NVIDIA unveils professional mobile graphics cards at Ampere and Turing
NVIDIA unveiled five NVIDIA RTX Axxx professional series mobile graphics cards based on the Ampere architecture at GTC 2021 and two mobile graphics accelerators based on the previous Turing architecture. The series of new products includes mobile versions of NVIDIA RTX A2000, RTX A3000, RTX A4000, and RTX A5000, as well as NVIDIA T600 and T1200.
A feature of all NVIDIA mobile video cards presented today is support for the third generation Max-Q technology set, which includes such energy efficiency features as Dynamic Boost 2.0, WhisperMode 2.0. There is also support for Resizable BAR and NVIDIA DLSS technologies.
The youngest segment of mobile professional graphics accelerators, RTX Axxx, opens the NVIDIA RTX A2000 Laptop GPU model. It includes 2560 CUDA cores, 80 third-generation NVIDIA tensor cores, and 20 second-generation RT cores. The peak performance of a video card in single-precision operations (FP32) can reach 9.3 teraflops. RT cores can provide a level of performance in tasks related to ray tracing up to 18.2 teraflops, and the performance of tensor cores can reach 74.7 teraflops. … The graphics card is equipped with 4 GB of GDDR6 video memory with a 128-bit bus and a peak bandwidth of 192 GB / s. The range of values of the dissipated thermal energy (TDP indicator) of the novelty is 35 to 95 W. The card supports PCI Express 4.0 x8 and supports DisplayPort 1.4 and HDMI 2.1 connectors.
As part of the NVIDIA RTX A3000 Laptop GPU model, 4096 CUDA cores, 128 tensor, and 32 RT cores are declared. The maximum theoretical performance in FP32 computing can reach 12.8 teraflops, the performance of RT cores is up to 25 teraflops, and tensor cores are up to 102.2 teraflops. The novelty received 6 GB of GDDR6 video memory with a 192-bit bus and a theoretical bandwidth of up to 264 GB / s. The TDP range of this model can vary from 60 to 130 watts. The graphics accelerator supports PCI Express 4.0 x16, as well as DisplayPort 1.4 and HDMI 2.1 connectors. The specifications also indicate that the new product belongs to the VR-Ready class, which means using virtual reality helmets.
The NVIDIA RTX A4000 Laptop GPU model received 5120 CUDA cores, 160 tensor cores, and 40 RT cores. Mobile RTX A4000 is ready to ensure performance in operations FP32 at 17.8 teraflops. The RT cores’ performance in ray tracing tasks can reach 34.8 teraflops, and the performance of tensor cores can reach 142.5 teraflops. The graphics accelerator is equipped with 8 GB of GDDR6 video memory with a 256-bit bus and a theoretical bandwidth of up to 384 GB / s. The declared range of operating TDP values of new items varies from 80 to 140 watts. The card supports PCI Express 4.0 x16 and DisplayPort 1.4, and HDMI 2.1 and offers VR-Ready.
The flagship model of the series, the NVIDIA RTX A5000 Laptop GPU, is ready to offer 6144 CUDA cores, 192 Tensor cores, and 48 RT cores. The card’s performance in single-precision calculations is up to 21.7 teraflops, the level of ray tracing performance is 75.6 teraflops, and tensor cores – up to 174 teraflops. As befits a flagship, the novelty is equipped with the largest supply of video memory among all the models presented today. The card received 16 GB GDDR6 with a 256-bit bus. Its peak theoretical bandwidth can reach 448 GB / s. Like all of the above new items, the flagship will be offered in different modifications of the declared TDP level – from 80 to 165 W. The card supports PCI Express 4.0 x16 and supports DisplayPort 1.4 and HDMI 2.1 connectors. It also claims VR-Ready support.
Separately, it is worth highlighting two more entry-level professional mobile graphics accelerators – NVIDIA T600 Laptop GPU and NVIDIA T1200 Laptop GPU. Both are built on the Turing architecture that the company unveiled in October 2018. The NVIDIA T600 model uses 896 CUDA cores and 4 GB of GDDR6 video memory with a 128-bit bus and a bandwidth of up to 160 GB / s. The manufacturer indicates only the performance of the novelty in FP32 operations – 2.5 Teraflops. Thermal energy dissipation (TDP) is 25 W. The video card supports PCI Express 4.0 x8.
The older NVIDIA T1200 Laptop GPU received 1024 CUDA cores and the same 4 GB GDDR6 with a 128-bit bus, but a higher peak bandwidth is up to 192 GB / s. The TDP range of the video card is 35W to 95W. The performance of the novelty is 3.7 teraflops in single-precision operations.
The notebooks, which will use the NVIDIA professional graphics cards presented today, will be on sale in the second quarter of this year.