Fast Technology News on April 17, NVIDIA released two new professional graphics cards RTX A1000 and RTX A400 today, which can be seen from the number to be positioned at the entry level, and the architecture is not the latest Ada Lovelace, or the previous generation of Ampere.
You know, the last time the Ampere architecture pro card was released was two and a half years ago RTX A2000.
The RTX A1000/A400 all use a single-slot, low-profile design, which is very miniature, because the power consumption is only 50W, a single fan can do it, and there is no need for external auxiliary power supply.
Both of them integrate 2nd generation RT ray tracing cores, 3rd generation Tensor Tensor cores, 7th generation NVENC encoders, and 5th generation NVDEC decoders (with AV1 support).
The core of the RTX A1000 should be the same as the RTX A2000 is GA106, with only 2304 CUDA cores, 72 tensor cores, 18 ray tracing cores, a core boost frequency of 1463MHz, and FP32 floating-point performance of 6.74TFlops, which is even inferior to the RTX 3050.
The video memory is still 128-bit wide 8GB GDDR6, with a bandwidth of 192GB/s.
The core of the RTX A400 should be GA107, with only 768 CUDA cores, 24 tensor cores, 6 ray tracing cores, a boost frequency of 1758MHz, a floating-point performance of 207TFlops, and a video memory of only 64-bit 4GB GDDR6 with a bandwidth of 96GB/s.
The RTX A1000 is shipping now, with the RTX 400 to follow next month.