hattig

I have a mild suspicion that Apple may have intended to go ARM first with the Mac Mini, but delays in the project resulted in having this massive period of time without an update.

The Mac Mini isn't a Pro machine, and USB Type A is still very useful on a machine - and if you want to differentiate between Thunderbolt Type C and USB ports this is the obvious way to do it.

So I suspect the box will come with two Type C Thunderbolt/DP/USB 3.1, and two Type A USB 3.1, one mere GigE, and an HDMI.

TBH I'd quite like to see a Ryzen 2700U in such a box - a decent CPU coupled with a decent eGPU. Far nicer than the Intel option of a decent CPU and a lame-ass eGPU. But I don't see Apple going this route.

Obviously a discrete GPU would be a nice alternative (Intel's 10nm NUCs will come with discrete Radeon 550s because Intel's 10nm process is horribly broken still and they can barely make dual-core chips, never mind dual-core chips with working eGPUs). Yeah, dual-core, in 2018... I suspect Apple would stick with Intel's 14nm chips to get a quad-core option, and they don't need absolute low-power chips for a Mac Mini - 35W should be fine, but 15-25W will probably be used.

Meltdown affects all Intel x86 CPUs since the Pentium Pro, excluding Atom processors before 2013. It also apparently affects ARM Cortex A75 CPUs, but likely isn't an ISA level issue, so Apple's custom cores are possibly not affected. The solution to this affects performance badly (5-15%, peaks to 30% or more) in some circumstances (server workloads especially, and I/O heavy workloads). AMD processors are not affected. Home users, gamers, and light office users are not likely to notice but benchmarks and user experiences will come out in due course. Sceptre affects all Out of Order processors. Linux has a solution in progress that affects performance up to 1.5%, but the solution involves not just the kernel, but compilers and applications that include compilers (e.g., web browsers).

Bad luck America.

Services are going to be tiered to hell and back as soon as they can do that.

Oh, I'm sure the dross will still be available for all to see, and 'approved content' will be on the cheaper tiers as well.

But most service providers exist in localised monopolies, so without competition they will gauge you if you want streaming services they don't provide themselves, or news sources that aren't favoured, and so on. Oh, you want to do online gaming? That's only available on the $100pm tier.

Kak1 said:

I have a late 2012 Mac and a 2014 Mac book pro. I will update every 6 six years. 

You may find even that is optimistic. Apple's systems (unless you get unlucky, and if its a common fault you will get a free fix anyway) just keep on working.

I'm still on the old 2011 MacBook Pro as a personal laptop, which is six years old now. It's still fine for my personal uses. It did get a free new motherboard a couple of years ago for the motherboard problem, and I added an SSD, 

Sure, my work laptop is a 2015 MacBook Pro, but I don't pay for that. It's not significantly faster because of Intel's interminably slow CPU upgrade cycle, and drive to reduce power rather than increase performance. Screen is great of course.

Intel 6-cores will be a option for the next upgrade hopefully. That might be worth a punt, the combined cumulative upgrade feels worthwhile.

I just want a cheap desktop Mac, and there isn't an option that isn't an iMac. Mac Mini is not what I want, certainly not with the update cycle it has.

sully54 said:

What happens when we hit 1 nanometer? Does chip development end there?

Well we could consider measuring in picometers instead. That's assuming that technology is developed to take processes to that level of miniaturisation.

However there's a real problem in that a silicon atom is 210pm across itself. Carbon is a bit smaller. 1nm processes will likely use graphene.

If you want to know how long process technology takes to mature, the first 5nm transistor was made by NEC in 2003, and the first 3nm transistor was made in Korea in 2006. It takes a while to be able to put billions of them together happily on a chip.

7nm A12 may have 5-7 billion transistors, 5nm A14/A15 may have 8-12 billion. 3nm A16/A17/A18 may have 15-20 billion.