Over at chiphell website, user panzerlied has posted a thread showcasing the performance of the RTX AD106 GPU. The first benchmarks of NVIDIA RTX AD106 graphics die have been revealed. Powering the RTX 4060 and 4060 Ti is set to power, as always, the numbers come in the form of the AD106 GPU’s 3DMark scores from the Chiphale forums.
We don’t know what GPU was used. However, we can assume that the user obtained the RTX 4060 Ti, and while the Ada Lovelace silicon easily outperformed its turning-based counterpart, the Ampere cannot.
Speedway, Port Royal, TimeSpy Extreme, FireStrike Extreme, FireStrike Ultra, and FireStrike scores for the AD106 (RTX 4060 Ti), GA104 RTX 3070 Ti , and TU104 are among the benchmarks ,RTX 2080 Super,.
Spoiler alert, the NVIDIA RTX 4060 Ti is extremely promising, outperforming both GPUs. The RTX AD106 GPU will have 4608 Cuda Cores, or 36 SMs. Because we do not currently have a GPU that meets this criteria, the GPU used may have been an early engineering prototype. Furthermore, the full-fat AD106 chip has a maximum clock speed of 2.535GHz. Furthermore, the memory is believed to be GDDR6 running at 18Gbps on a narrow 128-bit memory. Total effective bandwidth is 228Gb/s.
This data was compiled by the Harukaze 5719 into a single data graph to show the relative performance of each GPU. Surprisingly, the AD106 is quite competitive and competes with higher end SKUs from the previous two generations. It narrowly outperforms the NVIDIA RTX 3070 but is significantly faster than the RTX 2080. This means that the RTX 4060 Ti (full-fat AD106) is comparable to the RTX 2080 and RTX 3070. NVIDIA RTX 2080 and RTX 3070 graphics cards. The RTX 4060 Ti is the SAMKSHYA of the RTX 2080/3070 while consuming 27% less power.