Intel Unveils Cutting-Edge AI Capabilities with New Xe2 GPU and Xeon 6 Series Processors

Intel has reportedly announced its latest breed of computer chips, dubbed Xe2 GPU and Xeon 6 Series Pro, at Computex 2024. Intel's newest GPUs and processors look to make artificial intelligence-centered chips available for everyone while being designed explicitly for next-gen computers.

Intel's Xeon 6 processor series offers Efficient-core (E-core) and Performance-core (P-core) choices. The business created these solutions for diverse use cases and workloads. At Computex 2024, it unveiled new E-core CPUs codenamed Sierra Forest. New P-core alternatives, codenamed Granite Rapids, will be available in the third quarter of 2024.

In the first batch, Intel introduced new CPUs for the 6700 and 6900 platforms. The former provides up to 1.4x more memory bandwidth and up to 1.1x more in and out (I/O) bandwidth than previous-generation systems.

Meanwhile, the latter increases inter-socket bandwidth by up to 1.8x, allowing quicker and more efficient communication between system components. Both CPUs support CXL 2.0 (Types 1, 2, and 3).

According to Intel, these processors are ideal for high-density, scale-out workloads, including cloud-native apps, content delivery networks, network microservices, and consumer digital services.

The business claims a 4.2x boost in rack-level performance, a 2.6x increase in performance per watt, a 2.7x increase in 5G-user plane function performance per watt, and a 3.5x increase in next-generation firewall performance per watt over the previous generation.

Intel's Next-Gen Xe2

The Xe2 GPU improves gaming performance by 80% and increases AI throughput more than fivefold over the previous generation. It provides over 60 tera-operations per second (TOPS) and a world-class visual experience.

The NPU 4 can handle up to 48 TOPS, a threefold increase over its predecessor, while consuming less power. Finally, the new IPU is touted to offer a superior camera experience while consuming less power.

Intel's Lunar Lake

Intel's newly released AI chips come alongside the chipmaker's latest laptop chip architecture, dubbed Lunar Lake, which is expected in fall 2024. A new architecture that promises a comprehensive overhaul of Intel's system-on-chip design.

Compared to its predecessor, Lunar Lake improves CPU performance by 14% at the same clock speed. It also has 50% higher graphics performance and up to 60% longer battery life.

The revolutionary system-on-chip architecture triples the size and more than quadruples the performance of the AI accelerator. These improvements make Lunar Lake a serious market rival, with Hallock convinced that it would "definitely" defeat Qualcomm.

One of the most noticeable innovations in Lunar Lake is the incorporation of memory directly into the chip package. This solution, available in 16 and 32GB of LPDDR5X memory, removes the necessity for distinct memory sticks or chips. This improvement decreases power usage for data transportation by around 40%, which helps the chip's overall efficiency.

The Lunar Lake chip also has a novel core configuration with eight cores: four "Lion Cove" Performance Cores (P-cores), which provide 14% gen-on-gen performance increases, and four "Skymont" Efficiency Cores (E-cores), which provide 38% and 68% IPC boosts, respectively.

These E-cores operate as quickly as prior Low Power Efficiency (LP-E) cores while using one-third of the power, resulting in considerable energy savings. The new thread director technology allows Windows to construct "containment zones" that keep the majority of workloads on E-cores, extending battery life.