NVIDIA is developing a new GeForce RTX 4070 SKU with GDDR6 memory to address the GDDR6X shortage. The transition may affect gaming performance and pricing, with the latest models expected to be slightly cheaper than current GDDR6X versions.
NVIDIA Addresses GDDR6X Shortage with New GeForce RTX 4070 SKU Featuring GDDR6 Memory
According to reports, NVIDIA is developing a promising new GeForce RTX 4070 SKU that will utilize GDDR6 memory, which could bring about exciting advancements in the gaming world.
Last week, allegations surfaced that NVIDIA's GeForce RTX 4070 GPU SKUs were experiencing shortages due to various factors, including an imminent GDDR6X shortage caused by a quality check failure in one of Micron's G6X batches. These issues may be linked to production and supply cutbacks on the RTX 40 "Ada" SKUs. This transition could significantly impact the gaming performance and pricing of these GPUs.
Speculation suggested that NVIDIA might use GDDR6 memory to address the GDDR6X supply issues. According to the leaker MEGAsizeGPU, NVIDIA's GeForce RTX 4070 will be available in a new SKU featuring the PG141 SKU 347 PCB, including GDDR6 memory. The VRAM may be configured at 20 Gbps, but all other specifications will remain unchanged.
The NVIDIA GeForce RTX 4070 GPU is equipped with the AD104-250-A1 GPU, which boasts 5888 CUDA cores, 184 TMUs, and 64 ROPs. The GPU operates at a base frequency of 1920 MHz and a turbo frequency of 2475 MHz. It has a 192-bit bus interface supporting 12 GB of VRAM, with a maximum bandwidth of 504.2 GB/s using 21 Gbps G6X modules. The device is rated at 200W TDP.
RTX 4070 Models to Downgrade to GDDR6 Memory, Reducing Bandwidth by 5% Amid GDDR6X Shortage
Report shared by Wccftech, the GeForce RTX 4070 and the GeForce RTX 4070 SUPER currently use 21 Gbps GDDR6X memory, but they will require a downgrade to 20 Gbps GDDR6 memory. If NVIDIA maintains 20 Gbps speeds, the bandwidth will be 480 GB/s, a 5% decrease from the GDDR6X variants.
The Ada GPU of the GeForce RTX 4070 relies heavily on bandwidth; even minor reductions can significantly impact gaming performance. However, the RTX 4070 GPUs have a full 48 MB L2 cache on the AD104 processor, enabling numerous memory compression techniques and reasonable memory overclocking. The GDDR6 memory modules can be overclocked to 21 Gbps, which is sufficient for gaming sessions but not adequate for continuous use.
The NVIDIA GeForce RTX 4070 GDDR6 SKUs are expected to be released soon, as GDDR6 memory is readily available. Pricing is anticipated to be slightly lower than the GDDR6X models, priced around $499 US. The standard models are priced at approximately $549 US, with some retailers offering them for as low as $499 US. Availability has not yet been announced, but further information is expected in the coming days.


China vs. NASA: The New Moon Race and What's at Stake by 2030
China's AI Stocks Surge as Zhipu and MiniMax Hit Record Highs
Chinese Brands Are Taking Over Brazil — And It's Just Getting Started
Lumentum Holdings Rides AI Wave With Order Book Filled Through 2028
Chinese Cars in Europe: Consumer Trust Is Shifting Fast
Australia's Social Media Ban for Under-16s Sparks Global Movement
OpenAI Addresses Security Vulnerability in macOS App Certification Process
U.S. Automakers Push Back Against EU Rules Blocking American Trucks from European Market
China Vanke Seeks Bond Extension Amid Mounting Debt Crisis
SpaceX Eyes Historic IPO at $1.75 Trillion Valuation
BHP's Incoming CEO Visits China Amid Pricing Dispute with CMRG
Annie Altman Amends Sexual Abuse Lawsuit Against OpenAI CEO Sam Altman
TSMC Posts Strong Q1 2025 Revenue, Riding AI Chip Demand Wave
Anthropic Fights Pentagon Blacklisting in Dual Federal Court Battles
TSMC Japan's Second Fab to Produce 3nm Chips by 2028
Alibaba Shares Slide as Jefferies Slashes Price Target Over AI Spending and Business Losses
Apple's Foldable iPhone Faces Engineering Setbacks, Mass Production Timeline at Risk 



