set other than the ‘next few years’…”
OS updates’ problem can’t be kicked down the road anymore [without paying consequences]. So, Another addressing could be taken, in the meantime.
or so it would seem.
IMHO, it will take a long time to create as good a SOC as Apple currently do.
So the question is, are Google prepared to be in it for the long term?
Past history says no. Time for the Chocolate Factory to think long, very long. Please don’t can the effort to produce custom silicon in 1, 2 or even three years like you seem to do with other products. That alone pisses users/customers off no end.
shotsman,
If I had the resources to build my own chips, this would be what I’d try:
Combine a few standard ARM cores along with an FPGA. Think of the possibilities this could open up:
An OS/application programs the FPGA to execute massively parallel application level logic directly in silicon. Even a superscalar SMP processor with dozens of cores would struggle to match the performance. Obviously software and hardware have a symbiotic relationship and both have been emphasizing sequential programming logic. But just imagine what could be done if commodity hardware included FPGAs, it could create a revolutionary new kind of parallel software.
Applications could include more accurate physical simulations, AI algorithms that are too slow on sequential CPUs, better compression, highly flexible SDR/DSP to build advanced networks, etc.
Edited 2016-10-06 21:47 UTC
Like a GPU?