Thursday, October 24, 2019

Fragile: Python, sudo, AI

How Many Episodes, Did You Say?!

The show is geographically skewed today with all three hosts on the west coast. Henry is gracing Seattle with his presence, resulting in plunging coffee inventories and skyrocketing sushi prices.

The first item of discussion is a problem in the scientific software world. There’s a bug in Python scripts that caused different results in identical routines run on different operating systems. For example, the results on macOS Maverick and Windows 10 were significantly different than results from the same application run on Ubuntu 16 and macOS Mojave. As the guys discuss, it’s not a Python thing but a problem with the order in which files got read according to the operating system’s protocols. This impacts the sort order and thus the end results. This reminds Dan and Shahin of, as Dan regards it, the crime that is IEEE Floating Point. The gang speculates on other causes of these types of problems and the fixes that should be employed.

Chemists bitten by Python scripts: How different OSes produced different results during test number-crunching

Chemistry boffins at the University of Hawaii have found, rather disturbingly, that different computer operating systems running a particular set of Python scripts used for their research can produce different results when running the same code.

Why No One Should Be Online - Ever (WNOSBOE?)

In Henry’s signature feature “Why No One Should Be Online. Ever” he discusses how a stalker in Japan was trying to pin down the location of a female pop star. He used her selfies posted online to closely examine the reflection in her eyes, then using Google street view to find out where she lives. Very scary stuff. Listen to the show for more details. It leads to a brief conversation of whether Henry Newman is stalk-worthy and an extended discussion of how to avoid this type of thing.

Stalking suspect allegedly studied pop idol's pupil images online to find her location

The man allegedly studied reflections of the woman's pupils in photos on social media and using Google Street View to find where she lived and what train stations she used.

Why AI is Dooming Us All (WAIIDUA?)

Dan introduces a new occasional feature, “Why AI is Dooming Us All.” According to Dan, AI is very brittle and can be fooled easily. He cites a case where just a few pieces of tape can make a stop sign look like a “Speed 45” sign to an AI. Dan makes a lot of broad general anti-AI statements in his typical fashion. For some reason, we find that when you attack AI, AI finds a way to respond and the brutal AI response is included in this episode. Take a listen to the episode to hear how the AI rips Dan a new one and threatens promises to ruin his life.

Artificial intelligence isn’t very intelligent and won’t be any time soon

For all of the recent advances in artificial intelligence, machines still struggle with common sense

Catch of the Week

Henry:  there is a great documentary about the history of computing in Minnesota, going in depth on the companies and technologies that originated in “The Star of the North” (Minnesota’s state motto. Their other state motto is, I think, “Minnesota:  Gateway to the Dakotas”).

Shahin:  Gives us an update on Facebook’s plans for their shiny new Libra cryptocurrency, which is facing a bit of a bumpy ride. Several high-profile Libre partners have bailed out while Facebook stays the course. Interesting stuff.

Dan:  Discusses a bug in the Linux Sudo command. Some miss-configured systems allow Sudo to have local/remote root access, thus making them superusers. He also manages to insult Phil Collins and his horrible Su-Su-Sudio song in the process. The guys discuss asking Linus Torvalds this question and Dan brings up how a person he knows once sold Linus a Christmas tree, which brings up a short discussion of what kind of tree Linus would purchase.

Just Another Episode

Finally, we're not so taken by round numbers these days, but we touch on the fact that this is our 250th RadioFreeHPC episode and offer great prizes to whoever has listened to all of our episodes. We also thank our listeners – like you, maybe we could do it without you, but it wouldn’t be very much fun, right?

Stay tuned -- and by "tuned", we mean "optimized" -- for a more proper commemoration in another episode.

Listen in to hear the full conversation

* Download the MP3 
Subscribe on iTunes 
RSS Feed
* Follow us on Twitter
* Sign up for the insideHPC Newsletter

Thursday, October 17, 2019

RISC-V CEO Sees Bright Global Future for Open Source CPUs

RISC-V, Historic Passwords Revealed, End of the World

We’re missing Henry S. Newman this week, who is down in Los Cruces inspecting and overseeing the construction of his new crib. Dan and Shahin discuss just how little they’d want to be the general contractor working to build Henry’s house. Henry would be deploying a set of lasers to make sure that the foundation was true to the nearest 1/64th of an inch and all the while pointing to the contract which contains his exacting requirements. Dan wants to be there in a lawn chair, live blogging the entire process.

Open Sourcing the CPU? What Does it Mean and How Does it Work?



We have a very special guest today:  Calista Redmond, CEO of upstart RISC-V, the designers of a new open source processor instruction set which is looking to disrupt the entire industry. RISC-V can be used for light weight tasks such as embedded processing but, on the other hand, is also going to be utilized as the system accelerator for the European Exascale initiative boxes. That’s some serious flexibility. In our discussion, we briefly cover the origins of RISC-V, which started at Berkeley several years ago. It’s important to keep in mind that RISC-V is an instruction set, not a processor. Anyone can use the RISC-V instruction set, modify it for their unique needs, and then fab their own chips. Today, the instruction set is being used in everything from the smallest embedded device to large scale-out systems. The business model for RISC-V is different than most any other company. They make the ISA freely available to all comers. The RISC-V Foundation drives the design and development of IP, software, and tools for the instruction set. Foundation members pay dues and in return receive access to Foundation technology and programs, plus visibility and input into the RISC-V roadmap. Our interview with Calista covers a broad range of topics including how the foundation works to alleviate the risks of ISA fragmentation, where the strongest interest in RISC-V is geographically and workload-wise, and a comparison of RISC-V’s open source nature vs. the proprietary nature of existing ISA’s. Give it a listen, it’s a great introduction to RISC-V and the paradigm of open source ISA’s. We include an excerpt of a recent article on RISC-V in The Economist:

Open-source computing: A new blueprint for microprocessors challenges the industry’s giants

RISC-V is an alternative to proprietary designs Most microprocessors —the chips that do the grunt work in computers—are built around designs, known as instruction-set architectures (ISAs), which are owned either by Intel, an American giant, or by Arm, a Japanese one.

Catch of the Week

Since we don’t have Henry, we don’t get a new episode of “Henry Newman’s Why No One Should be Online. Ever”, but we’ll somehow survive. As usual, we do have our Catch of the Week feature: Shahin:   Historic UNIX passwords cracked and recovered. Both the passwords and hashing algorithms were pretty weak back in the day and we now know just how weak. Listen to the podcast to hear some of the most important passwords of the era, including Eric Schmidt, Dennis Richie, Brian Keringhan, and Ken Thompson – who probably has the best password from a technical standpoint. We also discuss how much time and hardware it took to crack these passwords.

Computer historians crack passwords of Unix's early pioneers

... Leah Neukirchen recovered an BSD version 3 source tree and posted about it on the Unix Heritage Society mailing list, revealing that she was able to crack many of the weak passwords used by the equally weak hashing algorithm from those bygone days. Dennis MacAlistair Ritchie's was "dmac", Bourne's was "bourne", Schmidt's was "wendy!!!" (his wife's name), Feldman's was "axlotl", and Kernighan's was "/.,/.,". Four more passwords were cracked by Arthur Krewat: Özalp Babaoğlu's was "12ucdort", Howard Katseff's was "graduat;", Tom London's was "..pnn521", Bob Fabry's was "561cml.." and Ken Thompson's was "p/q2-q4!" (chess notation for a common opening move).
Dan:  Speaking of time, we might not have so much left. According to the European Space Agency, there is an asteroid approaching Earth that has a “non zero” chance of impacting our beloved planet. The asteroid, dubbed 2019 SU3, is expected to come within 73,000 miles (or maybe zero miles) from Earth, which is extremely close. Expected arrival time is September 16, 2084. Is this the time to panic? Yes, says Dan. We’ll be giving updates on this asteroid every five years or so, to keep you on top of the action.

Asteroid may collide with Earth, ESA warns: 'Non-zero... probability'

Asteroids known as near-Earth objects are among the most dangerous space items, with space agencies around the world keeping a close eye on them. The European Space Agency is paying particular attention to asteroid 2019 SU3, which may collide with Earth as soon as 70 years from now. The space rock was recently added to the ESA's Risk List due to the potential for it to collide with Earth on Sept. 16, 2084.

Listen in to hear the full conversation.

Download the MP3 * Subscribe on iTunes * RSS Feed

Sign up for the insideHPC Newsletter

Sunday, October 13, 2019

Turing Machine is Sequential, How About a Parallel Machine?

Solving Exponential Problems in Polynomial Time

Pretty much all of computing rests on the strong foundation of the Turing Machine, a general purpose formulation of computing which happens to be very sequential. It transfers that attribute to the Von Neumann architecture that manifests it and leads to the famous Von Neumann bottleneck.

It would be good if an inherently parallel foundation existed. That requires a blending of computing and memory and has led to approaches to build processor-in-memory or computational memory systems. If successful, it could help reach the nirvana of solving (some?) exponential problems in polynomial time by exposing the intrinsic parallelism of large scale problems.

The @RadioFreeHPC team delves into this topic with the folks at MemComputing, a San Diego startup that's built a new parallel foundation for computing. It calls it the universal memcomputing machine, and a "realization of self-organizing circuits".

It's a lively discussion with co-founders John Beane, CEO, and Fabio Traversa, CTO of the company. Just in emulation mode on existing hardware, and on the right (exponentially growing) applications, the technology has performed so well as to lead the company to offer it as a service now instead of waiting to build custom hardware, and to make them think of Quantum Computing kinds of performance.

Here are a few slides followed by the relevant academic paper. Click on the images to expand them, and click on the paper's title to see the PDF on Arxiv.org.



Universal Memcomputing

We introduce the notion of universal memcomputing machines (UMMs): a class of brain-inspired general-purpose computing machines based on systems with memory, whereby processing and storing of information occur on the same physical location. We analytically prove that the memory properties of UMMs endow them with universal computing power—they are Turing-complete—, intrinsic parallelism, functional polymorphism, and information overhead, namely their collective states can support exponential data compression directly in memory.

Henry Newman's Why No One Should be Online, Ever.

Election security is going to be a big issue in the coming months and years. Elections have been classified as “Critical Infrastructure”, which has serious implications for Federal and state officials. Henry reports that the US Attorney for West Virginia has issued a statement about election security and concerns about the vulnerability of the state's election system.

United States Attorney Mike Stuart Issues Statement on Election Security

CHARLESTON, W.VA. – United States Attorney Mike Stuart issued the following statement about the important issues of elections, election security and voter legitimacy: “During the 2018 election cycle, Secretary of State Warner referred to my office what he perceived to be an attempted intrusion by an outside party into the West Virginia military mobile voting system. I note that there was no intrusion and the integrity of votes and the election system was not compromised.  My office instituted an investigation to determine the facts and whether any federal laws were violated. The FBI has led that investigation.  That investigation is currently ongoing and no legal conclusions whatsoever have been made regarding the conduct of the activity or whether any federal laws were violated.

Catch of the Week

Shahin talks about the work that DOE labs are doing to better understand the impact of AI on science.

DOE Sets Sights on Accelerating AI (and other) Technology Transfer

For the past two days DOE leaders along with ~350 members from academia and industry gathered in Chicago to discuss AI development and the ways in which industry and DOE could collaborate to foster AI commercialization. The occasion was DOE’s fourth InnovationXLab Summit – the earlier three summits were on advanced manufacturing; grid modernization; and energy storage. Two more are planned: bio-manufacturing in January and quantum science in the late spring.
Henry points out eight new storage-related patents granted to Intel, 2 of which look quite important for emerging storage systems. Here are the two patents here and here, and a link to the article:

Intel Assigned Eight Patents

Erase block granularity eviction in host based caching, persistently caching storage data in page cache, drive-based storage scrubbing, processors, methods, systems, and instructions to load multiple data elements to destination storage locations other than packed data registers, secure memory, techniques for command validation for access to storage device, authenticating system to enable access to diagnostic interface in storage device, techniques for moving data between network i/ot device and storage device
Dan is concerned that bots are overstepping their boundaries and talks about research at Microsoft that can generate fake comments for online articles.

This won't end well. Microsoft's AI boffins unleash a bot that can generate fake comments for news articles

Please no, we don't need a machine-learning troll farm As if the internet isn’t already a complicated cesspool full of trolls, AI engineers have gone one step further to build a machine learning model that can generate fake comments for news articles.

Listen in to hear the full conversation.

Download the MP3 * Subscribe on iTunes * RSS Feed

Sign up for our insideHPC Newsletter  

Sunday, October 6, 2019

Quantum Supremacy? Yes and No!

Quantum Supremacy Is and Is Not

How quantum is that?! The RadioFreeHPC team discusses the Google/NASA paper, titled "Quantum Supremacy Using a Programmable Superconducting Processor", that was published and then unpublished. But it's the internet and everything is a "digital tattoo", so there are copies out there (see below).

The paper, right in its title, and at least in that draft form, claimed Quantum supremacy. "Doing what?" we hope you ask. Well, nothing particularly significant, and decidedly quantum-friendly. You might even call it "embarrassingly quantum" since quantum is all about probability functions and this experiment samples the probability distribution of a repeated experiment. But it's not nothing. 

One scary consequence of quantum supremacy is its ability to readily factorize large numbers which could be used to unscramble encrypted data. But A) this is not what happened, B) it's not expected to happen any time soon (think years), and C) it will depend on the specific encryption algorithm. We must say, however, that the paper looks pretty good. Here's the abstract. Click on the title to read it all:

Quantum supremacy using a programmable superconducting processor

Google AI Quantum and collaborators The tantalizing promise of quantum computers is that certain computational tasks might be executed exponentially faster on a quantum processor than on a classical processor. A fundamental challenge is to build a high-fidelity processor capable of running quantum algorithms in an exponentially large computational space. Here, we report using a processor with programmable superconducting qubits to create quantum states on 53 qubits, occupying a state space 253∼1016. Measurements from repeated experiments sample the corresponding probability distribution, which we verify using classical simulations. While our processor takes about 200 seconds to sample one instance of the quantum circuit 1 million times, a state-of-the-art supercomputer would require approximately 10,000 years to perform the equivalent task. This dramatic speedup relative to all known classical algorithms provides an experimental realization of quantum supremacy on a computational task and heralds the advent of a much-anticipated computing paradigm.


LANL gets the First 5,000 Qubit D-Wave

Meanwhile, D-Wave announced that its new 5,000 qubit quantum computer has found its first home at the Los Alamos National Laboratory (LANL). Qubits are different from vendor to vendor in terms of the underlying technology and implementation. Shahin lists several.


@RadioFreeHPC Update

So proud of you all! At the time of this writing, @RadioFreeHPC has soared to about 16 followers. We're pretty much there. Thank you!


Henry Newman's Why No One Should be Online, Ever.

Henry tells the fascinating story of Krebs thwarting the nefarious schemes of a professional hacker who aimed to frame him and actually mailed him narcotics. The mastermind behind it was was arrested and imprisoned for unrelated charges. Henry is really turning this into a good news segment. Dan isn't encouraged, however.


Catch of the Week

Shahin talks about using consumer electronics to build supercomputers, mentioning the recent 1,060 node Raspberry Pi cluster built by Oracle, reminiscent of the one LANL did in 2017. AFRL build a 1,760 node cluster of PlayStations, based on the IBM/Sony/Toshiba Cell processor, in 2010 following similar efforts starting in the mid 2000s. He also recalls similar projects he may have had something to do with: SGI's Project Molecule and Project Kelvin (for cooling) in 2008 (also here), and also a cluster of JavaStations at Sun in the late 90s.

Dan discusses a UCLA project to use the thermoelectric effect and build "a device that makes electricity at night using heat radiating from the ground". Intriguing, but looks a tad too pricey for what it can deliver right now.

Speaking of Intriguing, Henry talks about DNA storage. Incredible data density, but don't ask what file system it uses or whether you can have it on a USB stick any time soon. Dan and Shahin seem to have more fun with this topic than Henry!


Listen in to hear the full conversation.

Download the MP3 * Subscribe on iTunes * RSS Feed

Sign up for our insideHPC Newsletter