Good post! And, I agree with possibly 4 but potentially less than that. Either way, it won’t be anything like the wide variety of solutions that Intel provides, mainly because that “wide variety” is a situation Intel created to maximize profit.
Intel has way, way, too much stuff. But a respectible fraction of that is because they have are very diverse set of customers. Some of the bloat of Intel catalog is wanting to sell everything to everybody.
so Celeron series , a Intel Atom C3000 series , Xeon D series , Xeon W 1000 ( formerly Xeon E3) , Xeon W , and Xeon SP is because there is an extremely broad range of devices can build from small home/business NAS server, medium size business NAS servers , cell phone tower base station service process servers , nad up through multiple million dollar super computers with that line up.
Other stuff like the same die sold as Xeon W 1000 and Core i7 gets into the slippery slope of product segmentation bloat (where flipping features on same die off/on for different folks so they buy different chipsets. )
That is part of the catch-22 for Apple. Intel (and AMD) sell to Apple's systems competitors who all sell way more systems than Macs. So if 25 vendors all use a mid range laptop chip and only sell 100,000 each than is collective better volume than Apple selling 950 M units of the same type ( 2.5M versus just 0.95M) .
In the Mac sub categories that are smaller ( probably sub 100K Mac Pro + iMac Pro) the approach of being able to sell to multiple implementers will work better long term.
I start with the idea that the iPad Pro is benchmarking competitively against some of Intel’s higher end chips and
that wasn’t even the goal...
but it doesn't if look at multicore performance. An i9 in a Mac 27" beats the slop out of Apple's current stuff. The Mac Pro chip pulverizes it. Beats some older "high" end mobile chips Intel has ... perhaps.
they were just progressively iterating iPad performance. And, in most cases, they’re just taking the same processor design of the iPhone and repurposing it with more graphic or compute cores. A12, A12x, A12z
which didn't happen every year. The A12z die is EXACTLY the same as the A12X . They have done nothing in that time period. That is actually closer to Intel's gimmicks of slapping a different name on the same die and making a big deal out of it than some "grand technical prowlness" achievement. The A12Z just has less broken GPU subsections in some of them. ( a huge number of the A12x are sold with the working bit turned off to save a buck to sell the broken ones as "full" deal. Apple diving for extra 'dimes and nickels' . )
In my mind, 2, 3, 4 and 5 are all the same processor/processor iteration like the A and Ax versions.
Like the Ax and A those really aren't going to be the same die. The underlying core microarchitecture implementation could be the same ( same building blocks). So if designed carefully could fully validate 2 as working and then remove independent units out of a "2" to get to a "4" so as to have a smaller mask for a smaller die. How economically do a 4 where 25-30% of the die is just "switched off" so have a uniform top to bottom die would be expensive.
One factor AMD has used to pull out in front of Intel on dekstop is that there desktop version do not even try to do a iGPU. That whole transistor budget is just assigned to "more and better" CPU function. The issue will be long term is that the one size fits all will loose to folks who actually allocate the same transistor budget to actually doing what that specific system is suppose to do well in a better fashion If only have 8 cores and AMD has 12 then they'll beat you on "wide" workloads even if the individual cores bench slower. They just have more stuff. Similar thing with have 2-4 "lower power , low compute " cores. On big, wide jobs will get whipped if other folks have 2-4 "big" cores to your "small" ones.
For laptops sure lots of "mundane time" for a wide span of workloads. On cell phones more than sure ( lots of sitting around doing mostly nothing.)
Apple could stretch one baseline to fit those significant workload desktop cases, but most likley they'll get whipped.
If Apple’s willing to put a 12 core processor in a form factor that usually as 2-4 cores, that says to me that what was “appropriate” in the past due to Intel’s reluctance to compress their i3 through i9 stack may be going out the window. They could have 12 cores, top to bottom for the entire line.
This myopic focus on Intel is way past dubious. AMD is putting 8 cores there now.
https://www.amd.com/en/products/apu/amd-ryzen-9-4900h
Apple is only going to have 8 "big" ones so basically a tie. All that Apple is briing to table is 4 "low end" cores to consume less power. If doing nothing then that is great. If have lots of heavy lifting work to do is it really going to help??????
It runs macOS, Final Cut Pro X and Logic Pro X and other macOS apps. That’s really all any Mac
has to do.
Anything Apple does doesn’t have to be competitive against a generic AMD or Intel system, it just needs to run macOS better than the current comparable Intel Mac versions. If Final Cut Pro runs so fast that you can get twice the work done in the same time, that’s what will matter more than what Intel and AMD are doing. [/quote ]
Lots of stuff on the current Mac Pro
marketing page about benchmarks is multiple platform.
"...
"
the vast majority of that is avialbe on other platforms.
"...
Ditto
That Macs are on some remote island is relatively bogus. The more that Apple tries to drink that kool-aid with their own processor, the more likely that effort will fail long term.
[qoute]
The Mac ecosystem is really just an extension of the iOS ecosystem at this point (AppleTV+, Apple Music, iCloud, etc.)
That is almost exactly the mentality that is going see "let's just throw iPad Pro chips into macs and call it a day" that will probably result in dead Mac Pro in 3-4 years.
www.extremetech.com
www.theverge.com
