AntidoteDB : large scale ( planet-scale ) distributed DB system. Competing with the like of cockroachDB or spanner. The core differentiator the architecture heavily rely on CRDT for its core functionality. It is a spin-off from the SyncFree EU research project. Sadly like a lot of EU or research-driven startup spin-off the documentation and website are slightly lacking polish. The architecture reference link is broken and a lot of stuff seems to be work in progress. Common guys! If you want to build a community and a product you really need to pick up the pace. This project has great potential, don't let it go to waste.
Machine Learning Benchmarks - Hardware Provider : a very good survey of machine learning benchmark of the current cloud provider. What is even more useful from that benchmark is that you get a cost overview of running ML application. Which is often a big unknown at the moment.
DeepMind Control Suite : benchmark suite for machine learning algorithms using a set of continuous control tasks with a standardised structure and interpretable rewards
EdgeML : Microsoft research demonstrate how to push machine learning at the edge and run KB models. We could quickly see machine learning enabled IoT device popping around us. [slides] [github]
Efficient Methods and Hardware for Deep Learning : Quest for speed never stop, and often that means getting read of those pesky indirection layers that make your software architecture so flexible :)
Pure Operation-Based Replicated Data Types : CRDT for operations rather than just value. But the core concepts are a little bit tricky and there is some potential pitfall in the approach. Such as the performance limitation and the reliance on causal stability ( which is really hard to obtain in pure decentralised systems)
A Survey on Homomorphic Encryption Schemes : this paper presents a thorough survey of the state of homomorphic encryption schemes. Homomorphic encryption allows manipulation of the encrypted data without the need to decrypt it. This will allow when hardware will be fast enough to deal with the complexity of the operations, to have a true secure distributed multitenant database. As no operation on the hosting side will require clear text decryption of the data and everything can be done securely on the client side.
Efficient Methods and Hardware for Deep Learning : Standford lecture where guest lecturer Song Han present algorithms and specialized hardware (FPGA, GPU, ASIC, etc..) that can be used to accelerate training and inference of deep learning workloads. [video]
DeepSecure : a framework that enables scalable execution of the state-of-the-art Deep Learning models in a privacy-preserving setting. The authors propose a system that enables data owner and model owner to maintain segregation of information while allowing them to work together without data leak between the two parties.
Quantum-secured blockchain : The authors propose in this paper a quantum blockchain architecture specifically designed to solve the post-quantum computer cryptographic weakness of currently used crypto algorithms in Bitcoin and other blockchain frameworks. However, it seems that they conveniently ignore newer cryptographic solutions that are "quantum resistant".
Survey of Neuromorphic Computing and Neural Networks in Hardware : heterogeneous hardware solutions are becoming the norm as classic CPU are not able to handle the bandwidth and processing power. Seriously, how a Intel or AMD CPU can process 1 Tb/S of bandwidth ... Anyway, as machine learning is reaching peak hype, the hardware that comes to accelerate it is getting more mainstream and diverse. This paper provides a good overview of the various technic and hardware used in the field. Moreover, it references an exhaustive collection of papers of the field.
Push on Green : Great article on google roll out policy and process. A lot of common sense, and also some less common but as important. This is a great read for anybody involved in software delivery and especially if you are aiming at having an efficient CI/DI system.
BBR : google congestion protocol for maximising bandwidth usage. It's new TCP scheduling algorithm to fight buffer-bloat at the TCP level. Since the majority of internet traffic is TCP, wide adoption would cause a big improvement. TCP scheduling only affects outgoing packets
Applied AI hard problems : current and future AI hard problem, the interesting bit is the "emergent" behavior aspect that computer scientist are trying to achieve. Where AI is not tailored for a specific problem by adapt to the environment it encounter.
Dark silicon & Hardware Reliability : the authors look at the impact of the dark silicon approach ( when not all component are turned on when the system is up) and how to leverage the "dark" ratio to maximise lifespan of hardware. [slides]
TLDK : project lead by Intel within the fd.io framework. It is trying to adresse the lack of high level ( as in layer 4 ) packet processing capabilities. The project aim at delivering UDP/TCP etc.. packet processing on top of vector packet processing of FD.io (which can works on top of DPDK). By doing so Intel will be able to finally have a comprehensive framework which will enable DPDK based solution to flourish beyond the pure networking stack (NFV) solution.
3dxpoint performance evaporate : seems that Intel is heavily scaling back its xpoint NVM performance claim. From 1000x to 10x ( still good but a far cry from what was promised). It seems that Intel had to push the technology early in order to counter a potential acquisition of its partner, Micron, by a competitor. Announcing the technology surely propped the share price making an acquisition difficult.
VLDB : very large databases 2016 proceedings are out. Sadly its one big zip file and didn't have time to go through it.
CAF : the authors propose a hardware core to core communication offloading engine. Providing an efficient queuing mechanism for transferring data between cores. I am not sure 100% of the value but the concept is interesting, let see if it catch on and if it can plays well in heterogeneous environment of today's datacenter. As core to core is slowly replaced with cored to GPU or core to FPGA or core to NVM.
VISC : Researcher are trying to improve the security, reliability, and performance implications of shipping all software in virtual instruction set form. They use the LLVM virtual instruction set as the shipping representation of code to enable install-time and run-time compilation of all such software. Probably a lot of potential there if coupled with uni-kernel tech.
Communications fabric for The Machine : HP session presenting their own fabric, competing with Intel Omnipath, PCIe and others already out there. The risk is that it might be too specialized for an industry wide adoption unless they start to aggressively "give it way" the IBM openpower way.
Emu : New type of server where the processing is moved to the data with a distributed shared memory model and high speed fabric. Seems like a revival of the connection machine for real time analytic solution. What is interesting is that they use a RapidIO fabric.
Redfish : Intel and other main player data center and systems management systems specification that aims to supply improved performance, functionality, scalability and security. Basically they are trying to sort out the IPMI and other ILO interface mess in order to offer a simple clean interface. Let's hope they achieve it.
Hints for Computer System Design : nice presentation and discution on how how Lampson's hints can still be used, some thirty years later, to implement functionally accurate, high-performing fault-tolerant software systems.
Workshop on Containers 2015 (WoC) : some interesting bits in that conference, Obviously the performance comparison ( linked yesterday) but the lesson learned slides show a very interesting bit regarding containers. It seems that the cost of security group is quite heavy as you scale the number of VM / containers.
