CPU, APU & Chipsets - Page 3
Explore the latest news on CPUs, APUs, and chipsets, including updates on Intel, AMD, ARM, and Qualcomm processors, performance benchmarks, and architecture trends. - Page 3
As an Amazon Associate, we earn from qualifying purchases. TweakTown may also earn commissions from other affiliate partners at no extra cost to you.
AMD's next-gen 'Gorgon Point' APUs leaked: Zen 5, RDNA 3.5 refresh expected in 2026
AMD is cooking up a refreshed Ryzen AI series "Gorgon Point" APU family, based on Zen 5 and RDNA 3.5, expected to drop in 2026.
In a special event hosted by LG and AMD for its next-generation Gram laptop, we have AMD's new Gorgon Point APU. The flagship Ryzen AI 9 variant of Gorgon Point will feature 12 cores and 24 threads of Zen 5 processing power, the same core and thread count as the Ryzen AI 9 HX 370 "Strix Point" APU.
Gorgon Point bumps the CPU clocks up to 5.2GHz+ as well as boosting L3 cache up to 36MB (up from 34MB on Strix Point). The flagship Gorgon Point APU will feature 16 CUs of RDNA 3.5 GPU, the same as what ships on Strix Point. We've also got the Ryzen AI 7 variant of Gorgon Point, which will have 8 cores and 16 threads at 5.2GHz+ and 8 CUs of RDNA 3.5, once again, the same GPU core count as its Strix Point APU counterpart.
TSMC will begin accepting 2nm wafer orders starting April 1 with Apple the first customer
TSMC's bleeding-edge 2nm process node technology is expected to enter mass production in the second half of 2025, with Apple being the first customer of new 2nm chips for its upcoming A20 Pro chip for the iPhone 18 Pro and iPhone 18 Pro Max in 2026.
In a new report from Ctee, we're learning that TSMC's first batch of 2nm wafers will arrive at its Baoshan, Hsinchu fab plant in late April, with reservations for the second half of the year to begin on April 1. TSMC will be holding a 2nm production expansion ceremony at its Kaoshiung plant on March 31, too.
Analysts are estimating that TMSC's new 2nm process node is going smoothly, with monthly production capacity expected to reach 50,000 wafers by the end of this year. TSMC's customers for its new 2nm process node are Apple, AMD, Intel, Broadcom, Amazon, and more.
AMD's next-gen Sound Wave leaks: Arm-based APU to compete battle Qualcomm, Intel in 2026
AMD's next-generation Sound Wave APU is reportedly a new Arm-based chip that will feature RDNA 3.5 GPU cores, MALL cache, and compete in the Windows on Arm market against Qualcomm, Intel, and soon NVIDIA in 2026.
In a new video from leaker Moore's Law is Dead, we're hearing AMD's next-gen Sound Wave APU will feature 2 x P-Cores and 4 x E-Cores of Arm-based CPU that won't deliver powerhouse performance, but instead will have fantastic battery life as Windows on Arm laptops do.
AMD is expected to use 4 x GPU cores based on the RDNA 3.5 GPU, but with improved ML (machine learning) performance that could be more of an "RDNA 3.5+" GPU instead. Sound Wave is expected to feature 16MB of MALL cache, a 4th Generation AI engine, a 128-bit LPDDR5X-9600 memory controller and 16GB of RAM expected to be the "standard" for the new Arm-based Sound Wave APU.
AMD confirmed to skip RDNA 4 for Zen 6 APUs: RDNA 4 for discrete GPUs only, RDNA 3.5 on Zen 6
AMD's next-generation Zen 6-based APUs will not be using RDNA 4 GPU cores, and instead will rely on RDNA 3.5 GPU that is used inside of the Zen 5-based Strix Point and just-released Strix Halo APUs.
The news is coming from AMD GPUOpen Drivers code on GitHib with the device ID "Gfxlp12" for RDNA 4 being exclusively reserved for discrete GPUs like the just-released Radeon RX 9070 series graphics cards. AMD's current fleet of Radeon RX 9070 series graphics cards will be the exclusive use of RDNA 4, while next-gen Medusa Point and Medusa Halo APUs using RDNA 3.5 GPU cores.
AMD's next-gen Zen 6 "Medusa Point" and "Medusa Halo" APUs will feature next-gen Zen 6 cores, with Medusa Strix APUs expected to debut with between 12 CUs and 32 CUs of RDNA 3.X GPU, while Medusa Halo APUs will bump the RDNA 3.5 GPU core count up to 48 CUs. Zen 6-based Medusa Point and Medusa Halo APUs will also reportedly have a 384-bit memory bus that's expected to drive performance up 30-50% over the new Strix Halo APU.
Need a top-notch new gaming CPU? AMD's Ryzen 9800X3D is finally widely available at its MSRP
AMD's Ryzen 7 9800X3D now appears to be in plentiful stock at its MSRP at major retailers in the US, finally.
If you recall, when the Ryzen 9800X3D came onto the market last year, it received a positively glowing reception.
In our review, we called the processor the "undisputed gaming champion" and also noted that it's a "capable all-rounder across a variety of workloads." (Remember that its predecessor, AMD's Ryzen 7800X3D, while also a great gaming chip, didn't do so well with other workloads).
AMD's new Ryzen AI Max+ 395 'Strix Halo' APU is 3x faster in DeepSeek R1 AI bench than RTX 5080
AMD's new Ryzen AI Max+ 395 "Strix Halo" APU has been tested in DeepSeek R1 with some impressive results, with the AI benchmark running up to 3x faster than NVIDIA's discrete GeForce RTX 5080 graphics card.
First off, the major difference between the two is that the RTX 5080 has 16GB of VRAM and the Strix Halo APU has up to 128GB of VRAM at its disposal, as well as combining a 16-core, 32-thread Zen 5-based processor with 50 TOPS of AI workload performance with its XDNA 2-based NPU.
AMD ran some benchmarks using consumer AI workloads including llama.cpp-powered application, LM Studio, which is shaping up to be the ultimate support for client LLM workloads, allowing users to locally run the latest language model without any technical knowledge required.
Qualcomm's new Snapdragon G3 Gen 3 SoC is ready for next-gen gaming handhelds
Qualcomm has just unveiled its 2025 lineup of G Series Gaming Platforms, including the new Snapdragon G3 Gen 3, Snapdragon G2 Gen 2, and Snapdragon G1 Gen 2 for a new wave of gaming handhelds.
The new Snapdragon G3 Gen 3 processor is the first Snapdragon G Series chip to support Lumen, Unreal Engine 5's fully dynamic global illumination and reflections system for Android-powered gaming handhelds. Qualcomm promises 30% more CPU performance, 28% faster gaming capabilities, lower power consumption, Wi-Fi 7, and more.
Under that, Qualcomm has the Snapdragon G2 Gen 2 for gaming and cloud gaming at 144FPS on dedicated gaming devices, with 2.3x performance boosts on the CPU, and 3.8x the GPU performance over Snapdragon G2 Gen 1. Lastly, the Snapdragon G1 Gen 2 supports up to 1080p 120FPS over Wi-Fi, intended for cloud gaming on dedicated Android gaming devices, with 80% more CPU and 25% more GPU performance.
Check out this beautiful die shot of AMD's just-released Ryzen 9 9950X3D CPU
AMD has officially launched its new flagship Ryzen 9 9950X3D processor, offering 16 cores, 32 threads, and second-gen 3D V-Cache, and now we've got a beautiful die shot of all that silicon goodness.
In a new post on Chinese social media platform Bilibili, we're getting a look at that gorgeous Zen 5 CPU design with the die of the Ryzen 9 9950X3D shown off, with the beautiful dual CCD configuration and additional 3D V-Cache. It looks so good, and as a self-professed nerd, I totally love seeing these things.
The die design of the new 9950X3D was compared next to the Ryzen 7 9800X3D, where you can see the second CCD on 9950X3D (on the right). Each CCD is home to 8 cores and 16 threads of Zen 5 CPU power, with the second CCD on the 9950X3D bumping things up to 16 cores and 32 threads, with that juicy 3D V-Cache thrown into the mix.
Samsung rumored to have cancelled bleeding-edge 1.4nm process to shift focus
TSMC may soon gobble up more of the semiconductor market as rumors are indicating Samsung has abandoned development on its bleeding-edge 1.4nm process.
At the moment, TSMC holds approximately 67% of the semiconductor market, but according to the latest rumor, that number could soon increase as Samsung has decided to concentrate its efforts on increasing yields from its 2nm GAA node. The news comes from an insider who states Samsung is shifting its attention to its 2nm GAA process to increase yields ahead of the release of its Galaxy S26 series, which should be rocking the foundry's latest flagship SoC, the Exynos 2600. But why wouldn't Samsung be pressing ahead with its bleeding-edge tech?
There are a few reasons why Samsung could cancel its 1.4nm process and the first is production for the Exynos 2600 is scheduled to begin in May, and current reports indicate that trial runs of its 2nm GAA node only secured a 30% yield. Another reason is that Samung's fabrication plant is struggling quite a bit, perhaps more than what is publicly reported, and the decision to focus efforts on 2nm node yields indicates issues with the development of the 1.4nm process.
Intel's new Arizona fab testing 18A (1.8nm-class) wafers, major milestone for Made in America
Intel just announced its new CEO, Lip-Bu Tan, which was a big deal for the company... but another (more silent) milestone was reached: its new 18A process node wafers being tested at its Arizona fab in a major milestone for Made in America semiconductors.
In a hidden post on LinkedIn but shared on X, we see Pankaj Marria, an engineering manager at Intel, saying: "Exciting milestone for Intel 18A! Proud to be part of the Eagle Team, leading the way in bringing Intel 18A technology to life! Our team was at the forefront of running the initial lots rights here in Arizona, marking a key step in advancing cutting-edge semiconductor manufacturing".
He continued: "This achievement is a testament to the hard work, innovation, and dedication of everyone involved. The Eagle has landed, and this is just the beginning! Developed and Made in America the world's smallest node".