1 / 10

Shattering AI Performance Records

NVIDIA Volta Tensor Core GPU achieves new AI performance milestones in ResNet-50 for a single chip, single node, and single cloud instance. Explore the performance improvements.

nvidia
Télécharger la présentation

Shattering AI Performance Records

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. SHATTERING AI PERFORMANCE RECORDS NVIDIA Volta Tensor Core GPU Achieves New AI Performance Milestones

  2. GPU-POWERED DEEP LEARNING IS TRANSFORMING EVERY INDUSTRY, SOLVING CHALLENGES ONCE THOUGHT IMPOSSIBLE…

  3. THE IDEAL AI COMPUTING PLATFORM NEEDS TO PROVIDE IMPROVED PERFORMANCE, SCALABILITY AND PROGRAMMABILITY TO ADDRESS THE DIVERSITY OF MODEL ARCHITECTURES.

  4. NVIDIA’S VOLTA TENSOR CORE GPU ACHIEVED RECORD-SHATTERING RESNET-50 PERFORMANCE FOR A SINGLE CHIP, SINGLE NODE, AND SINGLE CLOUD INSTANCE.

  5. FASTEST SINGLE CHIP A single V100 Tensor Core GPU achieves 1,075 images/second when training ResNet-50, a 4X performance increase compared to the previous generation Pascal GPU. “New figures from NVIDIA illustrate the contribution hardware improvements can make to progress in machine learning: the AlexNet model that won ImageNet in 2012 took six days to train, can now be done in 18 minutes — a 500x speedup.” - Tom Simonite, WIRED

  6. FASTEST SINGLE NODE A single DGX-1 server powered by eight Tensor Core V100s achieves 7,850 images/second, almost 2X the 4,200 images/second from a year ago on the same system. “I feel like it’s important to note that these performance improvements [by NVIDIA] are more important than they immediately appear, because while these gains dramatically impact today’s workloads, they’re effectively preempting even more complex workloads of the future.” - Rob Williams, TechGage

  7. FASTEST SINGLE CLOUD INSTANCE A single AWS P3 cloud instance powered by eight Tensor Core V100 GPUs can train ResNet-50 in less than three hours, 3X faster than a TPU instance. “4 #TPU chips in a ‘Cloud TPU’ deliver 180 teraFLOPS of performance; by comparison, four V100 chips deliver 500 teraFLOPS. #NVIDIAwins.” - Karl Freund, Moor Insights

  8. NVIDIA TENSOR CORE GPU ARCHITECTURE ALLOWS US TO SIMULTANEOUSLY PROVIDE GREATER PERFORMANCE THAN SINGLE-FUNCTION ASICS, YET BE PROGRAMMABLE FOR DIVERSE WORKLOADS.

  9. EACH TESLA V100 TENSOR CORE GPU DELIVERS 125 TERAFLOPS OF PERFORMANCE FOR DEEP LEARNING COMPARED TO 45 TERAFLOPS BY A GOOGLE TPU CHIP. 4 TPU CHIPS IN A ‘CLOUD TPU’ V2 DELIVER 180 TERAFLOPS OF PERFORMANCE. BY COMPARISON, 4 NVIDIA V100 CHIPS DELIVER 500 TERAFLOPS OF PERFORMANCE.

  10. EXPLORE THE PERFORMANCE IMPROVEMENTS HERE

More Related