Apple to replace Intel and move to ARM - *** Confirmed as "Apple Silicon" ***
- Thread starter UberTiger
- Start date
Permabanned
- Joined
- 9 Aug 2008
- Posts
- 35,708
aw diddums
They could use multiple sockets, ARM chips are far cheaper than Intel, easier to cool etc.
Soldato
Apple and ARM will be designed them, but it’ll be Samsung or TNSC that actually manufacture them. Sure, Apple could flush several hundred million dollars down the drain building and operating their own silicon wafer plant, but why would they want additional overheads?
Unless Apple are friends with Qualcomm again, Intel will still be making the wireless baseband chipsets for Apple.
Commissario
It's all speculation but I don't think they'll keep two different lines of hardware.
Permabanned
- Joined
- 9 Aug 2008
- Posts
- 35,708
Not a chance, they love to streamline everything.
Apple and Qualcomm are ‘friends’ again, well as much as a few $b gives you but Apple will be making their own baseband chips because they bought the business unit from intel.
Has anyone actually tried to make their own monster ARM chip that could be used in a desktop?
Most arm chips are super lower power passively cooled chips because there isn’t a desktop OS or customer base for a big chip to make building one worth it.
How hard can you push a normal snap dragon chip if it has a cooler and proper power delivery?
Has anyone actually tried to make their own monster ARM chip that could be used in a desktop?
Most arm chips are super lower power passively cooled chips because there isn’t a desktop OS or customer base for a big chip to make building one worth it.
How hard can you push a normal snap dragon chip if it has a cooler and proper power delivery?
Thanks for the link, it’s pretty clear they can scale the cores, there’s plenty of evidence for that. I guess the main concern is still single core performance compared to intel, it’s still king for those bursty workloads you see day to day on a computer.
True. It's going to be interesting to see. I'm a power user, my work laptop is a top spec 2018 MBP and they're replaced every 3 years. Mine was given to me in August last year and is a 2018 model (due to how they're bought). So I'll likely end up with a 2021 model in 2020. I think it'll be one of the last Intel ones but I wouldn't be disappointed if it's ARM, as long as the performance is there. I'm quite fed up of the CPU constantly throttling, fans spinning up due to the heat and times the machine feeling like it's bogging down. I'm not saying ARM will cure that, but I'm hoping it'll made substantial improvements on power consumption, battery life and thermals. If the performance even matches that's a good thing.
Soldato
I actually agree, however hard they try though due to the very expensive access to the Mac market people keep the devices longer than ever. Imagine someone who buys a 2020 MacBook pro with the idea they'll keep it 5-7 years, like they did with their 2013 MacBook Pro, now to be told sorry but we aren't going to be supporting the applications or O/S after 2022-23. You'll need to use some form of emulation meaning the speed of your x86 CPU is now significantly reduced, either that or you sell your Mac which has been greatly devalued on the secondhand market due to the very reason you need to get rid of it.
I love the idea of providing competition and making the CPU's of the future, after all x86 is pretty long in the tooth even with all of the upgrades it has had overtime. I just don't think that Apple are the company that will be able to deliver what most people want, affordability, even more so in the current chaos we have going on. If MS pull it off, and get a Windows 10 system which is as fast as x86, then sell the CPU design to Dell/Lenovo/HP etc. then we are all good, but a single brand with tight control on IP and a very restrictive way of operating.
Don
The current 28 core Intel Xeon's in use have a die size of around 698mm2.
The current Apple A13 is 98.48mm2, which includes 2 high performance cores and 4 low power cores as well as a GPU. It's not inconceivable that the "little" cores and GPU could be dropped entirely for a Mac Pro, leaving plenty of opportunity to ramp up the number of high performance cores
https://en.wikipedia.org/wiki/Apple_A13
edit:
basically losing the GPU you could probably have an 8 high performance core chip in the same die size
Additionally they aren't going to be thermally constrained in the way they currently are in Phones and Tablets, so if they respond to higher voltage potentially the clock speed could scale as well.
Permabanned
- Joined
- 9 Aug 2008
- Posts
- 35,708
Got to wait and see what happens. Apple has a plan and sounds like they know which direction they want to take it.
Soldato
Lets say you can squeeze in 6 core, if you dump the GPU, and low power cores, in the same package. You'd need between 4-5 of these CPU's to get a a similar core count, assuming performance per core is equal, which means up to 492mm² of die space in total and that is at 7nm, which means harder to cool than the 14nm of the Intel due to the density of the transistors once you are pushing high clock speeds.
like I said above, I am eager to see what can be achieved but if only Apple get to use them, and sell them it's almost pointless.
Caporegime
Apple will probably make a separate die,for the new laptops. So maybe they will make a new SOC with double the cores of the latest iPad SOC. This way it should be easier for them to scale performance on apps from phone to iPad to laptop. For desktop,they probably could make an SMP motherboard,with different processor cards,each with their own cooling. Latency probably won't be great though.
The A64FX isn't that powerful on its own though. However,its designed for flexibility and scaleability. What Fujitsu did(if you read the technical descriptions of the system),was make a "homogeneous" system. Existing supercomputers are heterogeneous,ie,they use CPUs and GPUs,which can lead to integration problems. So the A64FX is a multicore ARM chip,which is more like a hybrid CPU/GPU.This makes each module very compact too compared to an existing CPU/GPU blade.
Fujitsu spent a lot of time on developing a very efficient fabric to interconnect to connect CPU module together,as these interconnects are what consume a ton of power. For AMD CPUs,IF actually is a big percentage of the power consumption of its Eypc CPUs,even to the extent that it consume as much power as the CPU itself in the 1st generation Eypc CPUs IIRC. Because of this the A64FX can scale to large numbers of modules. It's a very elegant solution.
The A64FX isn't that powerful on its own though. However,its designed for flexibility and scaleability. What Fujitsu did(if you read the technical descriptions of the system),was make a "homogeneous" system. Existing supercomputers are heterogeneous,ie,they use CPUs and GPUs,which can lead to integration problems. So the A64FX is a multicore ARM chip,which is more like a hybrid CPU/GPU.This makes each module very compact too compared to an existing CPU/GPU blade.
Fujitsu spent a lot of time on developing a very efficient fabric to interconnect to connect CPU module together,as these interconnects are what consume a ton of power. For AMD CPUs,IF actually is a big percentage of the power consumption of its Eypc CPUs,even to the extent that it consume as much power as the CPU itself in the 1st generation Eypc CPUs IIRC. Because of this the A64FX can scale to large numbers of modules. It's a very elegant solution.
Last edited:
Soldato
- Joined
- 30 Sep 2005
- Posts
- 16,706
I'm really pleased they are moving to their own SOC. It will takes years to mature though, but I can see a very powerful macbook air in the next five years....one that doesn't throttle.
Soldato
It's worth remembering that Apple computers are less than 5% of the consumer market, so it might not be that easy to convince everyone to re-develop apps, unless there is much more cross-over to the iPad, and thus extending the market.
I'm still sceptical on the ARM chips performance. Apple say they have FCPX already developed and working on their ARM architecture. Let's see it perform a BruceX benchmark with the T2 chips turned off. I'm willing to bet the current Macbook Pro I7s will trounce it for now.
On a separate note. It's so sad that ARM isn't a British owned company anymore. I was a shareholder, and so had a vested interest in its success. It seems British companies are always looking for way out out and sell themselves off to US, Chinese, or in this case Japanese firms.
I'm still sceptical on the ARM chips performance. Apple say they have FCPX already developed and working on their ARM architecture. Let's see it perform a BruceX benchmark with the T2 chips turned off. I'm willing to bet the current Macbook Pro I7s will trounce it for now.
On a separate note. It's so sad that ARM isn't a British owned company anymore. I was a shareholder, and so had a vested interest in its success. It seems British companies are always looking for way out out and sell themselves off to US, Chinese, or in this case Japanese firms.
Caporegime
Blame Apple for Imagination Technologies being Chinese owned,after pulling the stunt they did!
Why with the T2 chips disabled? Surely the as is configuration out of the box is the most representative benchmark.
Hate to be picky but Apples market share isn’t 5% it’s actually between 6 and 8 depending on who you believe.
https://9to5mac.com/2020/01/13/idc-gartner-pc-mac-sales-q4-2019/
While it’s share is ‘small’ (they are actually 4th in a crowded market) they have a huge proportion of the premium segment which is where all the money is in the consumer market. The likes of Dell, Lenovo, HP and Acer can only dream about shipping the volumes that Apple does in this segment at the kind of margins Apple has.
There is very little money to be made in shipping generic cheap laptops and desktops pcs which is what most of the market is made up of.
Soldato
Did IT have much to do with ARM being sold of?
Either way, there was a massive whiff to the whole IT saga and it reeked of inside shenanigans. I'm not even sure the British government are still looking at it unfortunately, so i doubt we'll really know what happened.