The ARM has unveiled its upcoming CPU and GPU designs of its upcoming pay -generation today and is also introducing the Lumax Compute Subsistm (CSS). Previously, MediaTek and Samsung’s choice CPU and GPU will license the design and do a chipset together. Lumax CSS is a turnkey solution.
To be clear, arm chips will not start selling. Instead, it has created a product-ready implementation for 3Nm semiconductor tumors of multiple founders.
In their own words, its silicone and OEM partners “implementation will be able to use as flexible building blocks, so they can focus on the difference at the CPU and GPU cluster level.”
| CPO | Main benefit | Benefit of performance and efficiency | Ideal use cases |
|---|---|---|---|
| C1-Altra | Main functional work | +25% single-thread display Double-digit IPC gain year-by-year | Large-up estimation, computing photography, material creation, generating A.I. |
| C1 Premium | C1-Altra performance with more field functionality | 35% smaller area than C1-Altra | Sub -Flagship Mobile Segments, Voice Is Assistants, Multitasking |
| C1 favor | Continuous efficiency | +16% continuous operation | Video playback, streaming guess |
| C1-Nano | High electrical | +26% efficiency, using less field | Verable, small form factors |
Therefore, there is no fixed design, however, the new parts are very customized. The new C1-DSU C1-Altra, C1-Promium, C1-Promium and C1-Nano, which we have discussed in another article, enables the design of 1 to 14 CPU cores with three core types. GPU For, Mali-G1 scales 1 to 24 shades.
Lumax CSS platform
The arm has a “secret sauce” in the arm to make Lumax better than the custom chipset design. The new system has a system-level cache (SLC) for interconnect L1 chipset and has a 71% leakage reduction compared to standard RAM design, reducing passive power consumption.
In addition, the new memory management unit (MMU) L1 enables secure and cost-efficient virtualization (commonly used to operate multiple OSs simultaneously on the same computer).
ARM has a power efficient and secure “connective tissue” for Lumax CSS
C1 CPU Compute Cluster receives an average of 30% more performance in six industry benchmarks. For applications such as gaming and video streaming, it is 15% fast on average. For other workloads, things like video playback, web browsing and social media, they are on average 12% more efficient. This is compared with the previous design of the hand.
By focusing on top-end hardware, ARM offers double-digit instructions per Cycle (IPC) on C1-Altra CPU Cortex-X925. The Mali-G1 ultra GPU is 20% in rastering and is twice faster on ray tracing tasks compared to immortal-G925.
The newly introduced Scalable Matrix Extension 2 (SME2) is at the center of hand pressure for a high-device AI display-new CPU is faster up to 5x and can be 3x more efficient than the previous design. In addition, the G1 GPU is 20% faster than the previous pay generation.
ARM’s new hardware brings a large amount of device AI correction
Here are some quotes of key industry partners, including Samsung, Honor and Google:
“In Samsung, we are excited to continue our collaboration with the arm by giving the benefit of Arm’s Compute Subsistam platform to develop flagship mobile products. This partnership can lead us to the boundaries of on-device AIs, smarter and more efficient for our users.” Head of Nose Hi Seong, Vice President and SOC IP Development Team in Samsung Electronics
“According to honor, our goal is to bring premium experiences for more users, especially by our high mid-range smartphones. By giving the benefit of the arm Lumax CSS platform, we are able to deliver easy performance, intelligent AI features and excellent power efficiency that stimulates everyday mobile experiences.” Honor
“SME 2-annual hardware, enables more advanced AI models to operate directly on wide devices like Gemma 3. SMEs continue to scale, they will be able to unite the upcoming Pay Generation of AI features in ecosystems.
https://www.youtube.com/watch?v=w-nvWHMTOCO
Source 1 | Source 2
