Revamped Nvidia RTX 5090D V2 Delivers Reduced AI Capabilities: A 25% decline in VRAM and bandwidth compromises performance, yet pricing remains consistent at $2299 MSRP in China despite being a more modest gaming flagship.
Nvidia has unveiled the GeForce RTX 5090D V2, a graphics card tailored for the Chinese market. This new model presents a 25% reduction in memory capacity and bandwidth compared to its predecessors, the GeForce RTX 5090 and original RTX 5090D.
Memory Specifications
The GeForce RTX 5090D V2 boasts 12 memory chips, totalling 24GB of GDDR7 memory. This is a significant decrease from the 32GB found in the RTX 5090 and RTX 5090D, resulting in a total memory bandwidth drop from roughly 1790 GB/s to 1344 GB/s. The memory interface has been reduced from a 512-bit design to a 384-bit one.
Performance Implications
While the impact on gaming performance is minimal, with early benchmarks showing about a 2% lower gaming performance compared to the RTX 5090D, the story is different for AI workloads and memory bandwidth-heavy professional applications. The reduced memory capacity and bandwidth directly limit large AI models and datasets, which rely heavily on fast and large VRAM to avoid bottlenecks.
In terms of AI performance, the RTX 5090D V2 offers 2,375 AI TOPS, representing a 29% drop compared to the RTX 5090. This substantial decrease in AI throughput may negatively impact the graphics card's performance in AI-heavy use cases.
Shared Features with Predecessors
Despite the downgrades, the GeForce RTX 5090D V2 shares many features with its predecessors, such as the same architecture, process technology, transistors, SMs, GPU shaders, Tensor Cores, Ray Tracing Cores, Boost Clock, VRAM speed, L2 / Infinity Cache, Render Output Units, Texture Mapping Units, TFLOPS FP32 (Boost), Ray Tracing Performance, TBP, and launch date.
Pricing and Availability
The GeForce RTX 5090D V2 retains the same $2,299 MSRP as the previous RTX 5090D generation, despite its reduced memory capacity and bandwidth. This pricing strategy may make it a less attractive option for AI-heavy users. The graphics card was launched in August 2025.
Uncertain L2 Cache and Market Pricing
One aspect that remains uncertain is the L2 cache of the GeForce RTX 5090D V2. Additionally, the market pricing of the graphics card is a significant point of concern for Chinese consumers. Nvidia has not disclosed any information regarding the L2 cache of the GeForce RTX 5090D V2.
In summary, the GeForce RTX 5090D V2 offers a compromised memory subsystem that impacts AI and productivity computing more significantly than gaming. While it retains many features of its predecessors, its reduced memory capacity and bandwidth may make it a less attractive option for AI-heavy users in the Chinese market.
| Aspect | RTX 5090 / RTX 5090D | RTX 5090D V2 | Performance Implications | |----------------------|----------------------|-----------------------|----------------------------------------------| | VRAM Capacity | 32 GB GDDR7 | 24 GB GDDR7 | 25% less VRAM hampers large datasets/models, critical for AI workloads | | Memory Bus Width | 512-bit | 384-bit | Lower bandwidth (1790 -> 1344 GB/s) marginal for gaming, severe for AI/professional compute | | AI Performance (TOPS)| 3352 TOPS | 2375 TOPS | 29% drop limits AI throughput substantially | | Gaming Performance | Baseline | ~2% slower | Minimal gaming impact, most titles unaffected |
Read also:
- Battlefield 3 receives acclaim as the greatest Battlefield title to date.
- Significant Expansion Forecasted for Gesture-Controlled Technology Wearables Sector at a Rate of 14.4%
- Robotics Revolution: China Debuts Global First - A Shopping Mall for Human-like Androids - Captured on Film
- Upcoming iPhone 17 Debut in India: Design Leaks, Battery Specs, and Estimated Cost Revealed
