Samsung has reportedly produced the world's first "standalone" DRAM module that uses a process tech below 10nm. Samsung's New Sub-10nm DRAM Technology To Boost Densities & Uses New Materials For long, the DRAM industry has relied on a 10nm process technology for producing integrated circuits. 10nm DRAM technologies range from 1x, 1y, 1x, 1a, 1b, 1c, and 1d. Now, Samsung is working on a brand new 10a process technology for DRAM that goes below the official "10nm" process limit. Samsung Electronics has produced the world's first single-digit nanometer DRAM working die. It is reported that the company plans to rapidly […]

Read full article at https://wccftech.com/samsung-breaks-10nm-dram-barrier-new-4f-cell-structure-boosts-density/

DeepSeek V4 is out, bringing major optimizations, including up to 1.6T model sizes, and NVIDIA is ready with Day-0 support on Blackwell GPUs using NVFP4. NVIDIA Blackwell NVFP4 Architecture Delivers Major Speed-Ups In DeepSeek v4 With More Optimizations On The Way With the launch of DeepSeek V4, we saw some major optimizations in compute & memory requirements. The updated AI model uses just 27% of single-token inference FLOPs & 10% of the KV cache when running a one-million-token context window. Two new models were also introduced, one being a Pro model with a parameter size of 1.6T, and a Flash version […]

Read full article at https://wccftech.com/nvidia-beats-everyone-to-deepseek-v4-day-0-blackwell-support-pushing-3500-tokens-on-1-6t-models/

Think of DRAM, but in a NAND-like structure, that's the basic concept of 3D X-DRAM, a revolution for the memory markets, bringing higher densities for AI. 3D X-DRAM Is Now Closer To Reality, An HBM-Replacement That Offers Higher Densities For AI In 2023, US-based NEO Semiconductor announced its brand new project called 3D X-DRAM, which was going to address the DRAM capacity bottleneck by leveraging a 3D NAND-like architecture. The company also unveiled two 3D X-DRAM cells, which will be integrated into memory solutions based on 3D X-DRAM. These include 1T1C and 3T0C DRAM cells, offering up to 512Gb, a […]

Read full article at https://wccftech.com/3d-x-dram-hits-proof-of-concept-hbm-replacement-10x-density-high-yield-memory/

AMD's next-generation SP8 & SP7 sockets, designed for the Zen 6-powered EPYC Venice & Verano CPUs, have been detailed. AMD SP8 & SP7 Sockets Are Bigger Than Their Predecessors, Showcasing The Compute Density Increase For Next-Gen EPYC Venice & Verano CPUs Last year, AMD announced its EPYC Venice & EPYC Verano CPUs. The former comes with up to 256 cores in Zen 6C flavors with a launch scheduled for 2026, while the latter is a cost-effective Zen 6 offering that's slated for a 2027 launch. Now, Taiwanese component manufacturer HELM Technology has revealed the new socket, each featuring a larger […]

Read full article at https://wccftech.com/amd-sp8-socket-for-epyc-venice-sp8-socket-for-verano-cpus-detailed/

Intel is making some big changes to its desktop gaming roadmap to deliver faster performance, with new hardware & software technologies. Intel Says They Want To Match AMD X3D CPUs By Changing Their Entire Gaming Desktop Strategy Over the Next Five Years In an interview with PCGameshardware, Robert Hallock has shared the plans of the client CPU team covering desktops, laptops, and handheld PC markets. Robert has been giving us some hints of their strategy over the past few weeks, and it looks like Intel is on the right path. A key change in Intel's desktop strategy came around after […]

Read full article at https://wccftech.com/intel-rewrites-five-year-desktop-gaming-roadmap-betting-on-software-big-time/

TSMC's key supplier has stated that no technology was leaked to China, but it also filed a lawsuit against its former General Manager. TSMC Supplier Denies That Key Technology Was Leaked To China, But Files A Lawsuit Against Its Former General Manager GPTC (Grand Process Technology Corporation), which is a key supplier for TSMC's Advanced Packaging solutions such as CoWoS, has denied recent reports that key TSMC technology was leaked to China. In a recent press release, the company confirmed that no technology leaks were observed after recent rumors emerged. However, at the same time, the company has filed a […]

Read full article at https://wccftech.com/tsmc-supplier-drags-ex-general-manager-to-court-denies-any-tech-reached-china/

Intel's next-gen Xeon "Diamond Rapids" CPUs will offer up to 512 cores, while Coral Rapids will bring back SMT on 8-channel platforms in 2028. Intel Diamond Rapids 16-Channel Slips Into 2027, Features Up To 512 Cores Intel Diamond Rapids "Xeon" CPUs were going to launch this year, but delays in plans have pushed it to 2027. The delay can be attributed to several reasons, such as yields and the fact that the 8-channel line was cancelled. Now, Intel plans to launch Diamond Rapids "Xeon" CPUs in 2027. As per Jaykihn, Intel's mid-2027 plans for Diamond Rapids include a volume launch […]

Read full article at https://wccftech.com/intel-diamond-rapids-xeon-2027-512-cores-16-channel-memory-coral-rapids-smt-in-2028/

Intel's upcoming GPU plans have been disclosed, highlighting Xe3P without discrete gaming parts, Xe4 in 2027 & a next-gen Xe architecture in 2028. Intel Xe GPU Plans: Xe3P In iGPU & AI Discrete Flavors, Xe4 In 2027, Xe-Next In 2028 Intel is working on three next-gen Xe GPU architectures, which will form the basis of its graphics lineup in the years ahead. At the same time, the lineups include both integrated and discrete GPUs; the discrete gaming lineup is missing out on the action. Intel Xe3P GPU Architecture In 2026 Starting with the first Xe microarchitecture, we have Xe3P, which […]

Read full article at https://wccftech.com/intel-xe3p-celestial-discrete-gaming-gpu-cancelled-xe4-druid-2027-xe-next-2028/

Qualcomm is rumored to be preparing a brand new datacenter CPU, which will be just in time to power growing Agentic AI needs. Rumors that Qualcomm's New Datacenter CPU is just a few months away may not sound crazy, given the demand there is for agentic AI There's a rumor going around that Qualcomm is working on its very own "dedicated" Datacenter CPU based on the Arm architecture. Qualcomm making its own Datacenter CPU at some point was expected, but what the new rumor is suggesting is that we could see that chip being announced as early as June this […]

Read full article at https://wccftech.com/qualcomm-datacenter-cpu-launch-in-june-as-agentic-ai-goes-in-hyperdrive-mode/

Meta has partnered with Amazon's AWS to bring tens of millions of Graviton CPU cores to its AI compute portfolio for Agentic AI. The CPU Is The New GPU For Agentic AI - Meta is adding millions of Amazon AWS Graviton CPU Cores To Its AI Infrastructure. As the Agentic AI era rages on, companies are rapidly expanding their AI infrastructure to meet rising compute demands. We have seen multi-GigaWatt deals being signed here and there, and CPU usage is on the rise. With all of this happening, Meta has also announced its blockbuster partnership with Amazon's AWS. The new […]

Read full article at https://wccftech.com/meta-is-adding-tens-of-millions-of-aws-graviton-cpu-cores-to-its-compute-portfolio/

Intel announced that its ZAM memory project, which is being developed with SoftBank's subsidiary, SAIMEMORY, has received a big boost from Japan. Japan Accelerates The 3.5-Year Development Plan of ZAM, A Memory Revolution By Intel & SoftBank In its latest presser, Intel Kabushiki Kaisha (Intel K.K.) and SoftBank Corp. subsidiary SAIMEMORY have unveiled that Japan's NEDO (New Energy and Industrial Technology Development Organization) has now selected ZAM, a next-generation memory standard which is being seen as an HBM replacement. With the selection of ZAM by NEDO, the program will fund the project through government subsidies, accelerating its development to address […]

Read full article at https://wccftech.com/intel-revolutionary-zam-memory-receives-big-boost-from-japan/

AMD EXPO 1.2 is launching soon, and with it, some new features, along with support for Chinese DDR5 memory, are being added too. AMD EXPO 1.2 Enables Support For Chinese DDR5 Memory Makers To Tackle DRAM Shortages & Increasing Prices New information about EXPO 1.2 comes from 1usmus, the author of HYDRA, CTR, and DRAM calculator for Ryzen, & leaker chi11eddog. The new information covers various aspects of the new memory tech, which is anticipated soon. Starting with the details, AMD EXPO 1.2 will be adding support for module geometry, allowing users to mix and match various memory capacities. But […]

Read full article at https://wccftech.com/amd-chinese-ddr5-memory-support-expo-1-2-cudimm-mrdimm-ultra-low-latency-support/

AMD's Ex-FSR Lead has claimed that most team members who worked on FSR 4 & were part of the FidelityFX and GPUOpen teams left the company to join NVIDIA or Intel. AMD's FSR Shortcomings May Be Due To Its Members Leaving The Company To Join NVIDIA or Intel Recently, there has been a lot of chatter regarding AMD's FSR support on older GPUs. While the latest FSR Redstone release and the updated FSR 4.1 release were seen as big and positive updates, at the same time, many users were disappointed in the lack of FSR 4 support for older generations […]

Read full article at https://wccftech.com/ex-amd-fsr-lead-claims-most-gpuopen-fidelityfx-team-members-now-at-nvidia-or-intel/

Google TurboQuant was seen as a savior to the memory crisis; unfortunately, things will remain the same or get even worse from here. Google TurboQuant Isn't Going To Fix or Solve The Memory Crisis Back in March, Google launched a new algorithm called TurboQuant, which significantly compressed the KV Cache. The result was a drastic savings, up to 6x, in memory requirements for AI workloads. As soon as the algorithm was announced, reports emerged that memory prices were seeing a drastic price reduction. It was easy to link Google's TurboQuant to the price drop, and some went into panic mode, […]

Read full article at https://wccftech.com/everyone-thought-google-turboquant-would-solve-the-memory-crisis-but-makes-it-worse/

The prices of AMD's Ryzen CPUs have increased by more than 50%, signaling a panic in the DIY market amid heightened AI demand. Japanese Users Are Paying Up To 57% More For AMD Ryzen CPUs As The AI-Related Price Bumps Continue In a report published by PC Watch, it is stated that the Japanese DIY market is seeing a huge surge in hardware prices. While DRAM and GPU prices are way above their original rates, the next component that is being affected is the CPU. The report lists several AMD Ryzen CPUs, mainly the new Ryzen 9000 parts, seeing over […]

Read full article at https://wccftech.com/amd-ryzen-cpu-prices-increase-over-50-percent/

Intel's upcoming Z970 motherboards for Nova Lake-S Desktop CPUs will replace both high-end Z890 and mainstream B860 options. Intel Z970 Motherboards To Cover An Extensive Market With Both High-End & Mainstream Options For Nova Lake Builders Intel's 900-series motherboards will have a wide range of options for PC builders. The flagship Z990 chipset will be the recommended choice for enthusiast Nova Lake Desktop CPUs, featuring a dual compute tile configuration, while the Z970 chipset will retain a primary focus on the high-end market. Based on a new post by Jaykihn at X, it looks like the Z970 chipset may not […]

Read full article at https://wccftech.com/intel-z970-chipset-cover-both-z890-high-end-b860-mainstream-tiers-for-nova-lake/

Bolt Graphics has successfully taped out its Zeus GPU, which is expected to offer up to 6x faster HPC & 5x faster Path Tracing performance than NVIDIA's RTX 5090. The Bolt Graphics Zeus "12nm" GPU Aims To Offer 5x Faster Path Tracing Performance Than An RTX 5090 "5nm" Graphics Card Last year, Bolt Graphics announced its Zeus chip and claimed some big numbers. Today, we can finally confirm that Zeus wasn't just a paper announcement, as the chip has been successfully taped out. In a press release shared with us, Bolt Graphics confirms that its Zeus GPU test chip has […]

Read full article at https://wccftech.com/bolt-graphics-tapes-out-zeus-gpu-5x-faster-vs-nvidia-rtx-5090-in-path-tracing/

OpenAI is on a mission, an incredible mission, you might say, to scale up its AI compute capacity to 30GW by the end of this decade. 30GW By 2030: OpenAI Sets Ambitious Goals, but Will The Semiconductor Industry Be Able To Supply This Much Compute? Yesterday, Amazon and Anthropic announced that they were going to bring up 6GW of AI compute capacity by the end of this year. But their primary competitor, OpenAI, has more ambitious & mighty plans. Today, OpenAI announced that since the launch of ChatGPT, its revenue and compute resources have scaled at a rapid pace. This […]

Read full article at https://wccftech.com/openai-sets-goal-to-scale-up-ai-compute-capacity-to-a-whopping-30gw-by-2030/

Samsung's 1d DRAM (7th Gen 10nm) for next-generation HBM solutions might not undergo production soon due to failure to meet yields. Samsung's Next-Gen DRAM Tech For Future HBM5E Memory May Not Be Ready For Production A report published by Korean outlet, IT Chosun, suggests that due to less than ideal yields of its 1d "D1d" DRAM based on the 10nm process technology, Samsung might be pulling the plug on mass producing its next-gen HBM solutions. The DRAM technology had already received a pre-production approval (PRA), but concerns have been raised regarding the ROI of initiating a trial run, let alone […]

Read full article at https://wccftech.com/samsung-postpones-hbm5e-memory-production-indefinitely-d1d-dram-yields-fall-short/