Apple unveils M2 Ultra, a semiconductor chip that maximizes Mac performance with a faster central processing unit (CPU) and graphics processing unit (GPU) and expanded integrated memory support.
Apple unveiled M2 Ultra, a new system-on-chip (SoC) that completes the M2 chip lineup, at its Worldwide Developers Conference (WWDC) 2023, which kicked off today. The M2 Ultra is the largest and most capable chip Apple has ever created.
Built on a second-generation 5-nanometer process technology, the M2 Ultra uses Apple’s UltraFusion technology to connect two M2 Max chip dies to boost performance. It has 134 billion transistors, 20 billion more than the M1 Ultra.
UltraFusion uses silicon interposers that connect the dies with more than 10,000 signals to deliver more than 2.5 terabytes per second of low-latency interprocessor bandwidth. Notably, UltraFusion’s architecture allows the M2 Ultra to appear as a single chip to software. This means you don’t need to rewrite code to take advantage of M2 Ultra’s performance.
More specifically, the M2 Ultra’s 24-core CPU is comprised of 16 next-generation high-performance cores and 8 next-generation high-efficiency cores, delivering up to 20% faster performance than the M1 Ultra. Colorists using DaVinci Resolve in a Mac Studio with M2 Ultra can experience up to 50% faster video processing speeds than with M1 Ultra.
The GPU is configurable with either 60 or 76 next-gen cores. It uses up to 12 more cores than the GPU in the M1 Ultra and increases its performance by up to 30%먹튀검증. Rendering Octane 3D effects on Mac Studio with M2 Ultra is up to three times faster than on Mac Studio with M1 Ultra.
[SEOUL, South Korea – Apple today at the Worldwide Developers Conference (WWDC) 2023 unveiled the M2 Ultra, a new system-on-chip (SoC) that completes the M2 chip lineup. The M2 Ultra is built with an ultrafusion packaging architecture that uses silicon interposers to connect two M2 Max chip dies. (Photo courtesy of Apple)
The M2 Ultra also offers 800GB of system memory bandwidth, a level of bandwidth not readily available on PCs. It can be configured with up to 192GB of integrated memory, enabling workflows that would not be possible on a traditional PC. For example, M2 Ultra can handle large-scale machine learning training jobs on a single system that would be difficult to run on even the highest-performance discrete GPUs.
In addition, M2 Ultra’s advanced customization technology significantly enhances its capabilities for machine learning, video, and more. First, Apple’s latest custom technologies are applied directly to the chip to maximize performance and efficiency, with a 32-core neural engine that processes 31.6 trillion operations per second, making performance 40% faster than the M1 Ultra.
A powerful media engine further accelerates video processing with 2x the capabilities of M2 Max. With professional hardware support for H.264, HEVC, and ProRes encoding and decoding, it can play up to 22 8K ProRes 422 video streams. The display engine supports up to six Pro Display XDRs, delivering more than 100 million pixels.
On the security front, the latest Secure Enclave delivers industry-leading security capabilities with hardware-authenticated secure boot and runtime exploit prevention technologies.
“The M2 Ultra delivers phenomenal performance and capabilities for professional users with ultra-demanding workflows, while still supporting the industry-leading power efficiency of Apple silicon,” said Johnny Thurge, senior vice president of hardware technology at Apple. “The massive performance improvements in the CPU, GPU, and Neural Engine, as well as the massive memory bandwidth, combined in one SoC, make the M2 Ultra the most powerful chip ever built for personal computers in the world.”