Discussion Apple Silicon SoC thread

Page 158 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Eug

Lifer
Mar 11, 2000
23,586
1,000
126
M1
5 nm
Unified memory architecture - LP-DDR4
16 billion transistors

8-core CPU

4 high-performance cores
192 KB instruction cache
128 KB data cache
Shared 12 MB L2 cache

4 high-efficiency cores
128 KB instruction cache
64 KB data cache
Shared 4 MB L2 cache
(Apple claims the 4 high-effiency cores alone perform like a dual-core Intel MacBook Air)

8-core iGPU (but there is a 7-core variant, likely with one inactive core)
128 execution units
Up to 24576 concurrent threads
2.6 Teraflops
82 Gigatexels/s
41 gigapixels/s

16-core neural engine
Secure Enclave
USB 4

Products:
$999 ($899 edu) 13" MacBook Air (fanless) - 18 hour video playback battery life
$699 Mac mini (with fan)
$1299 ($1199 edu) 13" MacBook Pro (with fan) - 20 hour video playback battery life

Memory options 8 GB and 16 GB. No 32 GB option (unless you go Intel).

It should be noted that the M1 chip in these three Macs is the same (aside from GPU core number). Basically, Apple is taking the same approach which these chips as they do the iPhones and iPads. Just one SKU (excluding the X variants), which is the same across all iDevices (aside from maybe slight clock speed differences occasionally).

EDIT:

Screen-Shot-2021-10-18-at-1.20.47-PM.jpg

M1 Pro 8-core CPU (6+2), 14-core GPU
M1 Pro 10-core CPU (8+2), 14-core GPU
M1 Pro 10-core CPU (8+2), 16-core GPU
M1 Max 10-core CPU (8+2), 24-core GPU
M1 Max 10-core CPU (8+2), 32-core GPU

M1 Pro and M1 Max discussion here:


M1 Ultra discussion here:


M2 discussion here:


Second Generation 5 nm
Unified memory architecture - LPDDR5, up to 24 GB and 100 GB/s
20 billion transistors

8-core CPU

4 high-performance cores
192 KB instruction cache
128 KB data cache
Shared 16 MB L2 cache

4 high-efficiency cores
128 KB instruction cache
64 KB data cache
Shared 4 MB L2 cache

10-core iGPU (but there is an 8-core variant)
3.6 Teraflops

16-core neural engine
Secure Enclave
USB 4

Hardware acceleration for 8K h.264, h.264, ProRes

M3 Family discussion here:

 
Last edited:

Roland00Address

Platinum Member
Dec 17, 2008
2,196
260
126
It turns out the Mac mini’s USB-A ports are only 5 Gbps, but the hub supports 10 Gbps over USB-A. The SSD is spec’d for 10 Gbps as well.

The Mac Studio may offer no advantage here since its USB-A ports are also 5 Gbps.
I thought of that last night but did not respond for your USB should be in a 4 perhaps the 500 range for 5 Gbps is still 625 MB/s plus the over hang.
 

Doug S

Platinum Member
Feb 8, 2020
2,251
3,481
136
It turns out the Mac mini’s USB-A ports are only 5 Gbps, but the hub supports 10 Gbps over USB-A. The SSD is spec’d for 10 Gbps as well.

The Mac Studio may offer no advantage here since its USB-A ports are also 5 Gbps.


Probably the only reason Apple hasn't dropped USB-A ports entirely are all the legacy devices like mice and USB sticks where performance is irrelevant.

I imagine they will at some point be the first to ship a PC without any USB-A ports, and when they do people will scream bloody murder and claim Apple is somehow going to profit from the move. Then a year or two later Windows PCs will start dropping USB-A ports as well.
 

Eug

Lifer
Mar 11, 2000
23,586
1,000
126
I thought of that last night but did not respond for your USB should be in a 4 perhaps the 500 range for 5 Gbps is still 625 MB/s plus the over hang.
I did more testing with the same drive, this time with Amorphous Disk Mark.

2017 iMac with Kaby Lake Core i5-7600
USB-C - 1034 MB/s
USB-A - 460 MB/s

M1 Mac mini
USB-C - 813 MB/s
USB-A - 364 MB/s

TB4/USB4 hub attached to the M1 Mac mini
USB-C - 744 MB/s
USB-A - 737 MB/s

It's not a huge deal for me, but I can't say I'm impressed. It makes me wonder if the M2 Mac mini will even have USB-A ports. I'm guessing it might have 4 USB-C ports: 2 TB/USB4 ports + 2 USB4 ports.

Probably the only reason Apple hasn't dropped USB-A ports entirely are all the legacy devices like mice and USB sticks where performance is irrelevant.

I imagine they will at some point be the first to ship a PC without any USB-A ports, and when they do people will scream bloody murder and claim Apple is somehow going to profit from the move. Then a year or two later Windows PCs will start dropping USB-A ports as well.
The Mac laptops dropped USB-A even in the Intel years, and the Apple Silicon iMacs have dropped USB-A as well.

Honestly, given these results, I'd rather have 4 USB-C than 2 USB-C plus 2 USB-A. You can always convert USB-C to USB-A with a $3 dongle, but you can't convert 5 Mbps to 10 Mbps.
 

Doug S

Platinum Member
Feb 8, 2020
2,251
3,481
136
it appears there wont be any more Mac updates for this year. Does this mean no M2 Pro/Max until they can leverage N3 or even N3E?

It sure seems like the bigger M2 is waiting for N3. If they are waiting for N3E then I expect we never see M2 Pro/Max and they skip directly to M3 Pro/Max shipping early/mid summer with the non-Pro M3 trailing them instead of leading them to market.
 
  • Like
Reactions: trivik12

Mopetar

Diamond Member
Jan 31, 2011
7,831
5,980
136
I'd rather Apple take their time to have a great chip worth shipping than trying to hit some kind of release schedule with minor incremental upgrades to show for it.

The M1 Max MBP I've got is still running strong and has become one of my favorite tech products of all time. Apple always says that all of their Macs are their best ever Mac, but they were probably right with that one.

If you've got a tough act to follow, no sense in rushing. They'd be better off waiting for the M3 at this point.
 
  • Like
Reactions: ashFTW

eek2121

Platinum Member
Aug 2, 2005
2,930
4,025
136
I wonder at what point they begin to shrink the size of their chips rather than keeping them constant and finding some use for twice as many transistors every 3 years? i.e. if Apple considers 100 mm^2 to be the nominal size for the iPhone SoC over the past decade, do they still plan to make it around the same size for N3, N2 and beyond even if the per wafer price goes to $20K, $25K, $30K over the next few process generations? Having an SoC that used to be a $30 part of your BOM balloon to over $100 would mean raising the price of an iPhone or making big cuts elsewhere (flash keeps getting cheaper and they are no longer doubling the amount of DRAM every other year like in the early days of the iPhone...so maybe?)

Apple has been adding a lot of cache despite its increasingly poor scaling, so an improvement in cache density/scaling (from vertical transistors and BPR) could allow them to slowly cut their chip size over time without compromising their designs.

I don't know about Apple, but after the pricing backlash, I suspect NVIDIA will start to use smaller dies in the future. Gamers will see more modest gains, but lower prices. The GPU industry as a whole will have to go through this phase unless they can get chiplets working for games (> 1 GCD). At some point, prices can't go up anymore, and something will have to give. I suspect we are close to that point.

Apple, on the other hand, still has some room to run pricing up, so I don't expect them to worry for quite a while.
 

DrMrLordX

Lifer
Apr 27, 2000
21,617
10,826
136
I suspect NVIDIA will start to use smaller dies in the future. Gamers will see more modest gains, but lower prices.

Hope you're right, but this is NV we're talking about here. They already bagged on most of their 4000-series lineup in order to move excess 3000-series volume, preferring only to sell a few halo parts using their new uarch. That ought to tell us that NV would prefer to release nothing at all if faced with the prospect of being forced to lower prices.
 
  • Like
Reactions: scineram

Doug S

Platinum Member
Feb 8, 2020
2,251
3,481
136
I don't know about Apple, but after the pricing backlash, I suspect NVIDIA will start to use smaller dies in the future. Gamers will see more modest gains, but lower prices. The GPU industry as a whole will have to go through this phase unless they can get chiplets working for games (> 1 GCD). At some point, prices can't go up anymore, and something will have to give. I suspect we are close to that point.

Apple, on the other hand, still has some room to run pricing up, so I don't expect them to worry for quite a while.


They aren't designing GPUs for gamers, they are designing them for GPGPU and cost is not a big hurdle for that customer base. So long as they have this monster GPU they are selling for thousands a pop to that market, they might as well make a version for gamers who don't care about cost. There will still be cut down versions (or last year's model) available for those who are more cost sensitive.

Apple may have more pricing room with the Mac, but I don't think they can or at least don't want to run up pricing on the iPhone to support ever more expensive SoCs, unless those more expensive SoCs subsume other functions currently in separate chips like a cellular modem.
 

Qwertilot

Golden Member
Nov 28, 2013
1,604
257
126
Hope you're right, but this is NV we're talking about here. They already bagged on most of their 4000-series lineup in order to move excess 3000-series volume, preferring only to sell a few halo parts using their new uarch. That ought to tell us that NV would prefer to release nothing at all if faced with the prospect of being forced to lower prices.

Unless/until the AI people stop demanding the biggest dies they can get, there's no way that NW will stop making the biggest dies they can.

Whether those will then still be able to be priced at levels that gamers might want to afford, I have no idea.
 

Tigerick

Senior member
Apr 1, 2022
649
535
106
NV Gaming GPU's revenue has dropped from all time high of $3.62B @ Q1 to $1.57B last quarter. The GPU market has been shrinking rapidly, so unless NV willing to drop their margin and set the prices aggressively, NV won't be growing like mining time.

Even hot sales of 4090 won't help much unless NV can produce 1 million cards. :p
 

Eug

Lifer
Mar 11, 2000
23,586
1,000
126
Seems like Apple is indeed doing an M2 Max chip with 12 cores ..

M1 Max 8+2 1785 / 12785

If we are to assume the efficiency cores perform at about ~1/3 performance cores, then the performance is equivalent to 8.67 cores or 1475 points per performance core.

If M1 Max were extrapolated to 10+2, it would be 1785 / 15735.
If M1 Max were extrapolated to 8+4, it would be 1785 / 13768.

M2 Max single-core here is 1889. Extrapolating single-core points it would be 1561 points per performance core.

M2 Max 8+4 extrapolation is 1889 / 14570.
M2 Max 10+2 extrapolation is 1880 / 16652.

The score we got is 1889 / 14586.

I know this is almost the mathematical equivalent to handwaving, but if this Geekbench 5 M2 Max score is legit, that would imply an 8+4 configuration (not the 10+2 I had favoured).

Meanwhile, my iMac Core i5-7600 is 1119 / 3808 and my Mac mini M1 is 1768 / 7812.
 
Last edited:

Tigerick

Senior member
Apr 1, 2022
649
535
106
M1 Max 8+2 1785 / 12785

If we are to assume the efficiency cores perform at about ~1/3 performance cores, then the performance is equivalent to 8.67 cores or 1475 points per performance core.

If M1 Max were extrapolated to 10+2, it would be 1785 / 15735.
If M1 Max were extrapolated to 8+4, it would be 1785 / 13768.

M2 Max single-core here is 1889. Extrapolating single-core it would be 1561.

M2 Max 8+4 extrapolation is 1889 / 14570.
M2 Max 10+2 extrapolation is 1880 / 16652.

The score we got is 1889 / 14586.

I know this is almost the mathematical equivalent to handwaving, but if this Geekbench 5 M2 Max score is legit, that would imply an 8+4 configuration (not the 10+2 I had favoured).

Meanwhile, my iMac Core i5-7600 is 1119 / 3808 and my Mac mini M1 is 1768 / 7812.
Apple definitely will double the amount of E cores, as regards P cores, what we seen might be binned version, I guess?
 

Eug

Lifer
Mar 11, 2000
23,586
1,000
126
Apple definitely will double the amount of E cores, as regards P cores, what we seen might be binned version, I guess?
What do you mean? Apple doesn't seem to want to bin its desktop clock speeds, so that leaves core binning. I don't expect more than 12 cores total. If the eff. cores are doubled, then that just leaves 8 cores for perf. We don't expect a 10+4 part. People were expecting either 10+2 or else 8+4, and these numbers imply the latter.


They don't seem interested in competing or taking marketshare away from Raptor Lake/Zen 4. Perhaps they have hit a frequency wall.
Well, as mentioned, Apple doesn't bin on clock speed in their Apple Silicon desktops. I'm sure if they cranked the power they could get single-core up there, but that doesn't align with their marketing preferences.

That said, all bets are off with the Mac Pro.
 

Eug

Lifer
Mar 11, 2000
23,586
1,000
126
The Geekbench listing for the M2 Max looks reasonable, but truth be told, to me this seems early for a leak. Apple early appeared to say we won't see any new Macs until the new year, but typically see the leaked numbers much closer to the release.

Maybe this time it's to dampen expectations, so that a bunch of people will buy M1 Max MacBook Pros and Mac Studios this season instead of waiting 2 quarters?


Did the last Mac Pro have exotic cooling?
No. Just beefy.


The only time I remember Macs with exotic cooling was with the G5 Power Macs and that didn't end well. Lots of reports of leaks after a few years from their water cooling system. Anyhow, it seemed Apple was stuck in a hard place because IBM wasn't ramping up clock speeds fast enough.


Apple M2 Max 12 Cores vs AMD 7700X 8 Cores

View attachment 71964

Funny you happened to chose the Ryzen listing that was called a Mac Pro. :)