Processors - Page 18

Apple's new M5 processors will be launching later this year, but the company is cooking up its next-gen M6, M7, and M8 processors with up to a game-changing 256 CPU cores, and an even more impressive 640 GPU cores.

In a new report from industry insider Mark Gurman on Bloomberg, we're hearing Apple is working on several new M-series processors including the M6 (codenamed Komodo), M7 (codenamed Borneo) and also another, more advanced Mac chip that is in development says Gurman, dubbed Sotra (this should be the M8).

Apple's new AI server chips have been rumored for a little while now, with Gurman writing that this could be the first processors "expressly made for that purpose". The new chips will help process Apple Intelligence requests remotely, and then feed the information back to consumers' iDevices.

NVIDIA and MediaTek will have their respective keynotes at Computex 2025, but rumors have persisted for a while now that we'll see a reveal of their collaborative project in codename NX1 and N1: new Arm-based AI PC processors.

NVIDIA CEO Jensen Huang will take the stage with its Computex 2025 keynote on May 19 at 11am local time, with MediaTek CEO Rick Tsai hosting his keynote exactly 24 hours later. The companies are expected to unveil the new N1X and N1 chips, both with Arm-based CPU cores designed by MediaTek, and GPU cores provided by NVIDIA.

In fact, NVIDIA recently unveiled its new GB10 chip for mini AI workstations, with both the N1X and N10 chips expected to be similar to the GB10, but slimmed-down on the CPU side of things. We've been hearing about the NVIDIA x MediaTek AI PC processor for a while now, with issues with the chip even reported causing delays, but it looks like we could have a Computex 2025 unveiling, with a launch in 2026 (most likely at CES 2026 in January).

Intel has officially lowered the price, or updated the "suggested pricing," for two Intel Core Ultra 200S series Arrow Lake desktop processors: the 20-core (8 P-cores + 12 E-cores) Intel Core Ultra 7 265K and the Intel Core Ultra 7 265KF. The Core Ultra 7 265 K's suggested retail price of $399 has dropped to $299, while the Core Ultra 7 265KF's suggested retail price of $384 has dropped to $284.

Intel is targeting PC gamers and creators with this discount, noting that these Intel Core Ultra 200S series CPUs offer "up to 50% faster" gaming performance compared to their 12th Gen Intel Core counterparts, with 65% quicker image rendering. This is a pretty significant $100 price cut that puts the price in line with AMD's 8-Core Ryzen 9700X.

As part of the latest Intel Spring Bundle, those who pick up either CPU will receive a free copy of Dying Light: The Beast and Sid Meier's Civilization VII. This price cut is currently limited to the Intel Core Ultra 7 265K and the Intel Core Ultra 7 265KF, with the flagship Intel Core Ultra 9 285/285K ($549/$589) and mainstream Intel Core Ultra 5 245/245 K/245 K/245KF ($270/$309/$294) remaining unchanged.

Intel announced the enablement of its high-end 14A process node at its Direct Connect 2025 event this week, where we get some juicy new details on the new semiconductor process node.

Intel's freshly-minted CEO Lip-Bu Tan took the stage to reveal an updated Intel Foundry roadmap, adding new Intel 18A derivatives, and announcing the enablement of its high-end Intel 14A process node. The company says that it's already working with partners and 14A, and that it's sharing early versions of the Product Design Kit (PDK), and that so far, customers are reportedly happy with the implementation of Intel's new 14A process node.

The new Intel 14A process node will feature the company's second-generation PowerVia implementation dubbed PowerDirect, a more power-efficient method that focuses directly on supplying and draining transistor power through specialized contacts. This means that with Intel 14A, the company is effectively two entire generations ahead of semiconductor leader TSMC when it comes to employing backside power delivery (BSPD).

Intel has admitted that its latest processors are not what laptop buyers want, but rather Team Blue's older mobile CPUs are currently more popular - to the point where Raptor Lake silicon is running short of inventory.

Tom's Hardware grabbed the scoop on this from an Intel earnings call, which revealed that the turbulent times at the company extend to include problems with new chips for AI PCs - Copilot+ laptops - not selling as well as the company hoped they would.

There's less demand for Lunar Lake and Meteor Lake laptops, as consumers are buying older notebooks with previous-gen CPUs inside. This is happening to the point that Intel's running short of production wherewithal when it comes to older Intel 7 chips - the 7nm Raptor Lake and Raptor Lake Refresh processors.

Intel has provided some more details on its next-gen automotive SoCs codenamed Frisco Lake and Grizzly Lake, which will feature Panther Lake and Nova Lake IP inside, detailed at the recent Shanghai Auto Show in China.

Intel's 2nd Gen SDV SoC is codenamed Frisco Lake and will feature a CPU architecture similar to what will be inside of Panther Lake CPUs, which will be launching later this year with volume production expected in 1H 2026. Frisco Lake will feature Panther Lake IP and core architecture, with TDPs ranging between 20W and 65W, with 10x the AI performance and 61% more efficiency over the previous-gen chip based on the Raptor Cove architecture with 12 cores at between 12W and 45W.

Another big upgrade inside of Frisco Lake is that Intel will be using its 3rd Gen Xe GPU architecture -- Xe3 -- codenamed Celestial. This new Xe3 GPU IP will replace the Battlemage GPU architecture, offering a huge upgrade over the 1st Gen Xe (96 EUs) design used inside current-gen automotive chips from Intel.

Intel has reportedly placed 2nm orders with TSMC for its upcoming Nova Lake CPUs, outsourcing the compute tile to TSMC, confirming rumors from over a year ago that Intel would be indeed using TSMC 2nm for its Nova Lake processors.

In a new report from Taiwan Economic Daily, Intel has reportedly outsourced the compute tile for Nova Lake to TSMC to fab on their new 2nm process node (N2). Intel has been recently bragging about its new in-house Intel 18A process node, yet the company has opted to ask TSMC to make its most advanced processor to date: Nova Lake.

I have been reporting on Nova Lake for a few years now, writing back in January 2024 from rumors all the way back in August 2021 that Nova Lake was the biggest architectural change in CPU architecture since the Core architecture was introduced in 2006. Intel is working on building an entirely new architecture from the ground up much like Ryzen with up to 50% CPU performance improvements from Lunar Lake.

Update - Intel contacted us with this message: "We would appreciate if you could clarify that only the new Boost BIOS settings are covered under warranty. Other warranty terms remain unchanged."

"Overclocking Intel processors with the 200S Boost profile will not void the limited processor warranty provided by Intel. All other warranty terms remain unchanged. This profile does not apply to processors overclocked before the profile launch date."

Original story continues...

Intel is reportedly working on a new LGA 1954 socket that will be used for its next-generation "Nova Lake-S" desktop CPUs, while LGA 1851 will be used for Arrow Lake, and its Arrow Lake Refresh processors.

In some new listings found in recent shipping manifests at nbd.ltd, we're seeing that Intel is working ona new voltage regulator test tool for the new LGA 1954 socket, confirming what we didn't want to hear: LGA 1700 to LGA 1851 didn't last long, and now an entirely new socket.

This means that Intel's next-gen Nova Lake-S desktop processors requiring an LGA 1954 socket will force users to buy a new motherboard alongside their new Core CPU. If you upgraded to an Intel Core i9-12900K for example, you would've purchased a new LGA 1700 socket-ready motherboard at the time... but let's say you upgraded to a new Core Ultra 9 285K a couple of months ago, you needed a NEW socket again: LGA 1851.

AMD has just announced that its next-gen EPYC processor codenamed "Venice" is the first HPC product in the industry to be taped out and brought up on TSMC's advanced 2nm (N2) process technology.

This announcement highlights the strength of AMD and TSMC semiconductor manufacturing partnership to co-optimize new design architectures with leading-edge process technology, with the stars of the show here being the new EPYC "Venice" CPU and 2nm (N2) process node tech.

AMD says that this news also marks a major step forward in the execution of the successful bring up and validation of its 5th Gen AMD EPYC CPU products at TSMC's new fabrication facility in Arizona, Texas, with the company adding "underscoring its commitment to US manufacturing".

Intel's Senior Principal Engineer in Core Design Ori Lempel has revealed the company is already working on the "great-grandchild" of Lion Cove, which will materialize into Griffin Cove.

We know the company has been suffering in its CPU division for many, many generations now, with the real issues kicking off around its 12th Gen Core "Alder Lake" CPUs. But in a recent interview with Intel's Senior Principal Engineer in Core Design Ori Lempel and KitGuru, we're finding out more details of what Team Blue has up its sleeve in the form of Griffin Cove.

Lempel reiterated that Intel has finished its work with Cougar Cove, which will be used inside of its next-gen Panther Lake SoCs which are expected to ship in 2H 2025. It looks like the successor to Cougar Cove will be Griffin Cove, but nothing is official yet... we have heard in rumors from MLID that Griffin Cove would drop in 2027 and beyond with a "Unified Core" that eliminates E-Cores from the desktop CPU.

AMD has confirmed a flaw within its signature verification for microcode updates within Zen 1 to Zen 5 CPUs. This security flaw has been dubbed "EntrySign" and enables attackers to gain kernel-level access.

Last month, AMD confirmed the flaw impacts the first four generations of its Zen CPUs, which include everything from desktop processors to the company's server-based EPYC chips. Now, AMD has confirmed its latest Zen 5 generation is also affected by this security flaw. The crux of the problem can be traced back to AMD's signature verification for microcode updates, which are critical updates AMD rolls out to fix any bugs, improve the stability of the chip, optimize performance, or provide security fixes to vulnerabilities.

The typical order of process is that an operating system or firmware loads the microcode that AMD has signed off on as secure and safe by running it through its verification process, but EntrySign is a vulnerability that enables attackers with ring 0 or kernel-level access to bypass safeguards. What is reassuring is that AMD has already rolled out a fix for the vulnerability to motherboard vendors via the ComboAM5PI 1.2.0.3c AGESA update.

AMD is reportedly gearing up to add "AI" to the name of its new gaming handheld APU, with rumors of the new Ryzen Z2 family of processors seeing a flagship Ryzen AI Z2 Extreme APU.

In a new post on X by leaker Hoang Anh Phu, we're hearing that the Ryzen Z2, Ryzen Z2 A, Ryzen Z2 Go, Ryzen Z2 Extreme, and Ryzen AI Z2 Extreme APUs will take the gaming handheld processor battle to Intel and its Core Ultra 200V "Lunar Lake" series of processors.

Each of the Ryzen Z2 series APUs will have a mix of Zen 3, Zen 4, or Zen 5 processor cores with integrated GPUs that range between RDNA 2, RDNA 3, and RDNA 3.5. The flagship Ryzen AI Z2 Extreme APU will reportedly feature Strix Point silicon with up to 8 cores, and 16 threads of Zen 5/c cores, and up to 16 cores of RDNA 3.5-based GPU... and we can't miss out on that XDNA 2-based NPU for AI workloads.

Intel has officially announced that its next-generation Core Ultra 300 series "Panther Lake" CPUs will be launching in 2H 2025, with a volume supply planned for early 2026, with the new processors made on its bleeding-edge in-house Intel 18A process node.

Intel's next-gen Panther Lake CPUs are the first to debut on its new 18A process node, which is a huge advancement over its previous-gen in-house Intel 3 process node. We have new packaging technologies including BSPDN (Backside Power Delivery) which will move the power delivery process to the backside of the wafer.

Panther Lake CPUs will feature new Cougar Cove P-Cores with Skymont E-Cores, with Intel expected to use 6 P-Cores and 8 E-Cores for mobile platforms, but the mobile GPU is the highlight here: new Xe3 "Celestial" GPU architecture that will reportedly feature up to 12 Xe3 GPU cores packaged in a chiplet standard, ready to compete against AMD's new RDNA 3.5-based Strix Point and Strix Halo APUs.

Intel announced that its bleeding-edge 18A process node has entered "risk production" and its mass production will begin shortly after.

At its recent Intel Vision 2025 event, the company announced that its in-house Intel 18A process node has entered risk production, and that mass production kicks off later this year. Risk production is the stage before mass production, usually involving a limited-scale production to see how the process and its manufacturability and performance are, before it goes into mass production for consumer markets.

The new Intel 18A process node will be first seen with the company's next-gen Core Ultra 300 series "Panther Lake" processors, where it will show off its new semiconductor manufacturing prowess against TSMC.

Apple is working on a next-generation M6 processor that will be found inside of a new iPad Pro in 2027, with a custom in-house C2 or C3 5G modem in 2027.

In a new post from Bloomberg insider Mark Gurman, we're learning that the new M5 processor will enter mass production in 2026 at its new factory in Houston, expected to debut inside of new iPad Pros and MacBook Pro laptops. Meanwhile, the future-gen M6 processor will feature a new in-house C2/C3 5G modem and be ready for release sometime in 2027.

Gurman writes: "Apple's M5 iPad Pro goes into advanced testing ahead of launch this year. The company is planning to make the iPad Pro one of the first devices to include the next-generation M5 chip. The new versions of that model, code-named J817, J818, J820 and J821, are in late testing within Apple and on track for production in the second half of this year. The company has also started early work on M6 versions with in-house modem chips, which are being targeted for 2027".

AMD is reportedly working on releasing its Strix Point APUs to the desktop, under the Ryzen 9000G branding according to the latest rumors.

AMD CEO Lisa Su was recently asked if we'd see the company's new flagship "Strix Halo" APU on the desktop, which she confirmed with "yes, yes, of course!" and now it seems that Strix Point on the desktop will come first with the Ryzen 9000G series APUs.

In a new post on X by leaker @9550pro, we're hearing that there is a new AM5 APU coming and that it should be release in Q4 2025. As it stands, the highest core/thread count we've had on an AMD APU has been 8 cores and 16 threads, but the newer Strix Point APU features 12 cores and 24 threads of Zen 5 processing power (and a far better integrated RDNA 3.5 GPU, and more).

AMD's next-generation consumer-focused Zen 6-based "Medusa Range" desktop CPUs, with new leaks teasing that AMD will be using TSMC's bleeding-edge N2X process node that is made for ultra-high clock speeds on CPUs.

In a video posted by leaker Moore's Law is Dead, we're hearing that "Medusa Range" will be AMD's next-gen Zen 6 desktop processor family, made on TSMC's new N2X process node, meaning AMD is skipping over the new N3 process node, and next-gen N2 process nodes, and right into the (warm arms) N2X node.

We are to expect up to dual TSMC N2X-fabbed CCD chiplets (Medusa Range is expected to feature 2 x 12-core Zen 6 CCDs for a total of 24 cores, 48 threads) with CPU clock speeds targeting a juicy and hype-building 6.0GHz+ and over. We should expect a "decent" integrated GPU, while another variant could arrive with weaker or no integrated GPU.

Intel's rumored to be preparing a fresh take on an Arrow Lake Refresh generation, so we might yet get new desktop CPUs from Team Blue later this year.

As VideoCardz picked up, one of the well-known hardware leakers on X, Jaykihn, posted to inform us that an Arrow Lake Refresh is seemingly back on the table (see above).

The leaker asserts that this new retake on the Core Ultra 200S chips will be different to the previously rumored refresh. (If you recall, speculation was rife about an Arrow Lake Refresh mid-2024, which Intel apparently cancelled).

AMD has been doing a truly amazing job in the APU market, with its new Strix Point and more so its flagship Strix Halo APUs dominating the laptop and handheld chip market... but it looks like the powerhouse Strix Halo APU is coming to the desktop.

In a recent interview with ASUS China General Manager Tony Yu and AMD CEO Lisa Su, Tony requested Lisa to make its Strix Halo APU transition to the desktop as well, with Lisa replying "yes, yes, of course!". AMD makes its Ryzen AI 300 "Strix Point" and "Strix Halo" APU processors for mobile platforms on the FP11 socket, with the AM5 socket housing desktop processors.

FP11 processors are larger than AM5 processors, but we've seen companies like Minisforum using its BD770i and BD790i which allowed a Ryzen 7 7745HX and 7945HX on a custom motherboard that could be used as a regular desktop. It looks like AMD is cooking up version of its Strix Halo APU for the desktop, and that is some exciting news.