Apple\u2019s transition to\u00a0its\u00a0own silicon for its Mac products is\u00a0easily\u00a0one of the best things\u00a0ever\u00a0to happen for computing and Mac users.\u00a0I\u2019m old enough to have tested the M1 MacBook Pro when it was released and remember being utterly flabbergasted by its performance and efficiency. I called it the dawn of a new era for Macs, and with the benefit of hindsight, I now know I was right.<\/p>\n The M1 chip was a game-changer, not just for Apple, but for mobile computing as a whole. Pictured here is the M1 MacBook Pro.<\/p>\n Photo: Apple<\/p>\n<\/figcaption><\/figure>\n Fast forward to today, and we are five generations deep into Apple Silicon. The M5 chip made its debut last year, and just weeks ago, we were joined by the M5 Pro and M5 Max<\/b>.\u00a0If you paid close attention to the press release, you\u2019ll\u00a0see that these are\u00a0very\u00a0different\u00a0chips\u00a0from their predecessors.\u00a0Not only are they built on a new Fusion architecture, but they also contain an entirely new type of core called \u201cperformance cores\u201d.\u00a0<\/p>\n This all seems like a huge deal, but Apple didn\u2019t elaborate on them. In fact, in the press release announcing the new chips, you\u2019ll find just a single paragraph talking about the new Fusion architecture. Happily, I had the opportunity to speak to a some executives from Apple to learn more. They were:<\/p>\n Here are my takeaways from\u00a0my\u00a0brief\u00a0chat with them.<\/p>\n To put it briefly, Fusion Architecture involves connecting two heterogeneous dies to form a single SoC<\/b>.\u00a0According to Anand Shimpi from Apple\u2019s platform architecture team, in the case of M5 Pro and Max,\u00a0it\u00a0means\u00a0to connect\u00a0two third-generation 3nm dies,\u00a0using low-latency high-bandwidth die-to-die interconnect,\u00a0to form a single new logical chip.\u00a0<\/p>\n If this sounds familiar, it\u2019s because Intel and AMD both use a similar approach in designing their more recent processors. They call it a \u201cchiplet architecture\u201d.<\/p>\n There are\u00a0a couple of\u00a0advantages\u00a0in\u00a0pursuing\u00a0such an\u00a0approach over traditional monolithic designs.\u00a0Going into detail here\u00a0will\u00a0take a long time, but in\u00a0a nutshell, this approach can\u00a0improve\u00a0design and\u00a0manufacture\u00a0in terms of both\u00a0yields and efficiency.\u00a0Crucially, Anand told me that this approach enables Apple to scale multi-threaded CPU performance, improve GPU performance, and maintain its unified memory architecture \u2013 critical design goals for every Apple chip.<\/p>\n Anand also clarified that the Fusion Architecture in the M5 Pro and Max is fundamentally different from the UltraFusion technology<\/b> used in the M1, M2, and M3 Ultra chips. While they are\u00a0interconnect\u00a0technologies, UltraFusion is used to combine two independent SoCs into a single logical\u00a0<\/p>\n Performance cores are all-new and excel at multi-threaded workloads.<\/p>\n Photo: Apple<\/p>\n<\/figcaption><\/figure>\n With the release of M5 Pro and Max, Apple also introduced a new type of core and changed its nomenclature. It\u2019s a bit confusing, so bear with me. Previously, we had just two types of cores: performance cores and efficiency cores. Now, we have three.\u00a0<\/p>\n Super cores are what used to be known as performance cores<\/b>. These cores offer the best in single-threaded performance. In short, it\u2019s a renaming exercise.\u00a0If you\u00a0go\u00a0pore through the updated spec sheet of the M5 MacBook Pro, it now\u00a0says\u00a0it has 4 super cores and 6 efficiency cores.<\/p>\n Efficiency cores are unchanged and remain the most power-efficient cores that a chip can have. These cores are designed for background tasks and to have low power consumption. Finally, what are now known as performance cores are an all-new design that blends multi-threaded performance and efficiency.\u00a0<\/p>\n When I asked Anand to elaborate\u00a0more\u00a0on the performance cores, he explained that they are \u201cderived from the super core, but they are a unique microarchitecture\u201d<\/b>, and\u00a0that they are\u00a0really\u00a0designed\u00a0for delivering\u00a0great multi-threaded performance,\u00a0while maintaining low power consumption for\u00a0lightly-threaded\u00a0workloads.<\/p>\n To recap:<\/p>\n Having speedy storage can benefit\u00a0numerous types\u00a0of tasks and workloads. And in the M5 Pro and Max MacBook Pros, their SSDs can reach read and write speeds of up to 14.5GB\/s<\/b> \u2013 that\u2019s double the storage speeds of previous-generation MacBook Pros. To achieve such numbers, Doug Brooks from Mac product marketing explained that M5 Pro and Max use a custom controller and firmware.<\/p>\n Additionally, Aaron Coday from the pro workflow team added that improving storage speeds can vastly speed up workflows for professionals. For example, photographers and video editors can enjoy faster import and export times. Having tested the new MacBook Pros, I can certainly attest to this point. Loading massive high-resolution images (think over 100MB) is noticeably faster.<\/p>\n Customers\u2019 needs are at the forefront of every new chip design.<\/p>\n Photo: Apple<\/p>\n<\/figcaption><\/figure>\n According to Aaron, Anand, and Doug, there are a few core tenets that Apple hold dear when it comes to designing chips.\u00a0This has been true when they were\u00a0dreaming up\u00a0M1, and\u00a0it\u2019s\u00a0true to this day when it comes to the latest generation of M5 chips.<\/p>\n One of the most important things that the team looks at is the use cases of customers<\/b>.\u00a0There are system architects at Apple whose jobs are to\u00a0tear apart\u00a0a typical customer\u2019s workload to\u00a0see where the\u00a0bottlenecks\u00a0are\u00a0and how\u00a0that\u00a0might factor into future\u00a0chips.\u00a0Anand said Apple do not build \u201carbitrary silicon\u201d and that the company is clear on what types of chips they are building and what problems they are trying to address.<\/p>\n A corollary to fulfilling customers\u2019 needs is balance<\/b>. This was another point that Anand and Doug mentioned numerous times during my chat with them. It\u2019s incredibly important for Apple to build balanced systems that excel across various tasks and to be power-efficient.\u00a0It\u2019s no point having great CPU performance if your GPU is weak, or have many GPU cores, but not the memory subsystem to support it.<\/p>\n Because of these guiding principles, Apple doesn\u2019t \u201crinse and repeat\u201d as Anand calls it, and simply add more cores and call it a day. The M5 Pro and Max are a testament to that because they feature an entirely different core configuration from M4 Pro and Max. These chips were designed to meet customers\u2019 needs, and that might mean changing certain things up. In summary, with Apple, they are always trying to deliver the most performance in the most power-efficient way possible.<\/p>\n<\/div>\n\n
What is Fusion Architecture?<\/h2>\n
Performance cores are an entirely new type of core<\/h2>\n
\n
Super fast storage<\/h2>\n
Apple\u2019s approach to designing chips<\/h2>\n