Every 7.8μs your computer’s memory has a hiccup

Every 7.8μs your computer’s memory has a hiccup

Every 7.8μs your computer’s memory has a hiccup
Modern DDR3 SDRAM. Source: BY-SA/4.0 by Kjerish

During my recent visit to the Computer History Museum in Mountain View, I found myself staring at some ancient magnetic core memory.

Every 7.8μs your computer’s memory has a hiccup
Source: BY-SA/3.0 by Konstantin Lanzet

I promptly concluded I had absolutely no idea on how these things could ever work. I wondered if the rings rotate (they don't), and why each ring has three wires woven through it (and I still don’t understand exactly how these work). More importantly, I realized I have very little understanding on how the modern computer memory - dynamic RAM - works!

Every 7.8μs your computer’s memory has a hiccup
Source: Ulrich Drepper's series about memory

I was particularly interested in one of the consequences of how dynamic RAM works. You see, each bit of data is stored by the charge (or lack of it) on a tiny capacitor within the RAM chip. But these capacitors gradually lose their charge over time. To avoid losing the stored data, they must regularly get refreshed to restore the charge (if present) to its original level. This refresh process involves reading the value of every bit and then writing it back. During this "refresh" time, the memory is busy and it can't perform normal operations Continue reading

Applying Machine Learning At The Front End Of HPC

IBM and the other vendors who are bidding on the CORAL2 systems for the US Department of Energy can’t talk about those bids, which are in flight, and Big Blue and its partners in building the “Summit” supercomputer at Oak Ridge National Laboratory and “Sierra” at Lawrence Livermore National Laboratory – that would be Nvidia for GPUs and Mellanox Technologies for InfiniBand interconnect – are all about publicly focusing on the present, since these two machines are at the top of the flops charts now.

Applying Machine Learning At The Front End Of HPC was written by Timothy Prickett Morgan at .

Serverless Progressive Web Apps using React with Cloudflare Workers

Let me tell you the story of how I learned that you can build Progressive Web Apps on Cloudflare’s network around the globe with one JavaScript bundle that runs both in the browser and on Cloudflare Workers with no modification and no separate bundling for client and server. And when registered as a Service Worker, the same JavaScript bundle will turn your page into a Progressive Web App that doesn’t even make network requests. Here's how that works...

"Any resemblance to actual startups, living or IPO'd, is purely coincidental and unintended" - @sevki

A (possibly apocryphal) Story

I recently met up with some old friends in London who told me they were starting a new business. They did what every coder would do... they quickly hacked something together, bought a domain, and registered the GitHub org and thus Buzzwords was born.

The idea was simple: you could feed the name of your application into a machine learning model and it would generate the configuration files for your deployment for various container orchestrators. They achieved this by going through millions of deployment configurations and training a linear regression model by gamifying quantum computing because blockchain, or something (I told you this Continue reading

Uncertainty propagation in data processing systems

Uncertainty propagation in data processing systems Manousakis et al., SoCC’18

When I’m writing an edition of The Morning Paper, I often imagine a conversation with a hypothetical reader sat in a coffee shop somewhere at the start of their day. There are three levels of takeaway from today’s paper choice:

  • If you’re downing a quick espresso, then it’s good to know that uncertainty can creep into our data in lots of different ways, and if you compute with those uncertain values as if they were precise, errors can compound quickly leading to incorrect results or false confidence.
  • If you’re savouring a cortado, then you might also want to dip into the techniques we can use to propagate uncertainty through a computation.
  • If you’re lingering over a latte, then the UP (Uncertainty Propagation) framework additionally shows how to integrate these techniques into a dataflow framework.

We implement this framework in a system called UP-MapReduce, and use it to modify ten applications, including AI/ML, image processing, and trend analysis applications to process uncertain data. Our evaluation shows that UP-MapReduce propagates uncertainties with high accuracy and, in many cases, low performance overheads.

Are you sure?

Uncertainty can arise from a number of Continue reading

How to Avoid Being the Network Turkey

In honor of the Thanksgiving holiday, we’ve put together a few handy tips on how to avoid becoming the network turkey. While many may enjoy it at the family dinner table, nobody wants to be the turkey at work, causing disruptions for their coworkers and organization.

Fast Google Fonts with Cloudflare Workers

Fast Google Fonts with Cloudflare Workers

Google Fonts is one of the most common third-party resources on the web, but carries with it significant user-facing performance issues. Cloudflare Workers running at the edge is a great solution for fixing these performance issues, without having to modify the publishing system for every site using Google Fonts.

This post walks through the implementation details for how to fix the performance of Google Fonts with Cloudflare Workers. More importantly, it also provides code for doing high-performance content modification at the edge using Cloudflare Workers.

Google fonts are SLOW

First, some background. Google Fonts provides a rich selection of royalty-free fonts for sites to use. You select the fonts you want to use, and end up with a simple stylesheet URL to include on your pages, as well as styles to use for applying the fonts to parts of the page:

<link href="https://fonts.googleapis.com/css?family=Open+Sans|Roboto+Slab"
      rel="stylesheet">
<style>
body {
 font-family: 'Open Sans', sans-serif;
}
h1 {
 font-family: 'Roboto Slab', serif;
}

Your visitor’s browser fetches the CSS file as soon as the HTML for the page is available. The browser will request the underlying font files when the browser does layout for the page and discovers that it needs Continue reading

Philippines Department of ICT Sets the Multistakeholder Model into Action

Early this year, we embarked on an initiative with the Philippines Department of ICT (DICT) to co-develop the country’s National ICT Ecosystem Framework (NIEF) in a multistakeholder fashion. The NIEF, which succeeds the Philippine Digital Strategy, will guide the course of ICT use and development, as well as the priority areas for government, until 2022.

Our collaboration builds upon the success of the Philippine Chapter’s work with key stakeholders to advance open Internet development in the country, particularly in the policy sphere, and DICT’s sustained drive to expand avenues for participation in its policy formulation. Just last year, DICT and the Chapter, together with the Foundation for Media Alternatives, spearheaded the first Philippine Internet Governance Colloquium, which has been scaled up to a countrywide roadshow this year to help address pertinent Internet issues in different localities.

Having formalized our partnership in a memorandum of understanding, signed in July by DICT’s Secretary, Eliseo M. Rio, and the Internet Society’s Regional Bureau Director for Asia-Pacific, Rajnesh D. Singh, we pledged to support the DICT in embedding the multistakeholder approach not only in the framework’s development but in its implementation. Our engagement was complemented by an Internet Governance training workshop Continue reading