World’s Fastest SFF Graphics Card Hits Retail
Nvidia has launched a tiny RTX 4000 SFF Ada Generation graphics card that promises GeForce RTX 3070-like performance at 70W of power and fits in virtually any desktop PC. Thin dual-wide boards don’t come cheap. For example, they’re more expensive than the RTX 4080 because they’re aimed at professional users, but nothing prevents them from being installed in a regular gaming computer.
PNY’s Nvidia RTX 4000 SFF Ada Generation graphics card currently From $1,444 ShopBLTis a retailer known for selling hardware ahead of its rivals. This is why the board sells for more than its official suggested retail price of $1,250. Note that the board is equipped with four Mini-DisplayPort connectors, so the price of an mDP-DP or mDP-HDMI adapter should be added to the cost of this miniature his solution.
The Nvidia RTX 4000 SFF Ada Generation board features the company’s AD104 GPU with 6,144 active CUDA cores out of a total of 7,680 and 20GB of GDDR6 ECC memory connected to the GPU via a 160-bit interface. . The GPU is capped at about 1560 MHz boost frequency to reduce power consumption across the board and is rated at just 70W. This means it can be installed in almost any desktop computer, even those without an auxiliary PCIe power connector.
row 0 – cell 0 | GPUs | FP32 CUDA core | FP32 TFLOPS | INT8 TFLOPS | memory configuration | TBP | Manufacturer’s suggested retail price |
GeForce RTX 4070 Ti | AD104 | 7680 | 40 TFLOPS | 160/320 TFLOPS | 12GB 192bit 21GT/s GDDR6X | 285W | $799 |
GeForce RTX 4070 | AD104 | 5888 | 29 TFLOPS | 116/233 | 12GB 192bit 21GT/s GDDR6X | 200W | $599 |
RTX 4000 Ada Generation | AD104 | 6144 | 19.2 TFLOPS | 153/307 TFLOPS | 20GB 160bit 16GT/s GDDR6 ECC | 70W | $1,250 |
GeForce RTX 3090 Ti | GA102 | 10,752 | 40 TFLOPS | 160/320 TFLOPS | 24GB 384bit 20GT/s GDDR6X | 450W | $1,999 |
GeForce RTX 3070 | GA104 | 5888 | 20.31 TFLOPS | 81/160 TFLOPS | 8GB 256bit 14GT/s GDDR6 | 220W | $499 |
From a performance perspective, Nvidia’s GA104 graphics processor on this constituent chip offers a peak FP32 performance of 19.2 TFLOPS, which is theoretically similar to the GeForce RTX 3070. Still, with 20GB of memory onboard, the card is a bit more future-proof. Better than the RTX 3070 and potentially more useful for professionals and AI researchers. The memory configuration will likely use 2GB GDDR6 chips on each side of the PCB, otherwise the 160-bit interface would limit the maximum memory to just 10GB.
Nearly 20 FP32 TFLOPS overshadows the excellent performance of the recently launched GeForce RTX 4070 (29 FP32 TFLOPS). The board also boasts a peak RT performance of 44.3 TFLOPS and a peak FP8/INT8 tensor performance of 153/306.8 TFLOPS/TOPS (with and without sparse). Of course, FP8/INT8 performance has nothing to do with gaming, but it’s an added bonus for the pro market. In fact, 153/306.8 TFLOPS/TOPS is comparable to the more expensive and power hungry Nvidia GeForce RTX 3090 Ti (if you don’t care about accuracy as the 3090 Ti only supports native FP16).
Other benefits of the RTX 4000 SFF Ada include professional drivers from Nvidia and support for professional software ISVs. Additionally, it comes with a 3-pin mini-DIN connector for stereoscopic 3D output (such as Nvidia 3D Vision) and supports frame lock functionality for multi-display applications.
Pricing and availability should increase in the coming weeks as the card becomes more widely available. I am targeting.