Jump to content

How Apple's M1 uses high-bandwidth memory to run like the clappers


steven36

Recommended Posts

Expandability traded for performance

 

AboRWEj.jpg

 

Apple last week set the cat among Intel's pigeons with the launch of its first PCs incorporating silicon designed in-house.

 

The company claims its M1 Arm chip delivers up to 3.5x faster CPU performance, up to 6x faster GPU performance, up to 15x faster machine learning, and up to 2x longer battery life than previous-generation Macs, which use Intel x86 CPUs.

 

Let's take a closer look at how Apple uses high-bandwidth memory in the M1 system-on-chip (SoC) to deliver this rocket boost.

 

High-bandwidth memory (HBM) avoids the traditional CPU socket-memory channel design by pooling memory connected to a processor via an interposer layer. HBM combines memory chips and gives them closer and faster access to the CPU as the distance to the processor is only a few micrometer units. This on its own speeds data transfers.

 

The M1, Apple's first Mac SoC, is built by chip foundry TSMC using 16 billion transistors with 5nm technology. It includes an eight-core CPU, an eight-core GPU, a 16-core neural engine, storage controller, image signal processor, and media code/decode engines.

 

This Apple diagram of the M1 SoC shows two blocks of DRAM:

 

5S9x0XC.jpg

Apple M1 unified memory architecture

 

The SoC has access to 16GB of unified memory. This uses 4266 MT/s LPDDR4X SDRAM (synchronous DRAM) and is mounted with the SoC using a system-in-package (SiP) design. A SoC is built from a single semiconductor die whereas a SiP connects two or more semiconductor dies.

 

SDRAM operations are synchronised to the SoC processing clock speed. Apple describes the SDRAM as a single pool of high-bandwidth, low-latency memory, allowing apps to share data between the CPU, GPU, and Neural Engine efficiently.

 

In other words, this memory is shared between the three different compute engines and their cores. The three don't have their own individual memory resources, which would need data moved into them. This would happen when, for example, an app executing in the CPU needs graphics processing – meaning the GPU swings into action, using data in its memory.

 

The downside of this design is that expandability is traded for performance. Users cannot simply add more memory to the configuration; they cannot plug more memory DIMMs into carriers as there are no carriers and DIMM technology isn't used.

 

We can envisage a future in which all storage controllers, SmartNICs, and DPUs could use Arm SoCs with a pool of unified memory to run their workloads much faster than traditional x86 controllers, which are hampered by memory sockets and DIMMs.

 

For instance, Nebulon's Storage Processing Unit (SPU) uses dual Arm processors. Conceivably this could move to a unified memory design, giving Nebulon additional power to run its storage processing workload, and so exceed x86-powered storage controllers in performance, cost, and efficiency terms even more than it does now.

 

Source

  • Like 2
Link to post
Share on other sites
2 minutes ago, Sylence said:

is it better than AMD's new Ryzen CPUs that obliterated the entire Intel CPU lineup and beyond?

we need benchmarks. that's what we need

Benchmarks to don't mean nothing  much  since  Apple don't have any native apps much yet   and most everything  has to run under Apple’s Rosetta 2 Developer Transition Kit.  To run any windows  apps you have to use crossover under Rosetta 2  two  layers  of emulation . There no VM support or docker for it yet .  For what support  it does has it runs good   for  a mobile  processor about the same as  a 3  year old  desktop  when compared  to a   a work station that   is 2 or 3 times faster.

Link to post
Share on other sites
23 minutes ago, steven36 said:

Benchmarks to don't mean nothing  much  since  Apple don't have any native apps much yet   and most everything  has to run under Apple’s Rosetta 2 Developer Transition Kit.  To run any windows  apps you have to use crossover under Rosetta 2  two  layers  of emulation . There no VM support or docker for it yet .  For what support  it does has it runs good   for  a mobile  processor about the same as  a 3  year old  desktop  when compared  to a   a work station that   is 2 or 3 times faster.

 

so until then it's only whatever apple says, can't be verified.

laptops based on AMD's mobile Ryzen CPUs should be more powerful and power efficient.

https://www.amd.com/en/products/ryzen-processors-laptop

Link to post
Share on other sites
1 hour ago, Sylence said:

 

so until then it's only whatever apple says, can't be verified.

laptops based on AMD's mobile Ryzen CPUs should be more powerful and power efficient.

https://www.amd.com/en/products/ryzen-processors-laptop

You can look at the benchmarks at https://www.cpu-monkey.com/en/

The ones in the media are Biased  in favor of apple they  act like because they win  just one Geekbench 5 single-core performance against Ryzen 9 5950X that  they done something.  But what about all the other Benchmarks it under preforms against   Ryzen ? dont take no apple fanboys  word for it ,go look for yourself  at cpu monkey were there tested with lots of benchmarks not just 1 :tooth:

 

Quote: Entrope

Every single performance comparison I've seen of M1 versus other CPUs uses Geekbench, either in its single-core or all-core (multi-core) version. This gives an extremely limited basis for comparison. Even when AMD is comparing itself to Intel, while it uses Cinebench as the top-line comparison, it generally provides comparison on other benchmarks as well. Why is Apple focusing so much on Geekbench?

 

-----

 

The reason why  is because  Apple  scores low on Cinebench  and negative PR  dont sell  Macs that dont have any native software . M1 Macs  reminds me of Windows Vista  when  it 1st came out they was no support for it much  and Windows fanboys claim it was the  best windows ever  at the time of release . It didn't get stable tell Windows 7 came out. :lmao:

Apple M1 vs. AMD Ryzen 5 5600X

https://www.cpu-monkey.com/en/compare_cpu-apple_m1-1804-vs-amd_ryzen_5_5600x-1750

 

All the Blender 2.81  software  test i seen against different CPU  . Intel  and AMD  vs Apple M1.   Intel  and AMD leaves Apple in the dust.

Edited by steven36
Link to post
Share on other sites
10 hours ago, Sylence said:

we need benchmarks. that's what we need

 

There's a bunch of M1 benchmarks here.

Link to post
Share on other sites
9 hours ago, Karlston said:

 

There's a bunch of M1 benchmarks here.

Onnly 2  Geekbench 5.3.0 and  Cinebench R23 and when thy 1st posted it was only Geekbench 5.3.0 they updated  it days latter instead making  a new post   and  posting the facts that  Macbook  Air and MacMini are low end machines  Cinebench are better benchmarks  and the latest R23 shows M1 is about  the same as some of the 11 Generation  Intel  Chips the reason Apple started making there own Chips was  they save billions a year. ARS waited tell everyone commented  then updated  and its a very misleading article like most of the articles  that are based on only Geekbench results.

 

cpu-monkey.com has 18  different benchmarks BTW  and M1  scores even worse  on older Cinebench versions.

https://www.cpu-monkey.com/en/benchmarks

 

PS: Only real thing it  has going for it is battery life and new PC laptops last 8 or 9 hours who heck is going to use a laptop for 13 or 14 hours without plugging it in?  But that don't sell me because i'm not a laptop user  i use desktops and the only thing they made was a M1 Mac Mini and it's not upgradeble and only Apple can repair it so its a no go for me. Plus i dont like using  a closed off OS  were  most  the apps cost money and are not even  native apps  yet.  Apple are not dumb they released  the  M1 Macbook  Air and M1 MacMini for consumers they will wait  tell they get most apps ported over to come out with the highend ones for Devs  and Business they said that could take up to a few years .

Edited by steven36
Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

  • Recently Browsing   0 members

    No registered users viewing this page.

×
×
  • Create New...