At present, NVIDIA has officially announced the launch of a number of professional graphics cards based on ampere architecture, including RTX A5500, RTX A4500, RTX A3000 12GB, RTX A2000 8GB and RTX A550 and other 5 mobile platforms, as well as 1 RTX A5500 desktop platform to meet the needs of different professionals.
It is worth mentioning that the desktop platform RTX A5500 is positioned between the RTX A6000 and the RTX A5000, which seems to replace the position of the RTX A5000. The new professional graphics card is a dual-slot design, with GA102, 24GB of video memory, a rate of 16 Gbps, equipped with 80 sets of SM (a total of 10,240 CUDA cores), 16 more than the RTX A5000, only 4 fewer than the RTX A6000, and 34.1 TFLOPs of the FP32, which is 23% higher than the RTX A5000. However, NVIDIA did not provide specific specification information, such as frequency and TDP. The professional graphics cards used by workstations are different from consumer-grade gaming cards, which use GDDR6 memory with ECC and the latter with GDDR6X memory.
The RTX A5500 of the new flagship mobile platform uses GA103, equipped with 58 sets of SM (a total of 7424 CUDA cores), with 16GB of video memory; the RTX A4500 replaces the original RTX A4000, the number of SM is increased to 46, the video memory is also increased to 16GB, and the memory rate is also increased to 16 Gbps; the RTX A3000 12GB and RTX A2000 8GB double the video memory capacity on the original basis. The RTX A1000 and RTX A500 are the latest additions to the series, equipped with 16 groups of SM (2048 CUDA cores in total) with 4GB of video memory, the difference between which is the memory width, with the former being 128 bits and the latter being 64 bits.
It should be noted that the RTX A5500 for desktop platforms is already on the market, while the new RTX series of professional graphics cards for mobile platforms will be shipped later.
(7891028)