|
The New Wave of GPU Computing
|
|
|
|
Take Advantage: Limited Time Offer
|
|
For a limited time we are offering select GPU systems
and corresponding NVIDIA Tesla GPUs for a discounted price!
|
|
Right now is an unbelievable time to be in the HPC industry. NVIDIA, the world's leading GPU manufacturer, boasts a fully-developed line of Tesla GPU Accelerators tailored to the various computational needs within our community.
The Kepler Generation is the only generation to offer an actively-cooled solution, Tesla K40, to deliver the computational performance of Tesla in a workstation. In addition, the dual-GPU Tesla K80 remains the general purpose, mixed-workload HPC GPU Accelerator of choice for its double-precision performance and overall versatility. For more detailed information about the Kepler architecture GPUs, read our in-depth comparison.
The Maxwell Generation is the first generation to offer a Deep Learning specific GPU option, the Tesla M40. Maxwell based GPUs offer larger shared memory in each SMM (multiprocessor) - memory that is now separate from the L1 cache. Dual NVENC H.264 encoders allow for increased video throughput; large memory makes this GPU ideal for Deep Learning deployments using extremely deep neural networks. We offer a full list of specifications and comparison against the Kepler generation here for your reference.
Lastly, the new Pascal Architecture will offer unprecedented throughput and memory bandwidth. The Tesla P100 is by far the most powerful GPU in the market today with a peak performance of 21.2 TFLOPS half precision, 10.6 TFLOPS single precision, and 5.3 TFLOPS double precision. Combined with high-speed 720GB/s memory and the new NVLink interconnect, the P100 GPUs will offer exciting new possibilities. To learn more, continue reading below or check out additional details in our knowledge center.
|
|
|
Applied Deep Learning in Science and Engineering
|
|
We have all heard the buzz surrounding the current wave of Artificial Intelligence and Deep Learning being actively applied to industry and research today. But what does that mean for you?
Our new blog post summarizes recent projects using Deep Learning in scientific research and engineering. We want to help you understand if you can use deep learning. And if so, how it will help you. Read the one page summary or jump into the full blog post to learn how Deep Learning is improving everything from medicine to theoretical physics and material science.
|
|
|
Performance Density that Makes the Difference
|
|
As the ability to solve the world's toughest computational problems develops, the space needed to house those systems is getting smaller. Check out the most recent compute-dense systems such as the NVIDIA DGX-1 Supercomputer-in-a-Box, our 8-GPU 4U Octoputer, or our 4-GPU 1U NumberSmasher. These systems include the latest GPUs and offer a variety of density, interconnect and storage options.
|
|
|
NVIDIA DGX-1 Deep Learning System
|
|
|
|
- 8 NVIDIA Tesla P100 “Pascal” GPUs
- 28,672 NVIDIA CUDA cores, total
- Total of 128GB high-bandwidth GPU memory
- 42.4 TFLOPS double-precision, 84.8 TFLOPS single-precision, 169.6 TFLOPS FP16 half-precision
- Two 20-core Intel Xeon E5-2698v4 CPUs
- 512GB DDR4 2133MHz System Memory
- Four 1.92TB SSD in RAID0 (High-Speed Storage Cache)
|
|
|
4-GPU 1U NumberSmasher Server
|
|
|
|
- Up to 4 NVIDIA Tesla K40, K80, or M40 GPUs
- NVIDIA CUDA SDK installed and configured – Ready to run CUDA jobs!
- Up to 44 processor cores
- Up to 1 TB memory
- Up to 4 TB storage built-in
|
|
|
- Up to 8 NVIDIA Tesla M40, K80, or K40 GPUs
- Additional slots for EDR/FDR InfiniBand, 10G/40G Ethernet and storage controllers
- NVIDIA CUDA Toolkit installed and configured – Ready to run CUDA jobs!
- Up to 44 processor cores
- Up to 1.50 TB memory
- Up to 96 TB storage built-in
- Available with divided PCI-Express regions or with single-root I/O
|
|
|
Stay tuned to for more GPU offerings arriving later this year!
|
|
Please follow the links we've provided for the most detailed information about the exciting progress and products in the HPC world.
As always, if you have any questions, Microway's experts are happy to offer advice and share our technical expertise.
Feel free to call us when you design your next cluster or WhisperStation.
|
|
|
|
|
|
