Late in April, the NVIDIA RTX 4000 SFF Ada Generation graphics card was made available to the general public. Since then, however, relatively few reviews and benchmarks have been published. Japanese hardware website Jisaku Hibi has released a thorough assessment that mostly focuses on gaming performance. The RTX 4000 Ada SFF was created as a small workstation graphics card, but because it uses an AD104 die, it is related to NVIDIA’s gaming-focused GeForce RTX 4070 and 4070 Ti cards. Many PC hardware websites claimed that the 70 W RTX 4000 Ada SFF will “offer GeForce RTX 3070-like performance,” but according to Jisaku Hibi’s research, the RTX 3060 Ti is the closest equivalent card (based on benchmark results).

The RTX 4000 SFF Ada Generation has several shading units deactivated, per the TPU GPU database, in order to hit the product’s goal shader count. It has 192 texture mapping units, 80 ROPs, and 6144 shading units. 192 tensor cores are also incorporated, which accelerate machine learning applications. 48 acceleration cores for ray tracing are present on the board. The RTX 4000 SFF Ada Generation and 20 GB GDDR6 RAM from NVIDIA are connected through a 160-bit memory interface. Memory is running at 1750 MHz (14 Gbps effective), while the GPU operates at 1290 MHz with a boost to 1565 MHz. The card’s main weaknesses are thought to be the SKU’s 70 W TGP and its constrained memory interface, which lead to average clock rates and a maximum memory bandwidth of just 280 GB/s.

According to Jisaku Hibi’s benchmark tests, the RTX 4000 SFF performs most closely to the GeForce RTX 3060 Ti, yet using 65% less power. The previous card’s total graphics power (TGP) is 200 W, yet it was only marginally faster—by no more than 13%. The RTX 4000 SFF outperforms the RTX 3060 (non-Ti) 170 W card by 20% because to its superior efficiency and newer architecture. It is unable to compete with the significantly bigger and more powerful RTX 3070 and 4070 GPUs.