A Vietnamese retailer has listed a slew of GeForce RTX 4070 graphics cards from various brands. The surplus of trademarks and models can obviously point to plenty of hardware at launch, which is allegedly several days away. However, pricing for these add-in-boards — which start at ~$800 — seems rather high.  

Bpstore.vn listed GeForce RTX 4070 graphics cards from Asus, Gigabyte, MSI, and Zotac, according to screenshots from I_Leak_VN. Prices of these AIBs start at 17.9 million Vietnamese dong ($637 without VAT) for Zotac’s Gaming GeForce RTX 4070 Trinity OC (which is a premium product) and climb all the way to 25.9 million dong ($916) for a pre-overclocked Asus GeForce ROG Strix RTX 4070 OC Edition. 

These prices are not exactly close to the $599 recommended by Nvidia. Yet, one thing to keep in mind about the preliminary listing is that retailers tend to monetize on the demand for yet-to-be-announced hardware and price AIBs well above recommended prices.  

We already know that these boards feature pretty comprehensive cooling systems with three fans (and possibly more). They are aimed at enthusiasts, which is unsurprising given the price points we are discussing.  

Speaking of prices, we should, of course, wait to see what kind of performance these cards deliver before we make conclusions about whether or not the GeForce RTX 4070 will join the ranks of the best graphics cards available. 

Image 1 of 2

Nvidia is expected to launch the GeForce RTX 4070 graphics card in mid-April. However, much of its technical details have been public knowledge for a while, despite not being officially disclosed by the company. The standard edition of Nvidia’s GeForce RTX 4070 shares the AD104 graphics processor with the RTX 4070 Ti but offers only 5888 CUDA cores running at 1920 MHz to 2475 MHz, down from 7680 CUDA cores on the ‘Ti’ version. Like its ‘Titanium’ counterpart, the GeForce RTX 4070 boasts a 12GB GDDR6X memory subsystem and a 192-bit interface. 

Even though the GeForce RTX 4070 utilizes a scaled-down AD104 GPU, it is predicted to deliver around 29 FP32 TFLOPS of computational power, a figure that rivals the performance of the GeForce RTX 3080. Nevertheless, the RTX 3080 features a 320-bit memory bus and an impressive peak bandwidth of 760 GB/s, which is significantly higher than the 504 GB/s bandwidth offered by the AD104’s 21 GT/s GDDR6X memory. Even the GeForce RTX 3070 Ti has a superior memory bandwidth (608 GB/s) compared to the upcoming card. On the other hand, the new graphics card is expected to have a sizable L2 cache, which will likely offset the limitations of its slower memory subsystem.