CPU, APU & Chipsets News - Page 1

AMD's next-gen "Strix Halo" Zen 5-based APU has been pictured on Chiphell, showing off its chiplet-based deisgn, using 256-bit LPDDR5X memory.

The new Strix Halo APU will feature a chiplet-based design, with one or two Zen 5 CCDs and a huge SoC die that features an oversized integrated RDNA 3+ GPU, and 256-bit LPDDR5X memory controllers not found on the cIOD. AMD is aiming at CPU and GPU performance that competes with Apple's M3 Pro and M3 Max SoCs, at comparable PCB and power consumption.

AMD is using an integrated RDNA 3+ GPU in its next-gen Strix Halo APU, with a huge 40 Compute Units that works out to 2560 Stream Processors, 80 AI accelerators, 40 Ray accelerators, 160 TMUs, and an unknown amount of ROPs, with GPU clock speeds of up to 3.0GHz.

Intel teased its next-gen mobile-focused Lunar Lake CPUs today, but also confirmed its next-gen Arrow Lake-S, Arrow Lake-HX, and Arrow Lake-H processors for both desktop and laptops are coming in Q4 2024.

Intel's new high-performance Arrow Lake CPUs will hit the desktop later this year in Arrow Lake-S form, something that we should see confirmed (and detailed) at Computex 2024 in just a couple of weeks time. Intel has now provided a preview of its next-gen Core Ultra 200 "Arrow Lake-S" desktop CPUs, with a tease of the monolithic design that has multiple tiles inside.

Arrow Lake will feature a CPU, GPU, SoC, and I/O tiles, with the biggest of the tiles being the CPU and SoC which are the larger rectangular-shaped ones at the bottom, and the square one in the middle. There's also a dummy tile used, which is something Intel is also doing with its new Lunar Lake CPUs, too.

Intel has teased its next-generation Lunar Lake CPUs ahead of Computex 2024, teasing the new AI NPU and latest Battlemage "Xe2" integrated GPU.

The next-gen Lunar Lake CPUs will set the foundation for the next-generation AI PC platform, with Intel planning to have the new Lunar Lake CPUs out and in products in Q3 2024. Intel's next-gen Lunar Lake is something truly special: it features all-new architectures across the board with new CPU cores, new GPU cores, new NPU cores, and more thanks to advances in 3D packaging technology and power innovations.

Intel will be using next-gen Lion Cove P-Cores and Skymont E-Cores for Lunar Lake, with some hopefully very delicious IPC improvements. Intel has teased that we'll have 3x the AI performance from Lunar Lake over Meteor Lake, with 45+ TOPS of NPU power for AI workloads. This is just from the NPU alone, while Lunar Lake offers 100+ TOPS of AI performance in total (NPU + CPU + GPU).

AMD's next-gen Zen 5 processors will be launched at Computex 2024 next month, while next-next-gen Zen 6 processors are being teased with 32 cores per CCD.

In some new posts on X from leakers "Kepler_L2" and "InstLatX64" we're discovering the Zen 5 and Zen 6 core configurations, with Zen 5 hitting in just a few weeks time. Zen 5 and Zen 5c cores are expected to both be smaller than existing Zen 4 cores, which means there'll be more room to fire more CCDs on the CPU package.

Intel's next-generation Lunar Lake CPUs are still in the ovens inside of their labs, but now we're hearing about the new Core Ultra 5 238V processor with 32GB of LPDDR5X on-package memory.

If you don't know about on-package memory, Intel's new Lunar Lake CPUs will have 16GB or 32GB of on-package memory, right next to the CPU, GPU, and NPU on the chip. Lunar Lake is for low-power devices with 8W to 30W TDP designs, using Intel's latest CPU and GPU architectures combined with a brand-new AI accelerator.

Intel's next-generation Core Ultra 200 series "Arrow Lake" CPUs have been spotted in new shipping manifestos, in both Arrow Lake-S (desktop) and Arrow Lake-HX (laptop) form.

We've seen new Intel Core Ultra 200 series "Arrow Lake" CPUs in shipping manifestos already, but now we have more of the range: 6-core, 14-core, 16-core, and 24-core variants for both desktops and laptops have been spotted this time. Here's the full list so far:

So we'll have a trio of desktop Arrow Lake-S processors with a 24-core CPU with 36MB of L3 cache, a 14-core CPU with 24MB of L3 cache, and a 6-core CPU with 18MB of L3 cache. On the laptop side of things, we've got the Arrow Lake-HX processors in a 16-core part with 30MB of L3 cache, and a 14-core CPU with 24MB of L3 cache.

Earlier this year, AMD launched its latest Ryzen 8000G Series of desktop processors, which featured built-in Radeon graphics (up to a Radeon 780M), offering PC gamers an entry-level option that doesn't require a discrete graphics card. With the AMD Ryzen 7 8700G, you've got a CPU that can play modern PC games in 1080p without a GPU.

The Ryzen 8000G Series is AMD's first desktop processor lineup to include dedicated AI NPU hardware - with up to 39 TFLOPs of AI processing power. Combining AMD XDNA AI hardware, Zen 4 CPU architecture, and RDNA 3 graphics, you've got a series of AMD chips to suit all workloads.

This week, AMD unveiled two new entries to the lineup: the AMD Ryzen 7 8700F and AMD Ryzen 5 8400F. These variants ship without Radeon graphics and are optimized for efficiency and low-power draw.

NVIDIA and MediaTek are reportedly working on a new Arm-based AI processor, co-developing the new SoC that will be finalized in Q3 2024... just months from now.

The new silicon will support advanced technologies, and will be mass-produced on TSMC's new 3nm process node, and will compete against Apple's new M4 chip, and Qualcomm's upcoming Snapdragon X Elite SoC according to Chinese media outlet UDN.

The site reports that the price of NVIDIA and MediaTek's new Arm-based AI processor will be "as high as $300" adding that this is the first time NVIDIA and MediaTek have cooperated on PC processors, and that it's expected we'll get details on the new AI PC processor during Computex 2024 in just a few weeks' time in Taipei, Taiwan.

DIY gadget YouTuber "Bitluni" has built his own personal RISC-V megacluster, inside of a compact design, check it out:

Overall, the megacluster features 256 x RISC-V-based microcontrollers, operating at a mind-boggling 14.7GHz single-core frequency. Impressive. Remember, these are RISC-V chips, not x86-based processors inside of a regular PC.

The idea behind this is to fit 16 of the superclusters into a single interface, but that created issues down the line, so he designed his own "cluster blade" that featured two microcontrollers mounted on the board, each of them managing the supercluster on top of it. The cluster blade allowed Bitluni to give each supercluster its bus interface.

AMD's next-generation high-end Ryzen AI "Strix Halo" APUs have turned up in new shipping manifestos, featuring up to 64GB capacities and up to 120W TDPs.

The shipping manifestos were noticed by "nbd.ltd" and @harukaze5719, with multiple shipments of the same APU that would've been sent out for evaluation and testing purposes. These manifestos show the Maple DAP reference evaluation platform, which is available in both 32GB and 64GB memory capacities.

The Maple DAP reference systems feature 32GB or 64GB of pre-soldered memory, set by AMD, unlike Intel which is using on-package memory with its next-generation Lunar Lake CPUs.