Author: wavepoint

Home Author is wavepoint

Multi-Tenant I/O Isolation with Open-Channel SSDs

wavepoint
Multi-Tenant I/O Isolation with Open-Channel SSDs Abstract: As SSDs become commonplace in data-centers and storage arrays, there is a growing demand for predictable latency and I/O isolation in multi-tenant environments. Traditional SSDs, serving block I/Os, fail to meet this demand. They offer a high-level of abstraction at the cost of unpredictable performance and suboptimal resource...

Using Crail for Analytics on Disaggregated Storage

wavepoint
Using Crail for Analytics on Disaggregated Storage Abstract: Storage disaggregation is an attractive storage consumption model as it comes with a lot of benefits, including the ability to independently scale compute and storage resources and achieve higher hardware utilization. In this presentation we introduce Crail, a system built specifically for fast networking and storage hardware...

Virtual Reality as a Workload Requiring Very High Performance Non-Volatile Memory

wavepoint
Virtual Reality as a Workload Requiring Very High Performance Non-Volatile Memory Abstract: Virtual reality presents an enjoyable, immersive, believable experience for the end user. This requires a guaranteed frame rate of 90 FPS, meaning a new frame must be rendered each and every 11ms. As pixel density increases, the number and detail of textures per...

Part 1: Problem Statement & Industry Trends

wavepoint
Part 1: Problem Statement & Industry Trends Abstract: As NAND flash density continues to double, it is becoming increasingly important that SSDs can be shared across multiple applications or hosts without impacting read QoS. In a Hyperscale environment, it is critical to provide all applications with the right ratio of compute to flash capacity while...

Part 2: Renegotiating the Host/Device Contract

wavepoint
Part 2: Renegotiating the Host/Device Contract Abstract: SSDs are becoming very large (e.g., 16 TB in a single device). In Cloud environment, these SSDs are often hosting many different latency sensitive applications. By renegotiating the contract between the host and device, P99 and P9999 read latency can be bounded. The new contract provides other knobs,...

The N3XT technology for Brain Inspired Computing

wavepoint
The N3XT technology for Brain Inspired Computing Abstract: Advances in brain-inspired computing are making rapid progress to meet the demands of abundant-data processing using a variety of techniques, including spiking neural networks, hyperdimensional computing using sparse vectors, deep neural nets, deep belief nets, restricted Boltzmann machines, and their variants. It is therefore crucial to create...

Persistent Memory Programming Tutorial

wavepoint
Persistent Memory Programming Tutorial Speaker Name <Speaker Affiliation> Abstract: How can applications best leverage persistent memory? What programming support & tools are available? The existence of NVDIMMs and the expectation of an explosion in capacity has sparked wide interest in these areas. This tutorial will start with a description of persistent memory products such as the...
Menu