Nvidia’s newly introduced A100 Ampere Tensor Core GPU is now the fastest GPU ever recorded, or at least the card has had the best results ever in the OctaneBench benchmarking tool.
The A100 has a 7nm GA100 GPU, 54 billion transistors, 108 SMs, 6912 CUDA cores, an effective memory clock speed of 2340 MHz, a bandwidth of 1555 GB / s, a 5120-bit memory bus and 40 GB HBM2e -Storage. Here is the official description of Nvidia:
“The NVIDIA A100 Tensor Core GPU offers unprecedented acceleration for AI, data analysis and high-performance computing (HPC) of all sizes to meet the world’s toughest computer challenges. Driving the NVIDIA data center platform, the A100 can efficiently scale to thousands of GPUs or split into seven GPU instances using NVIDIA Multi-Instance GPU (MIG) technology to accelerate workloads of all sizes. And third generation tensor cores accelerate every precision for different workloads and accelerate the time to insight and the time to market launch. “
Although the A100 was launched in May, there are still no suitable benchmarks that show the true potential of the GPU. Jules Urbach, CEO of cloud graphics company OTOY, joined Twitter today share the OctaneBench benchmark results that put the Nvidia A100 through its paces.
“A record week – The @NVIDIA A100 is now the fastest GPU ever recorded on #OctaneBench: 446 OB4 * #Ampere seems ~ 43% faster than #Turing in #OctaneRender – even without #RTX! (* Standard Linux OB4 benchmark, RTX off, recompiled for CUDA11, Ref. 980 = 102 OB) “
A record week🚀
(* Standard Linux OB4 benchmark, RTX off, recompiled for CUDA11, Ref. 980 = 102 OB) pic.twitter.com/WlpG7ZUq2R
– Jules Urbach (@JulesUrbach) July 23, 2020
OctaneBench, developed by OTOY, uses the company’s OctaneRender software to create so-called “level playing fields” by ensuring that everyone uses the same version, scenes and settings. OctaneRender is specifically designed for Nvidia GPUs, thanks in part to using CUDA technology and ray tracing technology to push a GPU to its limits.
According to a picture shared by Urbach, the A100 had a total score of 446.03, which was determined by compiling the results from twelve separate tests. According to Urbach, the A100 is 43% faster than the existing Turing cards, although it is no longer specified which Turing GPUs are used, which means that it even outperforms the chips with the best scores from the previous lineup from Nvidia.
While this GPU is not intended for typical consumers, it may give us an insight into what we can expect from future high-performance GPUs from Nvidia. For information on our premium GPUs available today, see our guide to the best GPUs for games.