Discussion Apple Silicon SoC thread

[Eug]

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:



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:

Page 78 - Discussion - Apple Silicon SoC thread

Page 78 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com

M1 Ultra discussion here:

Page 109 - Discussion - Apple Silicon SoC thread

Page 109 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com

M2 discussion here:

Page 127 - Discussion - Apple Silicon SoC thread

Page 127 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com

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:

Page 215 - Discussion - Apple Silicon SoC thread

Page 215 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com

M4 Family discussion here:

Page 263 - Discussion - Apple Silicon SoC thread

Page 263 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com

 

[Tigerick]

Apple has apparently changed from (per core quadrant)
1 FP16 pipe and
1 FP32 pipe to

1 FP16 pipe
1 "mixed" pipe (which is basically FP16*FP16-> either FP32 or FP16
1 FP32 pipe (which can, under the right circumstances also execute FP16.This is new to A19.)

HOWEVER other infrastructure, most importantly the register cache, has not been changed much.
My assumption is that
- A18 could execute one FP16 and one FP32 in the same cycle. Theoretically interesting but practically useless because not much code uses both resolutions at the same time

- A19 VECTOR code (ie "traditional GPU code") can execute two FP16 instructions in the same cycle (more or less what we see above)

- A19 "special" code (which I assume is matrix multiply) can probably execute three FP16 instructions in the same cycle. The specific details of matrix multiply code (and a tweak in the A19) mean that only src registers per op are required rather than three, so the register cache can send out three sets of two registers to three pipelines. But you'll probably only see this boost on matrix multiply (and possibly requiring a recompile) code.

FP16 gets 85% boost in TFLOPs partly due to 10% boost in clock speed of GPU....

The boost also requires bigger memory bandwidth; that's why Apple is using LPDDR5x-9600 for 6 GPU cores; bump up from LPDDR5x-8533 (5 GPU).

I haven't watched the Geekerwan video; but based on GB6 Metal performance figures, each GPU core get around 43% boost per core. That's mean even with 10 GPU cores, M5's metal performance will surpass M3 Pro's 18 cores.

 

[LightningZ71]

...unless they cut down the number of cores by 2...

They can still achieve a good generational improvement and still save some silicon area while also reducing memory pressure by reducing the number of GPU cores for the base M5.

 

[Tigerick]

...unless they cut down the number of cores by 2...

They can still achieve a good generational improvement and still save some silicon area while also reducing memory pressure by reducing the number of GPU cores for the base M5.

Chances of lower than 10 GPU cores are pretty slim, but it does stress on memory bandwdith. Part of the reason M5 will beat M3 Pro is due to cut down of memory bus by Apple:

• M3 Pro - 192-bit LPDDR5 - 150GB/s
• M5 - 128-bit LPDDR5x-9600 - 153.6GB/s OR
• M5 - 96-bit LPDDR6-14400 - 153.6GB/s

Let's see how Apple opt for M5...

 

[Doug S]

A19P's outsized E core gains, along with bigger SLC and memory bandwidth improvements, give a nice boost to MT. On the GPU side the frame rate gains appear massive, and would be even better in a form factor that has more mass/area to limit throttling.

So an A19P based entry level Macbook is a pretty attractive prospect. I haven't compared the MT and GPU numbers to previous Apple Silicon generations to see exactly where it fits, but it is clearly much faster than M1 Macbooks which apparently Walmart still sells (with 8/256 config) for $599. That price would preserve a big enough gap to make the step up to M5 (plus whatever differences there are in display, build, etc.) worth it for those with need more. I wouldn't totally rule out $549 if they do the Y2K era iMac style translucent plastic case I suggested might appeal to Gen Z's retro sensibilities.

I wouldn't be surprised to see it sold with a fixed 12 GB / 512 GB config. If you need more, you step up to M5 where you have a few RAM/storage options.

 

[Eug]

So an A19P based entry level Macbook is a pretty attractive prospect. I haven't compared the MT and GPU numbers to previous Apple Silicon generations to see exactly where it fits, but it is clearly much faster than M1 Macbooks which apparently Walmart still sells (with 8/256 config) for $599. That price would preserve a big enough gap to make the step up to M5 (plus whatever differences there are in display, build, etc.) worth it for those with need more. I wouldn't totally rule out $549 if they do the Y2K era iMac style translucent plastic case I suggested might appeal to Gen Z's retro sensibilities.

I wouldn't be surprised to see it sold with a fixed 12 GB / 512 GB config. If you need more, you step up to M5 where you have a few RAM/storage options.

Geekerwan was able to get A19 Pro >4000/11000 in a phone. In a MacBook it may hit that without the fan they used, assuming the same clock speed. For MT, that's higher than M2. GPU is probably less of a concern. In fact for both CPU and GPU, it's already faster than the fastest iPad non-Air/Pro anyway so in terms of SoC speed, even A19 non-Pro would be fine. Hell, I could run a Mac with my business applications with the performance of A14 (Geekbench 6.5 - 2200/5400) no problem, but my main concern would be RAM and I/O.

Also, $549 is overly optimistic, especially for 512 GB. I'm thinking $799 retail / $699 education for USB 3 / 12 GB / 256 GB / 13” Retina, or at best $100 less, albeit with third party sale pricing going lower as usual. The M1 MBA is still at Wal-Mart, but Apple doesn't actually acknowledge its existence on its own retail website or even its education website. The only place you can find it at Apple is in the refurbished section. Furthermore even for the M1 at Wal-Mart, $599 is not the retail price. It's their sale price.

Actually are the Ax Pro chips even necessary for USB 3? Could A19 suffice, or is that limited to USB 2? All A19 iPhones are USB 2, but some A19 Pro iPhones are also USB 2.

 

[DZero]

Seeing this... now I am thinking that an 2+6 CPU config in a future AX Pro processor might be a real deal, using a "sealed" version with 2+4 and the full one on the Mac Books. If Apple do not release it the full version for phones too.

I understimated the E cores, Apple has cards on it. Pretty much is something that resembles Pentium M.