ASRock Z890I Nova WiFi R2.0 enables 256 GB CQDIMM at DDR5-7400 MT/s

ASRock’s Z890I Nova WiFi R2.0 mini-itx board now supports CQDIMM, running 256 GB of DDR5 memory at DDR5-7400 MT/s to push past typical capacity and speed limits. The design targets high-bandwidth, high-capacity workloads such as Artificial Intelligence computing and professional content creation.

ASRock has announced that its Z890I Nova WiFi R2.0 motherboard, when paired with two Kingston 4-rank DDR5 128 GB memory modules, has successfully enabled CQDIMM support through optimized circuit design and coordinated hardware and software tuning. The platform achieves DDR5-7400 MT/s operation with a total capacity of 256 GB, surpassing the DDR5-7200 MT/s limit commonly seen on most platforms today, which marks a notable advance for compact systems needing both high capacity and high bandwidth DDR5 memory.

The CQDIMM approach builds on a 4-rank CUDIMM (Clocked Unbuffered DIMM) design that is specified with a maximum capacity of 128 GB per module and a native speed of DDR5-7200 MT/s, directly addressing the long-standing challenge of combining large memory footprints with sustained high transfer rates. By validating CQDIMM at higher speeds, the Z890I Nova WiFi R2.0 demonstrates that dense memory configurations no longer have to accept significant tradeoffs in bandwidth, particularly in small form factor designs where motherboard real estate and slot count are limited.

With CQDIMM support, the Z890I Nova WiFi R2.0 further pushes memory performance to DDR5-7400 MT/s, delivering a memory configuration that combines both high capacity and high performance to meet the demanding requirements of Artificial Intelligence computing, content creation, and professional workloads. The capability to run 256 GB of DDR5 at this speed positions the board as a specialized option for builders who need workstation-class memory resources in a compact system, providing additional headroom for data-intensive applications and complex multitasking scenarios.

50

Impact Score

Oracle expands Artificial Intelligence database tools for business data

Oracle introduced new agentic Artificial Intelligence capabilities for Oracle Artificial Intelligence Database aimed at helping enterprises build, deploy, and secure production-grade applications on business data. The company is positioning the platform across multicloud, hybrid, and on-premises environments with open standards and stronger data controls.

Using Artificial Intelligence to predict insider trading

UW Bothell researcher Solmaz Batebi is studying how Artificial Intelligence can improve the detection of illegal insider trading. Her work found machine learning methods outperformed traditional approaches and sharply reduced the time needed to analyze suspicious trades.

AMD adds new rdna 4m gpu targets for Medusa Point

AMD has added GFX1171 and GFX1172 software IDs in LLVM, expanding the RDNA 4m graphics targets tied to its Ryzen 500 “Medusa Point” APUs. The changes reinforce the view that RDNA 4m is an extension of RDNA 3.5 adapted to support features such as FSR 4.

UK delays copyright reform for Artificial Intelligence training

The UK government has stepped back from its earlier preferred copyright approach for Artificial Intelligence training and will keep the current legal framework in place while it gathers more evidence. Officials plan further work on digital replicas, transparency, labeling, licensing, and creator control rather than immediate legislative change.

Contact Us

Got questions? Use the form to contact us.

Contact Form

Clicking next sends a verification code to your email. After verifying, you can enter your message.