MemArk (https://memark.io/en/) is a tech community leveraging modern memory architecture evolution. We are focusing on persistent memory (PMEM) technology as well as its applications with particular PMEM-aware optimization. We are building a leading community in both academia and industry that enables the persistent memory technology for both cutting-edge research and business values.
MemArk is led by the Artificial Intelligence (AI) unicorn 4Paradigm (https://en.4paradigm.com/), and sponsored by its industry partners including Intel. You are most welcomed to join our Slack channel for any feedback, technical collaboration, or just chatting:
At present, we have released two opensource projects that are built on PMEM with particular optimization.
- Pafka: A PMEM-Accelerated Streaming System
- It is an evolved version of Kafka, which is built based on PMEM for efficient data persistency, but with zero code change for your Kafka applications.
- It is able to achieve 10x higher throughput and 10x lower latency compared with Kafka
- Github project: https://github.com/4paradigm/pafka - PmemStore: A Persistent-Memory Based Data Storage Engine
- It is built based on Intel’s pmemkv, but specially optimized for AI workloads, with a new data structure named pskiplist.
- It contains our state-of-the-art research result, which is a persistent skip list. The research result is published in VLDB 2021 http://vldb.org/pvldb/vol14/p799-chen.pdf
- Github project: https://github.com/4paradigm/pmemstore
For any feedback, please join our Slack channel, or Email to us: contact@memark.io
