HHCJ6 Dell NVIDIA Tesla K80 24GB GDDR5 PCI-E 3.0 Server GPU Accelerator

NVIDIA Tesla K80 GPU Review

Nvidia Corporation is a multinational technology corporation based in the United States. It creates graphics processing units (GPUs) for gaming and professional applications, as well as system on a chip (SoCs) for mobile computing and automotive applications. Its principal GPU line, dubbed “GeForce,” competes directly with Advanced Micro Devices’ “Radeon”-branded GPUs (AMD).

NVIDIA Tesla K80 GPU Review

NVIDIA’s Tesla K80 professional graphics card was released on November 17th, 2014. This card supports DirectX 12 and is built on a 28 nm process and based on the GK210 graphics processor in its GK210-885-A1 model. The GK210 graphics processor has a 561 mm2 die size and 7,100 million transistors. To improve performance, the Tesla K80 involves two graphics processors.

The NVIDIA Tesla K80 24GB is a general-purpose GPU that can handle single and double precision applications. It’s a dual-chip card with a total performance of 2.91 TFLOPS (FP64 / FP32) and 24 GB (12GB per GPU) of GDDR5 memory.

On each GPU, there are 2496 shading units, 208 texture mapping units, and 48 ROPs. NVIDIA has paired the Tesla K80 with 24 GB GDDR5 memory, which is connected through a 384-bit memory interface per GPU (each GPU manages 12,288 MB). The GPU runs at a frequency of 562 MHz, which may be increased to 824 MHz, and the RAM runs at 1253 MHz (5 Gbps effective).

Basic information:

  • Type: Graphic Accelerator
  • Family: Tesla
  • Bus interface: PCI-e 3.0 x16
  • CUDA cores: 2496
  • Core clock: 560 MHz
  • Memory clock: 2500 MHz
  • Memory size: 24 GB
  • Memory bandwidth: 480 GB/s
  • Memory type: GDDR5
  • Thermal solution: Passive
  • Display connectors: None
  • Power connectors: One 8-pin CPU power connector
  • power consumptions: 300 W

Features:

  • 4992 NVIDIA CUDA cores with a dual-GPU design
  • Up to 2.91 teraflops double-precision performance with NVIDIA GPU Boost
  • Up to 8.73 teraflops single-precision performance with NVIDIA GPU Boost
  • 24 GB of GDDR5 memory
  • 480 GB/s aggregate memory bandwidth
  • ECC protection for increased reliability
  • Server-optimized to deliver the best throughput in the data center

Performance:

The NVIDIA Tesla K80 Accelerator helps data Centre’s save money by delivering great performance with fewer, more powerful servers. It’s designed to increase throughput in real-world applications by 5-10x while saving clients up to 50% on an accelerated data Centre against a CPU-only system. This improved throughput translates to more scientific discoveries for researchers every day, or better user engagement for web services.

NVIDIA GPUs are already supported by over 550 industry-leading applications, including 15 of the top 15 HPC applications. Tesla K80 is designed to give higher performance in these applications, including features such as dual-GPU design and Dynamic GPU Boost. With advances in interconnect technologies like GPU direct RDMA, popular programming models like NVIDIA CUDA and OpenACC, and hundreds of accelerated applications and libraries, Tesla is the world’s premier platform for the accelerated data Centre.

Power consumption:

The NVIDIA Tesla K80 is a dual-slot card that receives power from a single 8-pin power connector with a maximum power draw of 300 W. This product does not have display connectivity because it was not designed to be used with monitors. A PCI-Express 3.0 x16 interface connects the Tesla K80 to the rest of the system. The card is 267 mm long and comes with a dual-slot cooling solution.

Memory size:

The Tesla K80 GPU is a twin GPU board with 24 GB of memory, blazing fast memory bandwidth, and up to 2.91 Tflops double precision performance with NVIDIA GPU Boost. It’s perfect for single and double precision applications that demand high data throughput while also requiring leading computation performance. 

The Tesla K80 is built for the most demanding computational jobs, with 24 GB of memory, and top single- and double-precision compute capability. It is equipped with the most up-to-date NVIDIA GPU Boost technology, which automatically monitors GPU utilization to maximize throughput and outperforms CPUs by up to 10 times.

GPU accelerator:

With the NVIDIA Tesla Accelerated-Computing Platform, you can speed up your most demanding high-performance data analytics and scientific computing applications. Tesla GPU Accelerators are based on NVIDIA’s Kepler compute architecture and run on CUDA, the most widely used parallel computing model in the world.

As a result, NVIDIA is releasing the K80 with only 13 of the 15 SMXes activated on each GPU, for a total of 4,992 CUDA cores. This puts the clockspeed in the 562MHz to 870MHz range. Meanwhile, memory clockspeeds have been reduced slightly from Tesla K40; each GPU now has 12GB of GDDR5 clocked at 5GHz, resulting in 240GB/sec of memory bandwidth per GPU. This gives the two GPUs a total memory pool of 24GB, with 480GB/sec bandwidth between them.

Meanwhile, the Tesla K80 will push the power envelope still farther, this time in order to fit two GPUs on a single card. Tesla K80 is a 300W card, as opposed to the 235W Tesla K40 and K20X. The fact that NVIDIA was able to fit two high-performance GPUs into a 300W package is no easy accomplishment, but GPU Boost is key to making the total product viable. As a result, energy efficiency improvements are almost entirely dependent on how well the Tesla K80 performs at 300W; the worst case scenario is that it is only 2% more energy efficient than the K40, while the best case is 59 percent, with the realistic situation.


Leave a Comment

Your email address will not be published. Required fields are marked *