NVIDIA Tesla K80 Dual GPU Accelerator With 24 GB RAM Is World's Fastest
@thebigk
•
Oct 26, 2024
Oct 26, 2024
1.5K
The NVIDIA Tesla K80 Dual GPU accelerator is the world's fastest graphics card - and boy, it's driving the entire graphics industry to a new benchmark! Nvidia announced the updated GPU on Monday, the fastest card from the company till date. The GPU is apart of Nvidia's Tesla Accelerated Computing Platform, which finds place in world's supercomputers used for AI, machine learning, big data, data analytics, scientific and high-performance engineering research applications. The K80 builds on the best of GPU, CUDA parallel computing model and an ecosystem of developers, vendors and datacenter OEMs.
In terms of performance, Nvidia claims that it delivers 2x the performance of its predecessor, the mighty Tesla K40 GPU. The K80 is 10x faster than most of the fastest CPUs available in the market. It delivers 8.74 teraflops of single-precision performance vs. the 5 teraflops of GeForce GTS 980 (desktop class GPU).
Some of the key features of the K80 dual-GPU accelerator include 24 GB of uber fast GDDR5 memory, 480 GBps bandwidth and about 4992 CUDA parallel processing cores. NVIDIA promises that the users will be able to reach out to the untapped performance of applications through the dynamic Nvidia GPU Boost technology that scales GPU clock speed based on requirements of the application in real time.
But the company's not being complacent with their achievements. NVIDIA's engineers are already working on the Pascal architecture that will deliver even faster GPUs in the next two years.
Source: #-Link-Snipped-#
In terms of performance, Nvidia claims that it delivers 2x the performance of its predecessor, the mighty Tesla K40 GPU. The K80 is 10x faster than most of the fastest CPUs available in the market. It delivers 8.74 teraflops of single-precision performance vs. the 5 teraflops of GeForce GTS 980 (desktop class GPU).
Some of the key features of the K80 dual-GPU accelerator include 24 GB of uber fast GDDR5 memory, 480 GBps bandwidth and about 4992 CUDA parallel processing cores. NVIDIA promises that the users will be able to reach out to the untapped performance of applications through the dynamic Nvidia GPU Boost technology that scales GPU clock speed based on requirements of the application in real time.
But the company's not being complacent with their achievements. NVIDIA's engineers are already working on the Pascal architecture that will deliver even faster GPUs in the next two years.
Source: #-Link-Snipped-#
